Output-Based Event-Triggered Control with. Guaranteed L -gain and Improved and Decentralised Event-Triggering

Size: px
Start display at page:

Download "Output-Based Event-Triggered Control with. Guaranteed L -gain and Improved and Decentralised Event-Triggering"

Transcription

1 Output-Based Event-Trggered Control wth 1 Guaranteed L -gan and Improved and Decentralsed Event-Trggerng M.C.F. Donkers and W.P.M.H. Heemels Abstract Most event-trggered controllers avalable nowadays are based on statc state-feedback controllers. As n many control applcatons full state measurements are not avalable for feedback, t s the objectve of ths paper to propose event-trggered dynamcal output-based controllers. The fact that the controller s based on output feedback nstead of state feedback does not allow for straghtforward extensons of exstng event-trggerng mechansms f a mnmum tme between two subsequent events has to be guaranteed. Furthermore, snce sensor and actuator nodes can be physcally dstrbuted, centralsed event-trggerng mechansms are often prohbtve and, therefore, we wll propose a decentralsed event-trggerng mechansm. Ths event-trggerng mechansm nvokes transmsson of the outputs n a node when the dfference between the current values of the outputs n the node and ther prevously transmtted values becomes large compared to the current values and an addtonal threshold. For such event-trggerng mechansms, we wll study closed-loop stablty and L -performance and provde bounds on the mnmum tme between two subsequent events generated by each node, the so-called nter-event tme of a node. Ths enables us to make tradeoffs between closed-loop performance on the one hand and communcaton load on the other hand, or even between the communcaton load of ndvdual nodes. In addton, we wll model the event-trggered control system usng an mpulsve model, whch truly descrbes the behavour of the event-trggered control system. As a result, we wll be able to guarantee stablty and performance for event-trggered controllers wth larger mnmum nter-event tmes than the exstng results n the lterature. We llustrate the developed theory usng three numercal examples. Index Terms Event-Trggered Control Systems, Impulsve Systems, Hybrd Systems, Dynamc Output-Based Control, Decentralsed Control. Ths work s partally supported by the Dutch Scence Foundaton (STW) and the Dutch Organzaton for Scentfc Research (NWO) under the VICI grant Wreless controls systems: A new fronter n automaton, by the European 7th Framework Network of Excellence by the project Hghly-complex and networked control systems (HYCON ), and by the project Decentralsed and Wreless Control of Large- Scale Systems (WIDE ). Tjs Donkers and Maurce Heemels are wth the Hybrd and Networked Systems group of the Department of Mechancal Engneerng, Endhoven Unversty of Technology, PO Box 513, 56 MB Endhoven, the Netherlands, {m.c.f.donkers, m.heemels}@tue.nl. July 18, 211

2 2 I. INTRODUCTION In many control applcatons nowadays, the controller s mplemented on a dgtal platform. In such an mplementaton, the control task conssts of samplng the outputs of the plant and computng and mplementng new actuator sgnals. Typcally, the control task s executed perodcally, snce ths allows the closed-loop system to be analysed and the controller to be desgned usng the well-developed theory on sampled-data systems, see, e.g., 1, 2. Although perodc samplng s preferred from an analyss and desgn pont of vew, t s sometmes less preferable from a resource allocaton pont of vew. Namely, executng the control task at tmes when no dsturbances are actng on the system and the system s operatng desrably s clearly a waste of computaton resources. Moreover, n case the measured outputs and/or the actuator sgnals have to be transmtted over a shared (and possbly wreless) network, unnecessary utlsaton of the network (or power consumpton of the wreless rados) s ntroduced. To mtgate the unnecessary waste of communcaton and computaton resources, an alternatve to perodc control, namely, event-trggered control has been proposed, see 3 6. Event-trggered control s a control strategy n whch the control task s executed after the occurrence of an external event, generated by some well-desgned event-trggerng mechansm, rather than the elapse of a certan perod of tme as n conventonal perodc control. As expermental results show, see, e.g., 3 11, event-trggered control s capable of reducng the number of control task executons, whle retanng a satsfactory closed-loop performance. Although the advantages of ETC are well-motvated and practcal applcatons show ts potental, relatvely few theoretcal results exst that study ETC systems, see, e.g., In these references, several dfferent eventtrggerng mechansms and control strateges are proposed. For nstance, n 12, 13, an mpulsve control acton s appled to the system that resets the state to the orgn every tme the state of the plant exceeds a certan threshold. The analyss s performed for frst-order stochastc systems, as analyss of larger-dmensonal systems s dffcult, and t s shown that the varance of the state s smaller when compared to a sampled-data controller, whle havng approxmately the same number of control updates. Another nterestng approach to event-trggered control s presented n 14 16, n whch the system s controlled n open loop, usng an nput generator that uses a predcton of the states of the plant to produce a control sgnal. These predcted states are only corrected n case the true plant state devates too much from ts predcted value. Such a devaton can be caused by dsturbances, 14, 15, or by the fact that the plant model s ncorrect 16. A more basc emulaton-based approach s taken n By emulaton-based, we mean that the controller s desgned wthout consderng the event-trggered nature of the control system, and, subsequently, an event-trggerng mechansm s desgned to ensure that the event-trggered control system s stable, has some guaranteed lower bound on the performance and some guaranteed upper bound on the number of events wthn a certan tme nterval. The dfferences between the work dscussed n les n the fact that n 17, 18 the nfluence of unknown dsturbances are studed, whereas n only stablsaton s consdered. Another dfference s the condton to generate the events. In 17, events are generated n case the state of the plant s a larger than a certan threshold, n 18, 19 when the relatve dfference between the state of the plant and the prevously sampled state volates a July 18, 211

3 3 certan threshold, and n 2, 21 when the absolute dfference between the state of the plant and the prevously sampled state volates a certan threshold. An mportant observaton to be made about the aforected works s that most of them consder state-feedback controllers, whch assumes that all the plant states can be measured. To the best of the authors knowledge, the only theoretcal result on event-trggered control usng dynamcal output-based controllers s presented n 2. However, an analyss of the mnmum tme between two subsequent events, the so-called nter-event tme, s not avalable for 2 and, thereby, guarantees on the upper bound on the number of events cannot be made. Furthermore, extendng the event-trggerng mechansms n 18, 19 to output-based controllers s not straghtforward, snce for these event-trggerng mechansms, no mnmum nter-event tme can be shown to exst, even though they have a guaranteed mnmum nter-event tme for state-feedback controllers. For any event-trggered control system to be useful, we need such a lower bound on the nter-event tme, as our prmary reason to make control systems event-trggered s to save computaton and communcaton resources. In ths paper, we analyse stablty and L -performance of event-trggered control systems for gven dynamcal output-based controllers. We consder the case where the sensors and actuators, whch can be grouped nto nodes, and controllers can be physcally dstrbuted. Ths causes a centralsed event-trggerng mechansm to be prohbtve, as the condtons that generate events would need access to all the plant and controller outputs at all tmes and wthout any delays, whch would requre transmttng all the node data contnuously. To resolve ths ssue, we wll propose a decentralsed event-trggerng mechansm, n whch events are trggered on the bass of local nformaton only. Inspred by 19, we propose an event-trggerng mechansm that nvokes transmsson of the controller or the plant outputs of a node when the dfference between the current values n the node and ts prevously transmtted values becomes large compared to the current values and an addtonal threshold. Ths addtonal threshold ensures that each node has a nonzero mnmum nter-event tme, whch allows us to guarantee a bound on the total number of transmssons. Interestngly, the event-trggerng mechansm presented n ths paper can be seen as a unfcaton of the event-trggerng mechansms proposed n 18, 19 and As a second contrbuton of ths paper, we propose to model the event-trggered control system as an mpulsve system, see, e.g., 23, 24, whch truly descrbes the behavour of the event-trggered control system. Furthermore, we extend the framework presented n 19 towards output-feedback controllers and L -performance, and we formally show that the mpulsve systems framework provdes stablty guarantees for event-trggerng mechansms that result n larger mnmum nter-event tmes than the extended results of 19. These stablty condtons wll be based on lnear matrx nequaltes (LMIs), so that effcent verfcaton s possble. We wll provde three numercal examples to demonstrate varous aspects of the developed theory. In partcular, we wll llustrate that the guaranteed lower bounds on the mnmum nter-event tmes are ndeed mproved wth respect to exstng results n the lterature and that the ncluson of a nonzero threshold n the event-trggerng mechansm s necessary to guarantee a postve mnmum nter-event tme for each node. The remander of ths paper s organsed as follows. After ntroducng the necessary notatonal conventons, we ntroduce the model of the decentralsed output-based event-trggered control system n Secton II. We analyse ts July 18, 211

4 4 stablty and ts L -gan propertes n Secton III, and n Secton IV we provde a way to compute the lower bound on the mnmum nter-event tme of each node. In Secton V, we extend the work of 19 towards output-based dynamcal controllers and L -performance, and present a theorem that states that the mpulsve system formulaton of the event-trggered control problem allows us to guarantee stablty and performance for event-trggered controllers wth at least the same mnmum nter-event tmes as the results based on the reasonng of 19. Fnally, the presented theory s llustrated by numercal examples n Secton VI and we draw conclusons n Secton VII. The appendx contans the proofs of the more techncal lemmas and theorems. A. Nomenclature For a vector x R n, we denote by x := x x ts 2-norm, and by x J the subvector formed by all components of x n the ndex set J {1,..., n}. For a symmetrc matrx A R n n, λ max (A) and λ mn (A) denote the maxmum and mnmum egenvalue of A, respectvely. For a matrx A R n m, we denote by A R m n the transposed of A, and by A := λ max (A A) ts nduced 2-norm. Furthermore, by A J and A J, we denote the submatrces formed by takng all the rows of A n the ndex set J {1,..., n}, and by takng all the columns of A n the ndex set J {1,..., m}, respectvely. By dag(a 1,..., A N ), we denote a block-dagonal matrx wth the entres A 1,..., A N on the dagonal, and for A B A brevty we wrte symmetrc matrces of the form as. B C For a sgnal w : R + R n, where R + denotes the set of nonnegatve real numbers, we denote by w Lp = ( w(t) p dt) 1/p ts L p -norm for p N, provded that the ntegral s fnte, and by w L = ess sup t R+ w(t) ts L -norm. Furthermore, we defne the set of sgnals wth a fnte L p -norm as L p := {w : R + R n w Lp < } for p N { }. Fnally, for a sgnal w : R + R n we denote the lmt from above at tme t R + by w + (t) = lm s t w(s), provded that t exsts. B C II. EVENT-TRIGGERED CONTROL In ths secton, we present the event-trggered control problem and model the event-trggered control system as an mpulsve system. A. Problem Formulaton Let us consder a lnear tme-nvarant (LTI) plant gven by d dt x p = A p x p + B p û + B w w, y = C p x p, where x p R np denotes the state of the plant, û R nu the nput appled to the plant, w R nw an unknown dsturbance and y R ny gven by the output of the plant. The plant s controlled usng a contnuous-tme LTI controller d dt x c = A c x c + B c ŷ, u = C c x c, (1) (2) July 18, 211

5 5 where x c R nc denotes the state of the controller, ŷ R ny the nput of the controller, and u R nu the output of the controller. We assume that the controller s desgned to render (1) and (2) wth y(t) = ŷ(t) and u(t) = û(t), for all t R +, asymptotcally stable,.e., an emulaton-based approach s taken. In ths paper, however, we consder the case where the controller s mplemented n a sampled-data fashon, whch causes y(t) ŷ(t) and u(t) û(t) for almost all t R +. In partcular, we study decentralsed event-trggered control whch means that the outputs of the plant and controller are grouped nto N nodes and the outputs of node {1,..., N} are only sent at the transmsson nstants t k, k N. Hence, at transmsson nstant t k, node transmts ts respectve entres n y and u, and the correspondng entres n ŷ and û are updated accordngly, whle the other entres n ŷ and û reman the same. Such constraned data exchange can be expressed as n whch v = y u, ˆv = ŷ û, and ˆv + (t k ) = Γ v(t k ) + (I Γ )ˆv(t k ), (3) Γ = dag(γ 1,..., γ ny+nu ), (4) for all {1,..., N}. In between transmssons, we use a zero-order hold,.e., d dt ˆv(t) =, for all t R +\ ( N =1 {t k k N} ). (5) In (4), the elements γ j, wth {1,..., N} and j {1,..., n y}, are equal to 1 f plant output y j s n node and are elsewhere, the elements γ j+ny, wth {1,..., N} and j {1,..., n u }, are equal to 1 f controller output u j s n node and are elsewhere. We assume that for each j {1,..., n y + n u }, t holds that N =1 γj >,.e., we assume that each sensor and actuator s at least n one node 1. Furthermore, we assume that at tme t =, t holds that ˆv() = v(). Ths can be accomplshed by transmttng all sensor and actuator data at the tme the system s deployed 2. In the case that t k = t j k j for some k, k j N and some, j {1,..., N}, we assume that the updates as n (3) take place smultaneously or drectly after one another n a neglgble amount of tme. Obvously, the order of updatng s rrelevant as can be seen from (3). Moreover, note that performng multple successve transmssons at one tme nstant has exactly the same effect as dong these updates smultaneously. In a conventonal sampled-data mplementaton, the transmsson tmes are dstrbuted equdstantly n tme and are the same for each node, meanng that t k +1 = t k + h, for all k N and all {1,..., N}, and for some constant transmsson nterval h >, and that t k = tj k, for all k N and all, j {1,..., N}. In event-trggered control, however, these transmssons are orchestrated by a decentralsed event-trggerng mechansm, as s shown n Fg. 1. We consder a decentralsed event-trggerng mechansm that nvokes transmssons of node data when the dfference between the current values of outputs and ther prevously transmtted values becomes too large n 1 In case a sensor or actuator s not n any node, meanng that ths sensor or actuator s, effectvely, not part of the control loop, we smply remove the correspondng nput or output from the plant and controller model. 2 Ths assumpton could be removed, but t would ntroduce addtonal techncaltes later. For reasons of readablty, we opted to work under ths rather mld and natural assumpton. July 18, 211

6 6 Fg. 1: Event-trggered Control Schematc, ndcatng the event-trggerng mechansms (ETMs). an approprate sense. In partcular, the event-trggerng mechansm proposed n ths paper, results n transmttng the outputs of the plant or the controller n node {1,..., N} at tmes t k, satsfyng t k +1 = nf { t > t k e J (t) 2 = σ v J (t) 2 } + ε, (6) and t =, for some σ, ε. In these expressons, e J and v J denote the subvectors formed by takng the elements of the sgnals e and v, respectvely, that are n the set J = {j {1,..., n y + n u } γ j = 1}, and e(t) = ˆv(t) v(t) (7) denotes the error nduced by the event-trggered mplementaton of the controller at tme t R +. Hence, the eventtrggerng mechansm (6), whch s based on local nformaton avalable at each node, s such that when for some {1,..., N}, t holds that e J (t) 2 = σ v J (t) 2 +ε,.e., the norm of the error nduced by the event-trggered mplementaton of the sgnals n node becomes too large for the frst tme, node transmts ts correspondng sgnal n v(t) and, thus, the sgnal ˆv(t) s updated accordng to (3). Ths mples that e + (t k ) = (I Γ )e(t k ) and thus e + J (t k ) =. Usng ths update law, and the aforementoned assumpton that ˆv() = v(), yeldng e() =, we can observe that the error nduced by the event-trggered control scheme satsfes e J (t) 2 σ v J (t) 2 + ε, (8) for all t R + and all {1,..., N}. The man objectve of ths paper s to determne σ and ε for all {1,..., N}, such that the closed-loop event-trggered system s stable n an approprate sense and a certan level of dsturbance attenuaton s guaranteed, whle the number of transmssons of the outputs of the plant and the controller s mnmsed. Note that for ε =, {1,..., N}, the event-trggerng condtons n (6) can be seen as an extenson of the event-trggerng mechansm of 19 for output-based controllers, and for σ =, {1,..., N}, t s equvalent to the eventtrggerng mechansm of As such, the event-trggerng mechansm n (6) unfes two earler proposals, whle addtonally, output-based controllers and decentralsed event trggerng are consdered. Remark II.1 In ths paper, we assume that the controller s gven n contnuous tme as n (2). To mplement ths controller on a dgtal platform, the followng optons can be consdered: () the controller output s obtaned by July 18, 211

7 7 numercal ntegraton, or () the controller output s obtaned usng a dscrete-tme equvalent of the contnuoustme controller, based on a samplng nterval that s (suffcently) smaller than the smallest nter-event tme (see Theorem IV.1 below). Ths, however, means that the event-trggered control strategy presented n ths paper s partcularly useful when the objectve s to save communcaton resources and/or battery power of wreless rados, whch s mportant for many (wreless) networked control systems, see, e.g., 25 28, and s less useful for savng computaton resources. B. An Impulsve System Formulaton In ths secton, we reformulate the event-trggered control system as an mpulsve system, see, e.g., 23, 24, of the form d dt x = Ā x + Bw, when x C (9a) x + = Ḡ x, when x D, {1,..., N}, (9b) where x X R nx denotes the state of the system and w R nw an external dsturbance. The flow and the jump sets are denoted by C R nx and D R nx, {1,..., N}, respectvely, and X = C ( N =1 D ). Note that the transmsson tmes t k, k N, as n (6), are now related to the event tmes at whch the jumps of x, accordng to (9b) for {1,..., N}, take place. To arrve at a system descrpton of the event-trggered control system (1), (2), (3), (5), and (6) of the form (9), we combne (1), (2), (3), (5) and (7), and defne x := x e R nx, where x = x p x c and n x := n p +n c +n y +n u, yeldng the flow dynamcs of the system d dt x = A + BC B x + E w, (1) C(A + BC) CB CE }{{}}{{} =:Ā =: B n whch A p B p A =, B =, C = A c B c C p C c, E = B w. (11) The system flows as long as the event-trggerng condtons are not met,.e., as long as (8) holds for all {1,..., N}, whch can be reformulated as x C, wth C = { x R nx x Q x ε {1,..., N}}, (12) and Q = σ C Γ C, (13) Γ because x Q x ε s equvalent to Γ e(t) 2 σ Γ v(t) 2 + ε, as n (8). As mentoned before, when node transmts ts data, a reset accordng to e + = (I Γ )e occurs, whle x remans the same,.e., x + = x, see (3). July 18, 211

8 8 Ths can be expressed by for all x D, {1,..., N}, n whch x + = I x, (14) I Γ } {{ } =:Ḡ D = { x R nx x Q x = ε }, (15) accordng to (6). Combnng (1), (12), (14) and (15) yelds an mpulsve system of the form (9). C. Specal Cases: State Feedback and Centralsed Event Trggerng In the exstng lterature, the event-trggered control problem has mostly been consdered for state-feedback controlled systems, see, e.g., 18, 19. In ths case, the controller s gven by u = K ˆx p, (16) where ˆx p R np denotes the most recently sampled state of the plant and s defned n a smlar fashon as ˆv. We can regard ths as a specal case of the setup presented above. Namely, to formulate the event-trggered control system wth controller (16) as an mpulsve system, we combne (1), (3), (5), (6) and (16), where we take C p = I n (1), as all states are measurable, and take v := y = x p, ˆv := ŷ = ˆx p n (3), (5) and (6). In ths case, the resultng mpulsve model s gven by (9), wth (1), (12), (14) and (15), n whch x := x p e and A = A p, B = B p K, C = I, E = B w. (17) To arrve at the event trggerng mechansm that was proposed n 19, we take N = 1 and Γ 1 = I (.e., a centralsed event-trggerng mechansm), and ε 1 =. Remark II.2 Although we study event-trggerng condtons of the form (6), whch s an extenson of the one presented n 19, we can n prncple study every event-trggerng mechansm wth condtons that can be wrtten n the form x Q x = ε, such as the ones presented n 29. III. STABILITY AND L -GAIN In ths secton, we wll study stablty of the event-trggered control system n the sense of Lyapunov and ts L -gan. We wll frst revew some basc stablty and L -gan results for mpulsve systems of the form (9). A. Stablty and L -gan of Impulsve Systems Let us defne the notons of stablty and of Lyapunov functon canddate that can be used to analyse mpulsve systems of the form (9). Defnton III.1 23 Consder the mpulsve system, gven by (9) wth w = and a compact set A X. July 18, 211

9 9 The set A s sad to be stable for the mpulsve system (9) wth w =, f for each ε > there exsts δ >, such that mn x A x() x δ mples mn x A x(t) x ε, for all solutons x to the mpulsve system (9) wth w = and all t for whch the soluton x s defned. The set A s sad to be globally attractve f each soluton x to the mpulsve system (9), wth w =, satsfes mn x A x(t) x as t. The set A s globally asymptotcally stable for (9), wth w =, f t s stable and globally attractve. Defnton III.2 23 Consder the mpulsve system, gven by (9) wth w =, and a compact set A X. The functon W : X R s a Lyapunov functon canddate for the system (9) and the set A f the functon W () s contnuous and nonnegatve on (C N =1 D )\A X, () s locally Lpschtz on an open set O satsfyng C\A O X, () satsfes lm W ( x) =, and x A, x X (v) the sublevel sets of W on X are compact,.e., the sets { x X W ( x) c W } are compact for all c W >. To prove global asymptotc stablty of the set A of the system (9), we wll make use of the followng lemma. Lemma III.3 Consder the mpulsve system (9) wth w = and a compact set A X satsfyng Ḡ x A for all x D A, {1,..., N}. Assume that for w = and for all x X, a mnmum nter-event tme h mn > exsts for each {1,..., N},.e., t k +1 t k h mn for all k N, and assume there exsts a Lyapunov functon canddate W for the mpulsve system (9) wth w = and the set A X, such that dw ( x) d x Ā x <, for almost all x C\A, W (Ḡ x) W ( x), for all x D \A, {1,..., N}. (18a) (18b) Then, A s a globally asymptotcally stable set for the system (9) wth w =. Proof: The proof s gven n the Appendx. Let us now defne the noton of the L -gan of a system, whch was studed for LTI systems n, e.g., 3, for whch we ntroduce a performance varable z R nz gven by z = C x + Dw, (19) for some matrces C and D of approprate dmensons. Defnton III.4 The L -gan from w to z of the system (9), wth (19), s defned as κ = nf{ κ R + δ : X R +, such that z L κ w L + δ( x()), for all x() X, w L }, (2) n whch z s a soluton to (9) and (19) wth ntal condton x() X, and dsturbance w L. July 18, 211

10 1 B. Stablty and L -gan of the Event-Trggered Control System Usng the results presented above for mpulsve systems of the form (9), we now present the man result of ths secton. The man result conssts of condtons for stablty of a set A, and an explct expresson of ths set A, and an upper bound on the L -gan for the event-trggered control system (9), wth (1), (12), (13), (14) and (15), and (19). We wll also present smpler condtons to guarantee that A s a globally asymptotcally stable set for ths event-trggered control system for the case that the dsturbances are absent (.e., for w = ). Theorem III.5 Consder the event-trggered control system (9), wth (1), (12), (13), (14) and (15), and (19). Moreover, assume that for all x() X and all w L, a mnmum nter-event tme h mn > exsts for each {1,..., N},.e., t k +1 t k h mn for all k N. Now suppose there exst a postve defnte matrx P R (np+nc) (np+nc), a postve semdefnte matrx U R nx nx, scalars α, β, κ >, and µ, ν, {1,..., N}, satsfyng N =1 µ Q Ā P P Ā α P, (21a) B P βi α P C C D C (κ 2 β)i D, (21b) D P Ḡ for all {1,..., N}, n whch P := dag(p, ) + U. Then, P Ḡ ν Q, (21c) A = { x C ( N =1 D ) x N P x µ ε =1 α } (22) s a globally asymptotcally stable set for (9) wth w =. Moreover, the L -gan of (9), wth (19), s smaller than or equal to κ and δ( x()) n (2) can be taken as δ( x()) = (α x () P x() + N =1 µ ε ) 1/2 for x() X. Proof: The proof s gven n the Appendx. Let us now comment on the results presented n Theorem III.5. The frst comment s that the assumpton n the hypotheses of Theorem III.5 on the exstence of a strctly postve mnmum nter-event tme for each {1,..., N} s automatcally guaranteed, f ε > for all {1,..., N} and the LMIs n (21) are feasble. Ths wll be shown n Theorem IV.1 below. In case that ε = for some {1,..., N}, the assumpton on the exstence of a strctly postve mnmum nter-event tme for each {1,..., N} can be volated and the nter-event tmes t k +1 t k can converge to zero. In ths case, an nfnte number of events can occur n a fnte-length tme nterval (.e., the mpulsve system (9) exhbts Zeno behavour). Ths can happen at tmes t when v J (t) = and x(t), as we wll show n Example 2 n Secton VI. Therefore, we should generally take ε > for all {1,..., N} to guarantee mnmum nter-event tmes greater than zero. Another comment regardng Theorem III.5 s that feasblty of (21) s only determned by the choce of sutable α, β, κ, and σ, {1,..., N}, as Q depends on σ, and feasblty s not affected by the choce of ε, {1,..., N}. July 18, 211

11 11 Hence, once (21) s feasble, practcal stablty (for w = ) and the upper bound κ on the L -gan are guaranteed. The sze of the set A as n (22) (when w = ), s affected by α, κ and σ, through the resultng P, as well as ε. Hence, after choosng α, κ and σ that render the set A of the event-trggered control system globally asymptotcally stable and that guarantee the desred upper bound κ on the L -gan, the parameters ε can be freely chosen to adjust the sze of the set A. As we can see from (8), ths wll affect the number of events, enablng us to make trade-offs between the sze of the set A (related to the ultmate bound x reaches as t for w = ) and the number of transmssons n each channel. Indeed, larger ε, {1,..., N}, result n fewer events, and thus fewer transmssons, but n a larger set A (.e., a larger ultmate bound), when w =. In fact f ε, {1,..., N}, all approach zero (from above) we have that A {}. Hence, the set A can be made arbtrary small (at the cost of more transmssons). The nave choce to take ε =, for all {1,..., N}, seems appealng as t would yeld A = {}. However, as argued already above, ths mght result n zero mnmum nter-event tmes. In some cases, such as the case of a state-feedback controlled system wth centralsed event trggerng as dscussed n Secton II-C, a strctly postve mnmum nter-event tmes can guaranteed even for ε 1 =, and we have that A = {} s globally asymptotcally stable. We wll further dscuss the mnmum nter-event tmes below. Fnally, note that the functon δ s also affected by ε, also expressng that larger ε wll result n a larger ultmate bound (even for w ). In case dsturbances are absent (w = ), we can arrve at smpler condtons to guarantee that A s a globally asymptotcally stable set for the event-trggered control system (9), wth (1), (12), (13), (14) and (15). Corollary III.6 Consder the event-trggered control system (9), wth (1), (12), (13), (14) and (15), and w =. Moreover, assume that for all x() X, a mnmum nter-event tme h mn > exsts for each {1,..., N},.e., t k +1 t k h mn for all k N. Now suppose there exst a postve defnte matrx P R (np+nc) (np+nc), a postve semdefnte matrx U R nx nx, scalars α >, and µ, ν, {1,..., N}, satsfyng µ Q Ā P P Ā α P, (23a) =1 α P dag(i, ) (23b) and (21c) for all {1,..., N}, where P := dag(p, ) + U and I denotes the dentty matrx of sze (n p + n c ) (n p + n c ). Then, the set A as n (22) s a globally asymptotcally stable set for (9) wth w =. Furthermore, lm sup t x(t) ( N =1 µ ε ) 1/2. Proof: The proof follows the same lnes of reasonng as the proof of Theorem III.5. The fact that x(t) ( N =1 µ ε ) 1/2 as t follows from (52), wth w =, and the fact that (23b), mples that x(t) 2 αv ( x(t)) for all t R +. Remark III.7 In ths paper, we partcularly study the L -gan from w to z of the system (9), wth (19), nstead of the L p -gan from w to z, for some p N, defned as κ = nf{ κ R + δ : X R +, such that z Lp κ w Lp + δ( x()), for all x() X, w L p }, (24) July 18, 211

12 12 n whch z s a soluton to (9) and (19) for ntal condton x() X, and nput w L p. The reason for focussng on L -gans s that we are generally nterested n ε >, {1,..., N}, as ths guarantees nonzero mnmum nter-event tmes (see Theorem IV.1 below). In ths case, A {}, and thus x(t) wll not converge asymptotcally to the orgn for w =, and therefore z(t) wll typcally not converge to zero when t. Hence, z Lp = for all p. Consequently, a fnte L p -gan for p N cannot be guaranteed n case ε >, {1,..., N}. Snce the L -gan does not requre z(t) when t, but merely that z(t) s bounded for all t R +, we can arrve at a fnte L -gan for the event-trggered control system dscussed n ths paper. Note that n case ε =, {1,..., N}, for whch n some crcumstances t s possble to guarantee that h mn > (e.g., the case of a state-feedback controlled system wth centralsed event trggerng as dscussed n Secton II-C), the L p -gan mght be fnte snce n ths case A = {}. In partcular, the L 2 -gan s guaranteed to be smaller than κ for the system (9), wth (19) and ε = for all {1,..., N}, f there exst a postve defnte matrx P R (np+nc) (np+nc), a postve semdefnte matrx U R nx nx, such that P := dag(p, ) + U, scalars α >, and µ, ν, {1,..., N}, satsfyng N =1 µ Q Ā P P Ā α P B P κ 2 I, (25a) C D I P Ḡ P Ḡ ν Q, (25b) for all {1,..., N}. Of course, ths result only holds f all the mnmum nter-event tmes h mn, {1,..., N}, are strctly postve, as was also requred n Theorem III.5. IV. A LOWER BOUND ON THE INTER-EVENT TIMES In ths secton, we wll show that for each node {1,..., N}, the nter-event tmes t k +1 t k, k N, of the event-trggered control system are bounded from below by a strctly postve constant. The exstence of a lower bound on the nter-event tmes for every node means that the total number of transmssons n a fnte tme nterval s bounded from above, whch guarantees a certan maxmum utlsaton of the communcaton resources. We wll show that, although the stablty and L -gan propertes of the system hold globally, the guaranteed lower bound on the nter-event tmes s a local property, n the sense that t depends on the magntude of the ntal condton and the dsturbance. The analyss s based on studyng the solutons of (9), wth (1), (12), (13), (14) and (15), from t k to t k. To +1 do so, we study the solutons of the auxlary system d x I I dt = Ā e J Γ x Γ e J I (Ā + Γ Γ c e J c + Bw ) wth e + J (t k ) =, from t k to t k +1, for each {1,..., N}. In (26), the submatrces Γ := I J and Γ c := I J c, {1,..., N}, are formed by takng the columns of the dentty matrx I that are n the set J = {j {1,..., n y + n u } γ j = 1}, and n the set J c (26) = {1,..., n y + n u }\J, respectvely, and are used to select the July 18, 211

13 13 sgnals n e that correspond to node {1,..., N},.e., e J = Γ e, and that do not correspond to ths node,.e., e J c = ( Γ c ) e, respectvely. The auxlary system (26), {1,..., N}, s obtaned from (1), by consderng e J c as external nputs. Hence, the fact that the dynamcs of e J c depend on x and e J s gnored n (26), yeldng that the solutons to (9), wth (1), (12), (13), (14) and (15), from t k to t k +1 are ncluded n the solutons of (26) and e + J (t k ) = from t k to t k, for each {1,..., N}. Ths fact, and the fact that e +1 J c n (26) satsfes (8), wll be exploted to derve the lower bound on the mnmum nter-event tme. We now present the man result of ths secton. Theorem IV.1 Consder the event-trggered control system gven by (9), wth (1), (12), (13), (14) and (15), wth ε > for all {1,..., N}. For every δ x and every δ w, there exsts a strctly postve lower bound on the mnmum nter-event tmes h mn (δ x, δ w ) for each node {1,..., N},.e., t k +1 t k h mn, for all k N, for every soluton to (9) wth x() δ x, and w L δ w. An explct expresson for a lower bound h mn s gven by { ( I e I Γ mn h > λ max Ā h Q e Ā I h Γ I ) } ζ(h) η, (27) n whch ζ (h) = ε Q ( 2 ηρ (h) Ā e I Γ h I + ρ (h) ), (28) wth ρ (h) = h h ( e ϑs ds B δ w + Ā and η = λmax( P )αδ 2 x +βδ2 w + N =1 µε αλ mn(p ), ϑ = λ max (Ā Proof: The proof s gven n the Appendx. I Γ + Γ c I j I σ j Γ j C 2 η + ε j ) 2, (29) Γ Ā ), and I := {1,..., N}\{}. The mnmum n (27) n Theorem IV.1 can be solved by computng the maxmum egenvalue of the h-dependent matrx n the condton n (27) for ncreasng h > and check when the nequalty s satsfed for the frst tme. Ths determnes for node {1,..., N} the lower bound on the nter-event tmes h mn, as n (27). Even though a mnmum nter-event tmes s guaranteed for each node, no guarantees can be made about the tme between two events n dfferent nodes. Stll, the lower bound on the nter-event tmes for all nodes allows guarantee to be made about the total number of events wthn a certan tme nterval. The obtaned lower bounds decrease as δ x (whch s related to x() ) ncreases or as δ w (whch s related to w L ) ncreases, mplyng that the control task has to be executed more often f the system s ntal state s further away from the orgn and n case the norm of the dsturbance s larger. We wll llustrate ths observaton n Example 3 of Secton VI. In the specal case that C p and C c are nvertble and the event trggerng s centralsed,.e., the number of nodes N = 1 and Γ 1 = I (mplyng that Q 1 has full rank), and no dsturbances are present (mplyng that δ w = ), the mnmum nter-event tme h 1 mn > even for ε 1 =. Furthermore, we have that ρ 1 (h) = (due to δ w = and I 1 = ), and thus that ζ 1 (h) =, for all h R +, meanng that the obtaned bound s ndependent of δ x (and thus ndependent of x() X ). If July 18, 211

14 14 addtonally the controller s gven by a state-feedback controller (16), the resultng condton recovers the one presented n Theorem 5.1 n 31. In ths case, the bounds are tght n the sense that for some k 1 N, we have that t 1 k 1+1 t1 k 1 = h 1 mn. V. IMPROVED EVENT-TRIGGERING CONDITIONS In the prevous sectons, we modelled the event-trggered control system as an mpulsve system and presented condtons to guarantee ts stablty and an upper bound on the L -gan. The reason to take an mpulsve system approach s that t explctly descrbes the behavour of the event-trggered control system. Ths has the favourable consequence that t yelds less conservatve condtons than the (drect extensons of the) exstng results n the lterature, such as 18, 19. To formally demonstrate ths statement, we frst extend the reasonng of 19 towards dynamcal output-based controllers and by ncludng L -performance, and secondly, we show that the obtaned stablty condtons can be seen as a specal case of the condtons n Theorem III.5 (.e., usng the mpulsve system descrpton of the event-trggered control system). To extend the work of 19, let us consder the followng auxlary system: d dt B x = (A + BC)x + E e, w v = C x + e, z C x D w whch s obtaned from (1) and (19) by consderng the error e as an external nput, nstead of as a state varable as n (9), and by assumng that the performance output, as n (19), s gven by z = C x x + Dw, mplyng that C n (19) has the form C = C x (3). As mportant observaton s that n (3), the dynamcs of e, gven by d dt e = d dtv = C(A + BC)x CBe, s gnored n (3), whle t s captured explctly n the mpulsve system (9), wth (1), (12), (13), (14) and (15). System (3), wth e = and w =, s a globally asymptotcally stable LTI system, because of the assumpton made n Secton II that the controller stablses the plant when ˆv(t) = ŷ (t) û (t) = y (t) u (t) = v(t) for all t R +, meanng that A + BC s Hurwtz,.e., t has all the egenvalues n the left-half complex plane. Ths ensures that there exst a postve defnte storage functon of the form V (x) = x P x, see, e.g., 32, a (suffcently small) postve scalar α, (suffcently large) postve scalars β, κ, satsfyng β κ 2, and (suffcently small) postve scalars σ, {1,..., N}, that satsfy the dsspaton nequalty d dt V (x(t)) αv (x(t)) + ( β w(t) σ e J (t) 2 v J (t) 2) (31) and the nequalty =1 z(t) 2 + (β κ 2 ) w(t) 2 αv (x(t)). (32) Now snce (8) holds, for all {1,..., N} and all t R +, we have that ( vj (t) 2 + ε σ 1 σ e J (t) 2), (33) =1 July 18, 211

15 15 and all t R +. Combnng ths expresson wth (31) yelds that d dt V (x(t)) αv (x(t)) + β w(t) 2 + =1 ε σ, (34) whch allows us to show that for w = and for V (x(t)) N ε =1 σ, t holds that d α dtv (x(t)) <, whch means that the state x of (3), wth (8), converges asymptotcally to the set A = {x R np+nc V (x) N =1 ε σ α }. Furthermore, usng (32) and deas from 3, we can show that the system (3), wth (8), has a fnte L -gan from dsturbance w to performance output z. We wll formalse ths dea n the followng theorem. Theorem V.1 Assume that there exst scalars α, β, κ, σ P R (np+nc) (np+nc), satsfyng Z αp N =1 C Γ C B N P =1 1 σ Γ, E P βi αp C x C x D C x (κ 2 β)i D, D wth Z := (A + BC) P + P (A + BC). Then, the set >, {1,..., N}, and a postve defnte matrx (35a) (35b) A = {x R np+nc x P x N ε =1 σ } (36) α s a globally asymptotcally stable set of the system (3) wth (8) and for w =. Furthermore, the L -gan from w to z s smaller than or equal to κ and δ n (2) can be taken as δ(x()) = (αx ()P x() + N ε =1 σ ) 1/2 for all x R np+nc. Proof: The proof follows drectly from the dscusson above and the fact that (35a) and (35b) mply (31) and (32), respectvely. It also follows drectly from Theorem V.2 that we wll present below. In case the system s controlled by a state-feedback controller (as dscussed n Secton II-C), the event trggerng mechansm s centralsed (.e., N = 1 and Γ 1 = I), and when dsturbances are absent (E = and D = ), the condtons presented n Theorem V.1 provde LMI-based stablty condtons that can be used to analyse the stablty of the event-trggered control system studed n 19. Even though the results n 19 are vald for nonlnear systems as well, whle we focus n ths paper on lnear systems, Theorem V.1 provdes a computatonal procedure that allows us to obtan large values for σ and, thus, large values for the nter-event tmes, whereas 19 only presents exstence results and does not provde a constructve (optmsaton-based) way to obtan sutable choces for σ. Note that obtanng a LMI-based stablty analyss for the case studed n 19 s not the man result of ths secton. Namely, the man result of ths secton s presented below. Ths man result states that f Theorem V.1 guarantees global asymptotc stablty of the set A, as n (36), and guarantees an upper bound κ on the L -gan for the system (3) wth (8), for some scalars σ and the scalars ε, {1,..., N}, then, global asymptotc stablty of the same July 18, 211

16 16 set A and the same upper-bound κ on the L -gan can also be guaranteed for the mpulsve system (9) usng Theorem III.5. Theorem V.2 Consder the model of the event-trggered control system (3), wth (8), and the mpulsve system formulaton of the event-trggered control system (9), wth (1), (12), (13), (14) and (15), and (19) wth C = C x. If there exsts a postve defnte matrx P, and scalars α, β, κ, σ >, {1,..., N}, satsfyng the condtons of Theorem V.1, then P := dag(p, ), U =, µ = 1 σ and ν =, for all {1,..., N}, satsfy the condtons of Theorem III.5 for the same α, β, and κ. Proof: The proof s gven n the Appendx. Theorem V.2 formally shows that the condtons based on mpulsve system (9) are never more conservatve than the ones based on system (3), as the matrx P n Theorem III.5 can have a more general form than P := dag(p, ) (whch was used n the hypothess of Theorem V.2). Hence, ths creates the opportunty to guarantee stablty for event-trggerng condtons wth a larger nter-event tme or a smaller upper-bound on the L -gan. Furthermore, Theorem V.2 also guarantees that the condtons n Theorem III.5 can always be satsfed f all σ, {1,..., N}, are chosen suffcently small. Namely, for the auxlary system (3) the exstence of storage functon of the form V (x) = x P x satsfyng (31) for some postve α, β and σ, {1,..., N}, s guaranteed. Hence, the hypothess of Theorem V.1 can always be satsfed for some suffcently small α and σ, {1,..., N}, and some suffcently large β and κ, whch, n turn, mples feasblty of the condtons n Theorem III.5. VI. ILLUSTRATIVE EXAMPLES In ths secton, we llustrate the presented theory usng three numercal examples. The frst example s taken from 19, n whch an unstable plant s stablsed usng an event-trggered mplementaton of a state-feedback controller and a centralsed event-trggerng mechansm. We wll show that by formulatng the event-trggered control system as an mpulsve system and employng the theory as developed n ths paper, we can guarantee stablty for eventtrggered control systems wth larger mnmum nter-event tmes. In the second example, we stablse an unstable plant usng a dynamcal output-based controller and a decentralsed event-trggerng mechansm to llustrate that ndeed output-based controllers and decentralsed event trggerng can be desgned that perform well. In the last example, we consder a stable plant that s subject to dsturbances and show that outputs of the plant and the controller are only transmtted when dsturbances are actng on the system or durng transents, whle no transmssons occur when dsturbances are absent and the system s n steady state. Ths s a favourable property that tradtonal dgtal control systems wth perodc transmssons do not have. Example 1: Let us consder the numercal example taken from 19. The plant (1) s gven by d dt x p = 1 x p + u, (37) July 18, 211

17 the state-feedback controller s gven by (16), wth K = 1 4, and the event trggerng s centralsed,.e., we have that N = 1 and Γ 1 = I. In 19, global asymptotc stablty of the orgn s guaranteed for σ.55 for the event-trggerng condton e = σ x and was obtaned usng an alternatve approach. Ths yelds σ 1 =.55 2 =.3 and ε 1 = f the event-trggerng mechansm s formulated as n (6). For ths event-trggerng mechansm, Theorem IV.1, or ts counterpart Theorem 5.1 n 31, yelds a lower bound on the nter-event tmes 3 of.318. We now compare ths result wth the event-trggerng mechansm obtaned usng the results from Secton V,.e., obtaned by maxmsng σ 1 n the condtons of Theorem V.1, modfed accordng to Remark??. Takng ths approach allows us to guarantee stablty up to σ 1 =.273, resultng, for ε 1 =, n a lower bound on the nter-event tmes of.84. Therefore, we can conclude that takng the approach as n Secton V already ncreases the allowable mnmum nter-event tme wth respect to 19. However, f we analyse stablty usng the result of Theorem III.5, whch s based on the mpulsve system formulaton, we can guarantee stablty of ths event-trggered control system up to σ 1 =.588, whch yelds, for ε 1 =, a lower bound on the nter-event tmes of The ncrease of nter-event tmes s expected due to the formal result establshed n Theorem V.2. We therefore conclude that by modellng the event-trggered control system usng an mpulsve model, whch truly descrbes the behavour of ths event-trggered control system, stablty can be guaranteed for event-trggerng mechansms that yeld larger mnmum nter-event tmes. Example 2: Let us now consder the plant (1) gven by d dt x p = 1 x p + û y = 1 4 x p, and the controller (2) gven by d dt x c = 1 x c + ŷ 5 1 u = 1 4 x c. We assume that no dsturbances act on the plant,.e., B w =, and, therefore, we smply take C = and D =. Furthermore, we assume that the system s equpped wth an event-trggerng mechansms at both the sensorto-controller channel and the controller-to-actuator channel, whch means that we defne Γ 1 17 (38) (39) = dag(1, ) and Γ 2 = dag(, 1). Hence, we have two nodes. Practcal stablty of the event-trggered control system (1), (2), wth event-trggerng mechansm (6), wth σ 1 = σ 2 = 1 3, can be guaranteed usng the mpulsve system formulaton (9) and the results of Corollary III.6. If we take ε 1 = ε 2 =, the nter-event tmes wll become zero when v J (t) =, at some tme t R + and for some {1, 2}, as was dscussed n Secton III and Secton IV. By smulatng the response of the system to the ntal condton x() = 25 2, 25 2, 25 2, 25 2,,, we can observe that ndeed the nter-event tmes converge to 3 Note that for ths example, the obtaned lower-bound on the mnmum nter-event tmes s tght, as also was observed n Secton IV. July 18, 211

18 tempmage temp 211/4/2 tempmage temp 1:21 page 1 #1 211/4/2 1:2 page 1 #1 replacemen 18 outputs y and u y, node = 1 u, node = tme t nter-event tme t k+1 t k node = 1 node = tme t Fg. 2: Evoluton of y (node = 1) and u (node = 2) as a functon of tme (left) and the nter-event tmes for each node as a functon of tme (rght) for Example 2 wth ε 1 = ε 2 =. zero as v J2 (t) = u(t), see Fg. 2. Hence, ths llustrates that ε > s needed for all {1,..., N} to guarantee mnmum nter-event tmes larger than zero. If we now take ε 1 = ε 2 = 1 3, Corollary III.6 yelds that the states x(t) satsfy lm sup t x(t) 6.4. Usng the result of Theorem IV.1, we obtan that f the ntal condtons satsfy, e.g., x() 25, a lower bound on the nter-event tmes h 1 mn = h2 mn = s guaranteed for both nodes. When we compare these results wth a smulaton of the response of the system to the ntal condton x() = 25 2, 25 2, 25 2, 25 2,,, see Fg. 3, we observe that the states of the plant and the controller ndeed converge asymptotcally to a vcnty of the orgn and that the outputs of the plants and controllers have to be transmtted less often when the state approaches the orgn. However, x(t) even satsfes lm sup t x(t).12, whch s sgnfcantly smaller than the predcted upper bound of approxmately 6.4. In addton, the observed mnmum nter-event tme s h 1 mn h2 mn 1 4, whch s larger than the predcted value of Ths seems to hold for many ntal condtons satsfyng x() < 25. Ths shows that, although we can formally prove the exstence of a globally asymptotcally stable compact set and a nonzero lower bound on the mnmal nter-event tmes, the obtaned bounds can stll be mproved. Improvng these bounds s a topc of future research. Example 3: Let us now consder the (stable) plant (1) gven by d dt x p = 1 x p + û + w y = 1 x p, and the controller (2) gven by d dt x c = 2 1 x c + 2 ŷ u = 5 2 x c. Furthermore, we take C = 1 and D = for the performance output z n (19), and assume that the system s equpped wth event-trggerng mechansms at both the sensor-to-controller channel and the controller-to-actuator (4) (41) July 18, 211

19 tempmage temp 211/4/2 1:19 page 1 # tempmage temp 211/4/2 1:2 page 1 #1 states xp and xc tempmage temp 211/4/2 1:18 page 1 # tme t 1 nter-event tme t k+1 t k node = 1 node = Fg. 3: Evoluton of the states of the plant and controller as a functon of tme (top) and the nter-event tmes for each node as a functon of tme (bottom) for Example 2 wth ε 1 = ε 2 = 1 3. Inserted n the top fgure: A close-up tme t of the evoluton of the states of the plant and the controller at tmes t 17.5, 3 channel. Ths means that we, agan, defne Γ 1 = dag(1, ) and Γ 2 = dag(, 1). Asymptotc stablty of the compact set A and an upper bound of the L -gan of the event-trggered control system (1), (2), wth eventtrggerng mechansm (6), wth σ 1 = σ 1 = 1 3, can be guaranteed usng the mpulsve system formulaton (9) and the results of Theorem III.5. Ths leads to the smallest upper bound on the L -gan, gven by κ =.46. When we smulate the response of the system to the ntal condton x() = and a dsturbance satsfyng w(t) 1 for tme t, 1 and w(t) 1 4 for tme t 2, 3, as shown n Fg. 4, we obtan the trajectores of z as also shown n Fg. 4. In ths fgure, we can observe that the performance output z, as n (19), satsfes z(t).3 for tme t, 1 and z(t).9 for tme t 2, 3, whch satsfes z L κ w L + δ() =.46 w L +.17, whch s an upper bound of the L -gan of Defnton III.4. Furthermore, we can also observe that the nter-event tmes are larger than.22 for t, 1 and larger than.44 for t 2, 3. Ths observaton concurs wth the result of Secton IV, whch stated that transmssons occur less often f the magntude of the dsturbance s smaller. Fnally, snce the system (4) s stable, t seems that the outputs of the plant and controller only have to be transmtted when dsturbances are actng on the system or durng transents (.e., approxmately for t < 1 and t > 2), and no transmssons occur when no dsturbances are actng on the system and the systems s close to ts steady state. One could say that event-trggered control only acts when t s July 18, 211

20 tempmage temp 211/4/2 1:2 page 1 #1 2 dsturbance w and output z output z dsturbance w tempmage temp 211/4/2 1:2 page 1 # tme t nter-event tme t k+1 t k node = 1 node = tme t Fg. 4: Evoluton of the dsturbance w and the output z as a functon of tme (top) and the nter-event tmes for each node as a functon of tme (bottom) for Example 3. necessary from a stablty or performance pont of vew, whch s a favourable property that makes event-trggered control of hgh nterest. Tradtonal dgtal control systems wth perodc transmssons do not have ths appealng property. VII. CONCLUSIONS In ths paper, we studed stablty and L -performance of event-trggered control systems for dynamcal outputbased controllers havng decentralsed event-trggerng mechansms. The proposed event-trggerng mechansm unfes earler proposals for event-trggerng mechansms, whch were manly appled to state-feedback controllers. Va an example (Example 2), we showed that drect extensons of exstng event-trggerng mechansms for outputbased controllers and decentralsed event trggerng are not applcable, as they result n nter-event tmes that converge to zero. Such Zeno behavour s obvously undesrable n practcal mplementatons and, therefore, extensons as proposed n ths paper are necessary. To analyse the resultng event-trggered control system, we modelled the event-trggered control system as an mpulsve system that truly descrbes the behavour of the event-trggered control system. The stablty and L - performance are then analysed usng lnear matrx nequaltes. In addton, we provded expressons for lower bounds on the mnmum nter-event tmes and we formally proved that by usng an mpulsve systems approach, July 18, 211

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

Errors for Linear Systems

Errors for Linear Systems Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

MMA and GCMMA two methods for nonlinear optimization

MMA and GCMMA two methods for nonlinear optimization MMA and GCMMA two methods for nonlnear optmzaton Krster Svanberg Optmzaton and Systems Theory, KTH, Stockholm, Sweden. krlle@math.kth.se Ths note descrbes the algorthms used n the author s 2007 mplementatons

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

The Order Relation and Trace Inequalities for. Hermitian Operators

The Order Relation and Trace Inequalities for. Hermitian Operators Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence

More information

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016 U.C. Berkeley CS94: Spectral Methods and Expanders Handout 8 Luca Trevsan February 7, 06 Lecture 8: Spectral Algorthms Wrap-up In whch we talk about even more generalzatons of Cheeger s nequaltes, and

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

STAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 16

STAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 16 STAT 39: MATHEMATICAL COMPUTATIONS I FALL 218 LECTURE 16 1 why teratve methods f we have a lnear system Ax = b where A s very, very large but s ether sparse or structured (eg, banded, Toepltz, banded plus

More information

Appendix B. The Finite Difference Scheme

Appendix B. The Finite Difference Scheme 140 APPENDIXES Appendx B. The Fnte Dfference Scheme In ths appendx we present numercal technques whch are used to approxmate solutons of system 3.1 3.3. A comprehensve treatment of theoretcal and mplementaton

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Time-Varying Systems and Computations Lecture 6

Time-Varying Systems and Computations Lecture 6 Tme-Varyng Systems and Computatons Lecture 6 Klaus Depold 14. Januar 2014 The Kalman Flter The Kalman estmaton flter attempts to estmate the actual state of an unknown dscrete dynamcal system, gven nosy

More information

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng

More information

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0 MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Maximizing the number of nonnegative subsets

Maximizing the number of nonnegative subsets Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum

More information

Solutions to exam in SF1811 Optimization, Jan 14, 2015

Solutions to exam in SF1811 Optimization, Jan 14, 2015 Solutons to exam n SF8 Optmzaton, Jan 4, 25 3 3 O------O -4 \ / \ / The network: \/ where all lnks go from left to rght. /\ / \ / \ 6 O------O -5 2 4.(a) Let x = ( x 3, x 4, x 23, x 24 ) T, where the varable

More information

APPENDIX A Some Linear Algebra

APPENDIX A Some Linear Algebra APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,

More information

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg

princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg prnceton unv. F 17 cos 521: Advanced Algorthm Desgn Lecture 7: LP Dualty Lecturer: Matt Wenberg Scrbe: LP Dualty s an extremely useful tool for analyzng structural propertes of lnear programs. Whle there

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Bayesian predictive Configural Frequency Analysis

Bayesian predictive Configural Frequency Analysis Psychologcal Test and Assessment Modelng, Volume 54, 2012 (3), 285-292 Bayesan predctve Confgural Frequency Analyss Eduardo Gutérrez-Peña 1 Abstract Confgural Frequency Analyss s a method for cell-wse

More information

Difference Equations

Difference Equations Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1

More information

10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014)

10-801: Advanced Optimization and Randomized Methods Lecture 2: Convex functions (Jan 15, 2014) 0-80: Advanced Optmzaton and Randomzed Methods Lecture : Convex functons (Jan 5, 04) Lecturer: Suvrt Sra Addr: Carnege Mellon Unversty, Sprng 04 Scrbes: Avnava Dubey, Ahmed Hefny Dsclamer: These notes

More information

Perron Vectors of an Irreducible Nonnegative Interval Matrix

Perron Vectors of an Irreducible Nonnegative Interval Matrix Perron Vectors of an Irreducble Nonnegatve Interval Matrx Jr Rohn August 4 2005 Abstract As s well known an rreducble nonnegatve matrx possesses a unquely determned Perron vector. As the man result of

More information

Economics 101. Lecture 4 - Equilibrium and Efficiency

Economics 101. Lecture 4 - Equilibrium and Efficiency Economcs 0 Lecture 4 - Equlbrum and Effcency Intro As dscussed n the prevous lecture, we wll now move from an envronment where we looed at consumers mang decsons n solaton to analyzng economes full of

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

Some modelling aspects for the Matlab implementation of MMA

Some modelling aspects for the Matlab implementation of MMA Some modellng aspects for the Matlab mplementaton of MMA Krster Svanberg krlle@math.kth.se Optmzaton and Systems Theory Department of Mathematcs KTH, SE 10044 Stockholm September 2004 1. Consdered optmzaton

More information

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski

EPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski EPR Paradox and the Physcal Meanng of an Experment n Quantum Mechancs Vesseln C Nonnsk vesselnnonnsk@verzonnet Abstract It s shown that there s one purely determnstc outcome when measurement s made on

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Lecture 12: Discrete Laplacian

Lecture 12: Discrete Laplacian Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly

More information

The Minimum Universal Cost Flow in an Infeasible Flow Network

The Minimum Universal Cost Flow in an Infeasible Flow Network Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran

More information

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal Inner Product Defnton 1 () A Eucldean space s a fnte-dmensonal vector space over the reals R, wth an nner product,. Defnton 2 (Inner Product) An nner product, on a real vector space X s a symmetrc, blnear,

More information

Improved delay-dependent stability criteria for discrete-time stochastic neural networks with time-varying delays

Improved delay-dependent stability criteria for discrete-time stochastic neural networks with time-varying delays Avalable onlne at www.scencedrect.com Proceda Engneerng 5 ( 4456 446 Improved delay-dependent stablty crtera for dscrete-tme stochastc neural networs wth tme-varyng delays Meng-zhuo Luo a Shou-mng Zhong

More information

2.3 Nilpotent endomorphisms

2.3 Nilpotent endomorphisms s a block dagonal matrx, wth A Mat dm U (C) In fact, we can assume that B = B 1 B k, wth B an ordered bass of U, and that A = [f U ] B, where f U : U U s the restrcton of f to U 40 23 Nlpotent endomorphsms

More information

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009 College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

Appendix B. Criterion of Riemann-Stieltjes Integrability

Appendix B. Criterion of Riemann-Stieltjes Integrability Appendx B. Crteron of Remann-Steltes Integrablty Ths note s complementary to [R, Ch. 6] and [T, Sec. 3.5]. The man result of ths note s Theorem B.3, whch provdes the necessary and suffcent condtons for

More information

Representation theory and quantum mechanics tutorial Representation theory and quantum conservation laws

Representation theory and quantum mechanics tutorial Representation theory and quantum conservation laws Representaton theory and quantum mechancs tutoral Representaton theory and quantum conservaton laws Justn Campbell August 1, 2017 1 Generaltes on representaton theory 1.1 Let G GL m (R) be a real algebrac

More information

Notes on Frequency Estimation in Data Streams

Notes on Frequency Estimation in Data Streams Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to

More information

Digital Signal Processing

Digital Signal Processing Dgtal Sgnal Processng Dscrete-tme System Analyss Manar Mohasen Offce: F8 Emal: manar.subh@ut.ac.r School of IT Engneerng Revew of Precedent Class Contnuous Sgnal The value of the sgnal s avalable over

More information

On Event-Triggered Adaptive Architectures for Decentralized and Distributed Control of Large- Scale Modular Systems

On Event-Triggered Adaptive Architectures for Decentralized and Distributed Control of Large- Scale Modular Systems Unversty of South Florda Scholar Commons Mechancal Engneerng Faculty Publcatons Mechancal Engneerng 26 On Event-Trggered Adaptve Archtectures for Decentralzed and Dstrbuted Control of Large- Scale Modular

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

NUMERICAL DIFFERENTIATION

NUMERICAL DIFFERENTIATION NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the

More information

Perfect Competition and the Nash Bargaining Solution

Perfect Competition and the Nash Bargaining Solution Perfect Competton and the Nash Barganng Soluton Renhard John Department of Economcs Unversty of Bonn Adenauerallee 24-42 53113 Bonn, Germany emal: rohn@un-bonn.de May 2005 Abstract For a lnear exchange

More information

Finite Element Modelling of truss/cable structures

Finite Element Modelling of truss/cable structures Pet Schreurs Endhoven Unversty of echnology Department of Mechancal Engneerng Materals echnology November 3, 214 Fnte Element Modellng of truss/cable structures 1 Fnte Element Analyss of prestressed structures

More information

Neuro-Adaptive Design - I:

Neuro-Adaptive Design - I: Lecture 36 Neuro-Adaptve Desgn - I: A Robustfyng ool for Dynamc Inverson Desgn Dr. Radhakant Padh Asst. Professor Dept. of Aerospace Engneerng Indan Insttute of Scence - Bangalore Motvaton Perfect system

More information

Grover s Algorithm + Quantum Zeno Effect + Vaidman

Grover s Algorithm + Quantum Zeno Effect + Vaidman Grover s Algorthm + Quantum Zeno Effect + Vadman CS 294-2 Bomb 10/12/04 Fall 2004 Lecture 11 Grover s algorthm Recall that Grover s algorthm for searchng over a space of sze wors as follows: consder the

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Formulas for the Determinant

Formulas for the Determinant page 224 224 CHAPTER 3 Determnants e t te t e 2t 38 A = e t 2te t e 2t e t te t 2e 2t 39 If 123 A = 345, 456 compute the matrx product A adj(a) What can you conclude about det(a)? For Problems 40 43, use

More information

Chapter 11: Simple Linear Regression and Correlation

Chapter 11: Simple Linear Regression and Correlation Chapter 11: Smple Lnear Regresson and Correlaton 11-1 Emprcal Models 11-2 Smple Lnear Regresson 11-3 Propertes of the Least Squares Estmators 11-4 Hypothess Test n Smple Lnear Regresson 11-4.1 Use of t-tests

More information

PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY

PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY POZNAN UNIVE RSITY OF TE CHNOLOGY ACADE MIC JOURNALS No 86 Electrcal Engneerng 6 Volodymyr KONOVAL* Roman PRYTULA** PARTICIPATION FACTOR IN MODAL ANALYSIS OF POWER SYSTEMS STABILITY Ths paper provdes a

More information

Note 10. Modeling and Simulation of Dynamic Systems

Note 10. Modeling and Simulation of Dynamic Systems Lecture Notes of ME 475: Introducton to Mechatroncs Note 0 Modelng and Smulaton of Dynamc Systems Department of Mechancal Engneerng, Unversty Of Saskatchewan, 57 Campus Drve, Saskatoon, SK S7N 5A9, Canada

More information

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 )

Yong Joon Ryang. 1. Introduction Consider the multicommodity transportation problem with convex quadratic cost function. 1 2 (x x0 ) T Q(x x 0 ) Kangweon-Kyungk Math. Jour. 4 1996), No. 1, pp. 7 16 AN ITERATIVE ROW-ACTION METHOD FOR MULTICOMMODITY TRANSPORTATION PROBLEMS Yong Joon Ryang Abstract. The optmzaton problems wth quadratc constrants often

More information

Event-Triggering in Distributed Networked Systems with Data Dropouts and Delays

Event-Triggering in Distributed Networked Systems with Data Dropouts and Delays Event-Trggerng n Dstrbuted Networked Systems wth Data Dropouts and Delays Xaofeng Wang and Mchael D. Lemmon Unversty of Notre Dame, Department of Electrcal Engneerng, Notre Dame, IN, 46556, USA, xwang13,lemmon@nd.edu

More information

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

More information

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud Resource Allocaton wth a Budget Constrant for Computng Independent Tasks n the Cloud Wemng Sh and Bo Hong School of Electrcal and Computer Engneerng Georga Insttute of Technology, USA 2nd IEEE Internatonal

More information

The lower and upper bounds on Perron root of nonnegative irreducible matrices

The lower and upper bounds on Perron root of nonnegative irreducible matrices Journal of Computatonal Appled Mathematcs 217 (2008) 259 267 wwwelsevercom/locate/cam The lower upper bounds on Perron root of nonnegatve rreducble matrces Guang-Xn Huang a,, Feng Yn b,keguo a a College

More information

5 The Rational Canonical Form

5 The Rational Canonical Form 5 The Ratonal Canoncal Form Here p s a monc rreducble factor of the mnmum polynomal m T and s not necessarly of degree one Let F p denote the feld constructed earler n the course, consstng of all matrces

More information

Tokyo Institute of Technology Periodic Sequencing Control over Multi Communication Channels with Packet Losses

Tokyo Institute of Technology Periodic Sequencing Control over Multi Communication Channels with Packet Losses oyo Insttute of echnology Fujta Laboratory oyo Insttute of echnology erodc Sequencng Control over Mult Communcaton Channels wth acet Losses FL6-7- /8/6 zwrman Gusrald oyo Insttute of echnology Fujta Laboratory

More information

Chapter - 2. Distribution System Power Flow Analysis

Chapter - 2. Distribution System Power Flow Analysis Chapter - 2 Dstrbuton System Power Flow Analyss CHAPTER - 2 Radal Dstrbuton System Load Flow 2.1 Introducton Load flow s an mportant tool [66] for analyzng electrcal power system network performance. Load

More information

Physics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1

Physics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1 P. Guterrez Physcs 5153 Classcal Mechancs D Alembert s Prncple and The Lagrangan 1 Introducton The prncple of vrtual work provdes a method of solvng problems of statc equlbrum wthout havng to consder the

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

4DVAR, according to the name, is a four-dimensional variational method.

4DVAR, according to the name, is a four-dimensional variational method. 4D-Varatonal Data Assmlaton (4D-Var) 4DVAR, accordng to the name, s a four-dmensonal varatonal method. 4D-Var s actually a drect generalzaton of 3D-Var to handle observatons that are dstrbuted n tme. The

More information

The Feynman path integral

The Feynman path integral The Feynman path ntegral Aprl 3, 205 Hesenberg and Schrödnger pctures The Schrödnger wave functon places the tme dependence of a physcal system n the state, ψ, t, where the state s a vector n Hlbert space

More information

34 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 62, NO. 1, JANUARY 2017

34 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 62, NO. 1, JANUARY 2017 34 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 6, NO. 1, JANUARY 017 Output-Based and Decentralzed Dynamc Event-Trggered Control Wth Guaranteed L p -Gan Performance and Zeno-Freeness V. S. Dolk, D. P.

More information

Adaptive Consensus Control of Multi-Agent Systems with Large Uncertainty and Time Delays *

Adaptive Consensus Control of Multi-Agent Systems with Large Uncertainty and Time Delays * Journal of Robotcs, etworkng and Artfcal Lfe, Vol., o. (September 04), 5-9 Adaptve Consensus Control of Mult-Agent Systems wth Large Uncertanty and me Delays * L Lu School of Mechancal Engneerng Unversty

More information

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2 Salmon: Lectures on partal dfferental equatons 5. Classfcaton of second-order equatons There are general methods for classfyng hgher-order partal dfferental equatons. One s very general (applyng even to

More information

Introduction. - The Second Lyapunov Method. - The First Lyapunov Method

Introduction. - The Second Lyapunov Method. - The First Lyapunov Method Stablty Analyss A. Khak Sedgh Control Systems Group Faculty of Electrcal and Computer Engneerng K. N. Toos Unversty of Technology February 2009 1 Introducton Stablty s the most promnent characterstc of

More information

Norm Bounds for a Transformed Activity Level. Vector in Sraffian Systems: A Dual Exercise

Norm Bounds for a Transformed Activity Level. Vector in Sraffian Systems: A Dual Exercise ppled Mathematcal Scences, Vol. 4, 200, no. 60, 2955-296 Norm Bounds for a ransformed ctvty Level Vector n Sraffan Systems: Dual Exercse Nkolaos Rodousaks Department of Publc dmnstraton, Panteon Unversty

More information

COS 521: Advanced Algorithms Game Theory and Linear Programming

COS 521: Advanced Algorithms Game Theory and Linear Programming COS 521: Advanced Algorthms Game Theory and Lnear Programmng Moses Charkar February 27, 2013 In these notes, we ntroduce some basc concepts n game theory and lnear programmng (LP). We show a connecton

More information

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011 Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected

More information

= z 20 z n. (k 20) + 4 z k = 4

= z 20 z n. (k 20) + 4 z k = 4 Problem Set #7 solutons 7.2.. (a Fnd the coeffcent of z k n (z + z 5 + z 6 + z 7 + 5, k 20. We use the known seres expanson ( n+l ( z l l z n below: (z + z 5 + z 6 + z 7 + 5 (z 5 ( + z + z 2 + z + 5 5

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin Proceedngs of the 007 Wnter Smulaton Conference S G Henderson, B Bller, M-H Hseh, J Shortle, J D Tew, and R R Barton, eds LOW BIAS INTEGRATED PATH ESTIMATORS James M Calvn Department of Computer Scence

More information

High resolution entropy stable scheme for shallow water equations

High resolution entropy stable scheme for shallow water equations Internatonal Symposum on Computers & Informatcs (ISCI 05) Hgh resoluton entropy stable scheme for shallow water equatons Xaohan Cheng,a, Yufeng Ne,b, Department of Appled Mathematcs, Northwestern Polytechncal

More information

Remarks on the Properties of a Quasi-Fibonacci-like Polynomial Sequence

Remarks on the Properties of a Quasi-Fibonacci-like Polynomial Sequence Remarks on the Propertes of a Quas-Fbonacc-lke Polynomal Sequence Brce Merwne LIU Brooklyn Ilan Wenschelbaum Wesleyan Unversty Abstract Consder the Quas-Fbonacc-lke Polynomal Sequence gven by F 0 = 1,

More information

Uncertainty and auto-correlation in. Measurement

Uncertainty and auto-correlation in. Measurement Uncertanty and auto-correlaton n arxv:1707.03276v2 [physcs.data-an] 30 Dec 2017 Measurement Markus Schebl Federal Offce of Metrology and Surveyng (BEV), 1160 Venna, Austra E-mal: markus.schebl@bev.gv.at

More information

Convergence of random processes

Convergence of random processes DS-GA 12 Lecture notes 6 Fall 216 Convergence of random processes 1 Introducton In these notes we study convergence of dscrete random processes. Ths allows to characterze phenomena such as the law of large

More information

General viscosity iterative method for a sequence of quasi-nonexpansive mappings

General viscosity iterative method for a sequence of quasi-nonexpansive mappings Avalable onlne at www.tjnsa.com J. Nonlnear Sc. Appl. 9 (2016), 5672 5682 Research Artcle General vscosty teratve method for a sequence of quas-nonexpansve mappngs Cuje Zhang, Ynan Wang College of Scence,

More information

Lecture 4: November 17, Part 1 Single Buffer Management

Lecture 4: November 17, Part 1 Single Buffer Management Lecturer: Ad Rosén Algorthms for the anagement of Networs Fall 2003-2004 Lecture 4: November 7, 2003 Scrbe: Guy Grebla Part Sngle Buffer anagement In the prevous lecture we taled about the Combned Input

More information

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals Smultaneous Optmzaton of Berth Allocaton, Quay Crane Assgnment and Quay Crane Schedulng Problems n Contaner Termnals Necat Aras, Yavuz Türkoğulları, Z. Caner Taşkın, Kuban Altınel Abstract In ths work,

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 31 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 6. Rdge regresson The OLSE s the best lnear unbased

More information

Zeros and Zero Dynamics for Linear, Time-delay System

Zeros and Zero Dynamics for Linear, Time-delay System UNIVERSITA POLITECNICA DELLE MARCHE - FACOLTA DI INGEGNERIA Dpartmento d Ingegnerua Informatca, Gestonale e dell Automazone LabMACS Laboratory of Modelng, Analyss and Control of Dynamcal System Zeros and

More information

THERE ARE INFINITELY MANY FIBONACCI COMPOSITES WITH PRIME SUBSCRIPTS

THERE ARE INFINITELY MANY FIBONACCI COMPOSITES WITH PRIME SUBSCRIPTS Research and Communcatons n Mathematcs and Mathematcal Scences Vol 10, Issue 2, 2018, Pages 123-140 ISSN 2319-6939 Publshed Onlne on November 19, 2018 2018 Jyot Academc Press http://jyotacademcpressorg

More information

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA 4 Analyss of Varance (ANOVA) 5 ANOVA 51 Introducton ANOVA ANOVA s a way to estmate and test the means of multple populatons We wll start wth one-way ANOVA If the populatons ncluded n the study are selected

More information

Power law and dimension of the maximum value for belief distribution with the max Deng entropy

Power law and dimension of the maximum value for belief distribution with the max Deng entropy Power law and dmenson of the maxmum value for belef dstrbuton wth the max Deng entropy Bngy Kang a, a College of Informaton Engneerng, Northwest A&F Unversty, Yanglng, Shaanx, 712100, Chna. Abstract Deng

More information

Estimation: Part 2. Chapter GREG estimation

Estimation: Part 2. Chapter GREG estimation Chapter 9 Estmaton: Part 2 9. GREG estmaton In Chapter 8, we have seen that the regresson estmator s an effcent estmator when there s a lnear relatonshp between y and x. In ths chapter, we generalzed the

More information

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper Games of Threats Elon Kohlberg Abraham Neyman Workng Paper 18-023 Games of Threats Elon Kohlberg Harvard Busness School Abraham Neyman The Hebrew Unversty of Jerusalem Workng Paper 18-023 Copyrght 2017

More information

Limited circulation. For review only IEEE-TAC Submission no.: Periodic event-triggered control for nonlinear networked control systems

Limited circulation. For review only IEEE-TAC Submission no.: Periodic event-triggered control for nonlinear networked control systems Perodc event-trggered control for nonlnear networked control systems W. Wang, R. Postoyan, D. Nešć and W.P.M.H. Heemels Abstract Perodc event-trggered control PETC) s an appealng paradgm for the mplementaton

More information

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl RECURSIVE SPLINE INTERPOLATION METHOD FOR REAL TIME ENGINE CONTROL APPLICATIONS A. Stotsky Volvo Car Corporaton Engne Desgn and Development Dept. 97542, HA1N, SE- 405 31 Gothenburg Sweden. Emal: astotsky@volvocars.com

More information

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION Advanced Mathematcal Models & Applcatons Vol.3, No.3, 2018, pp.215-222 ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EUATION

More information

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS IJRRAS 8 (3 September 011 www.arpapress.com/volumes/vol8issue3/ijrras_8_3_08.pdf NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS H.O. Bakodah Dept. of Mathematc

More information

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling Real-Tme Systems Multprocessor schedulng Specfcaton Implementaton Verfcaton Multprocessor schedulng -- -- Global schedulng How are tasks assgned to processors? Statc assgnment The processor(s) used for

More information