COMPARED with AC microgrids, Direct-Current microgrids

Size: px
Start display at page:

Download "COMPARED with AC microgrids, Direct-Current microgrids"

Transcription

1 1 Optmal Power Flow n Stand-alone DC Mcrogrds Ja L, Feng Lu, Member, IEEE, Zhaojan Wang, Steen H Low, Fellow, IEEE, Shengwe Me, Fellow, IEEE, arx: [math.oc] 17 Aug 2017 Abstract Drect-current mcrogrds (DC-MGs) can operate n ether grd-connected or stand-alone mode. In partcular, stand-alone DC-MG has many dstnct applcatons. Howeer, the optmal power flow problem of a stand-alone DC-MG s nherently non-conex. In ths paper, the optmal power flow (OPF) problem of DC-MG s nestgated consderng conex relaxaton based on second-order cone programmng (SOCP). Mld assumptons are proposed to guarantee the exactness of relaxaton, whch only requre unform nodal oltage upper bounds and poste network loss. Furthermore, t s reealed that the exactness of SOCP relaxaton of DC-MGs does not rely on ether topology or operatng mode of DC-MGs, and an optmal soluton must be unque f t exsts. If lne constrants are consdered, the exactness of SOCP relaxaton may not hold. In ths regard, two heurstc methods are proposed to ge approxmate solutons. Smulatons are conducted to confrm the theoretc results. Index Terms DC mcrogrd, optmal power flow, conex relaxaton. I. INTRODUCTION COMPARED wth AC mcrogrds, Drect-Current mcrogrds (DC-MGs) hae been recognzed as an attracte alternate for numerous applcatons due to hgher effcency, more natural nterface to many types of renewable energy resources and energorage systems, better complance wth consumer electroncs, etc. [1]. Addtonally, when components are coupled around a DC bus, there are no ssues wth reacte power flow and frequencablty, resultng n a notably less complex power system [2]. DC-MGs may operate n ether grd-connected or stand-alone mode. The latter has many dstnct applcatons n shpboard [3], [4], arcraft [5], [6], automote [7], as well as the electrcty supply of remote rural areas. In ths regard, t s of great mportance to nestgate the optmal power flow of DC-MGs. Second-order cone programmng (SOCP) has been extensely used n AC networks for solng the optmal power flow (OPF) problem [8] [13]. In [8], radal dstrbuton load flow s formulated as a conc quadratc optmzaton problem, and soled effcently usng nteror-pont methods. In [10], a two-step SOCP relaxaton approach s proposed, whch consst of angle relaxaton and conc relaxaton. The exactness of angle relaxaton requres the so-called cyclc condton that the sum of angle dfferences on each cycle must be zero. Wth the cyclc condton satsfed, the conc relaxaton s exact, proded there are no upper bounds on loads. The formulatons of OPF problem and ther relaxatons are summarzed n J. L, F. Lu, Z. Wang and S. Me are wth the State Key Laboratory of Power System, the Department of Electrcal Engneerng, Tsnghua Unersty, Bejng , Chna.lfeng@tsnghua.edu.cn} S. H. Low s wth the Department of Electrcal Engneerng, Calforna Insttute of Technology, Pasadena, CA, USA, (emal:slow@caltech.edu) [11], and the suffcent condtons under whch the conex relaxatons are exact are presented n [12]. Another suffcent condton for the exact SOCP relaxaton of AC OPF n radal dstrbuton networks s proposed n [13], whch requres that the allowed reerse power flow s only reacte or acte, or none. As the OPF problem of DC power network s nherently nonlnear and non-conex, SOCP s extended to realze the conexfcaton of OPF problem n DC dstrbuted networks [14]. Such a result, howeer, reles on the assumpton that there exsts a large substaton wth unlmted capablty to prode power njecton and keep ts nodal oltage constant. Unfortunately, a stand-alone DC-MG s not the case because: 1) there s no substaton wth an unconstraned power njecton, and eery dstrbuted generaton theren has a lmted capacty; 2) there s no substaton wth a fxed nodal oltage, and all nodal oltages are allowed to ary wthn a certan range. Such dfferences motate us to extend the preous work [14] to ge rse to a better understandng of DC-MG n dfferent operatng modes. To ths end, the followng steps are taken. Step 1: Equalent Transformaton. By ntroducng slack arables, the orgnal problem OPF1 s transformed equalently nto OPF2 wth non-conex rank constrants. Step 2: Seocnd-Order Conc Relaxaton. By remong the rank constrants, the non-conex problem OPF2 s relaxed nto a conex SOCP problem,.e., RL1. Moreoer, extendng the results of [14], ths paper proes the exactness of SOCP relaxaton under mld assumptons, whch only requre unform nodal oltage upper bounds and poste network loss. Step 3: Equalent Conerson. By ntroducng alternate arables, RL1 s transformed equalently nto a branch flow model, namely RLS1, for the sake of mprong numercal stablty. The soluton approach and the relaton between the assocated theoretc results are llustrated n Fg. 1. OPF1 (Non-conex) Step 1 OPF2 (Non-conex) Step 2 Step 3 RL1 (Conex) RLS1 (Conex) Fg. 1. Proposed soluton approach. Theorem 1: Equalent Transformaton Theorem 2: Exact Conc Relaxaton Theorem 4: Unque Optmal Soluton Theorem 5: Equalent Conerson Theorem 3: Topologcal Independence

2 2 Under the proposed mld assumptons, the exactness of SOCP relaxaton and the unqueness of optmal soluton are reealed, whch are ndependent of topologes and operatng modes of DC-MGs. These propertes of DC-MG are ery helpful for optmal control and market desgn. When lne constrants are consdered, the stuaton turns to be much more complcated, snce the rank constrant cannot be guaranteed. In ths context, suffcent condtons are explored to justfy the exactness of SOCP relaxaton and two heurstc methods are suggested to construct feasble solutons when the rank constrant s not satsfed. The rest of ths paper s organzed as follows. The OPF problem of a stand-alone DC-MG s formulated and equalently transformed n Secton II. The SOCP relaxaton s gen n Secton III. Numercal studes are proded n Secton IV. The nfluence of lne constrants are dscussed n Secton V. Secton VI draws the conclusons. A. Basc Formulaton II. OPF PROBLEM OF DC-MGS Consder a graph G := (N, E), where N := 1,, n} denotes the set of all buses and E denotes the set of all lnes n the network of a DC-MG. G s assumed to be connected. Index the buses by 1,, n and abbreate, j} E as j. Denote ( j & < j) by j. For each bus N, denote V as ts oltage, and p as ts power njecton. For each lne j, y j denotes ts conductance. A letter wthout subscrpts denotes a ector of the correspondng quanttes, e.g., V = [V ] N. The notatons are summarzed n Fg.2. In p Bus V Fg. 2. Summary of notatons. y j Bus j a DC-MG, V, p, y j are real numbers, and y j > 0. Then the OPF problem of a stand-alone DC-MG reads V j... OPF1: mn h(p) f (p ) N oer: p, V ; s.t. p V (V V j ) y j, N ; p p p, N ; V V V, N. p j (1a) (1b) (1c) Here, f (p ) s strctly ncreasng n p. Eq. (1a) s the power njecton equaton for bus. The nodal oltages are constraned by (1c) wth the lower bound V > 0 and upper bound V. The power njectons are constraned by (1b) wth the lower bound p and upper bound p. It s assumed that p 0, whch ndcates the followng two cases: 1) Bus s a pure generaton bus wthout load. In ths case, p = 0 as the generators n a DC-MG can be turned off; 2) Bus s a pure load bus wthout generaton or a mxture power njecton of both load and generaton. In ths case, there s p < 0. In fact, the model of a grd-connected DC-MG [14] can be ewed as a specal case of OPF1 ncludng a substaton bus wth an unconstraned power njecton (p 0 =, p 0 = ) and a fxed nodal oltage (V 0 = V0 ref). Note that, n a DC-MG, lne constrants usually do not bnd n the normal operaton condton, snce the network s oerprosoned. Therefore, n ths paper, we frst consder the OPF problem of DC-MGs wthout lne constrants. Then, we dscuss the nfluence of lne constrants on the exactness of SOCP relaxaton. Throughout the paper, we do not assume any specfc topology of the power network. B. Equalent Transformaton The proposed OPF problem (1) s a non-lnear non-conex problem. By ntroducng slack arables, t can be transformed nto an equalent counterpart, where the non-conex power njecton equaton (1a) s conerted nto a rank constrant. Introduce slack arables to formulate a map f such that = V 2, N ; (2a) f := W j = V V j, j; (2b) and defne a matrx [ ] W R j := j (3) W j j for eery j. Then OPF1 (1) s transformed nto OPF2: mn h(p) oer: p,, W s.t. p p p p, N ( W j ) y j, N (4a) V 2 V 2, N W j 0, j W j = W j, j R j 0, j rank(r j ) = 1, j (4b) (4c) (4d) (4e) (4f) (4g) where the non-conexty n (1a) (n OPF1) s conerted nto the non-conexty n the rank constrant (4g) (n OPF2). R j s poste semdefnte as shown n (4f). Theorem 1: OPF1 and OPF2 are equalent. To proe Theorem 1, we frst ge the followng lemma. Lemma 1: Gen > 0 for N and W j 0 for j, let W j = W j for j. If rank(r j ) = 1 for j, then there exsts a unque V satsfyng V > 0 for N and (2). Moreoer, V s determned by V = for N. The proof of Lemma 1 can be found n Appendx A. Lemma 1 mples that for each (, W ), there exsts a unque V that satsfes the map gen by (2). Dfferng from [14], we

3 3 do not hae a substaton bus wth a fxed oltage. Instead, we assume V > 0 for all N. Wth Lemma 1, Theorem 1 can be proen. The proof can be found n Appendx B. It s worth notng that Theorem 1 does not rely on the topology of network. III. EXACTNESS OF CONIC RELAXATION A. SOCP Relaxaton of OPF n DC-MGs By remong (4g), the orgnal non-conex problem OPF2 s transformed to an SOCP problem (named as RL1) as below. RL1: mn h(p) oer: p,, W s.t. (4a) (4f) The only dfference between RL1 and OPF2 s that RL1 has no constrant (4g). Therefore, RL1 s exact, proded that ts eery optmal soluton satsfes (4g). To ensure the exactness of conc relaxaton, addtonal assumptons are requred. B. Assumptons Throughout the paper, we make the followng assumptons. Assumpton 1: V 1 = V 2 = V n > 0. Assumpton 2: N p > 0. Assumpton 1 requres all the nodal oltages hae the same upper bounds, whch s reasonable n DC-MGs, snce the scale of system s usually small. Assumpton 2 s tral as t means the total network loss s poste. Such assumptons relax those n [14], whch requre negate power njecton lower bounds and an unconstraned power njecton, to admt the features of stand-alone DC-MGs. C. Exactness of Conc Relaxton Wth Assumptons 1 and 2 mentoned aboe, we hae the followng man theorem: Theorem 2: RL1 s exact f Assumptons 1 and 2 hold for OPF1 (equalently OPF2). Theorem 2 clams that RL1 s an exact SOCP relaxaton of OPF2 (equalently OPF1) under Assumptons 1 and 2. To proe ths theorem, we ntroduce the followng lemmas. Lemma 2: Assume Assumpton 1 holds. Let (p,, W ) be feasble for RL1 but olate the rank constrant (4g) on a certan lne (s t) E. If p s = p s, then s < V 2 s. Meanwhle, f p t = p t, then t < V 2 t. The proof of Lemma 2 can be found n Appendx C. Lemma 2 mples that, for each bus, the power njecton s lower bound and the nodal oltage s upper bound cannot bnd at the same tme. Therefore, for any feasble soluton (p,, W ) of RL1, f t olates the rank constrant (4g) for a certan lne (s t) E and the constrant p s p s (or p t p t ) s bndng, then s V 2 s (or t V 2 t ) cannot bnd. Lemma 3: Assume Assumptons 1 and 2 hold for OPF1 and let (p,, W ) be a feasble soluton to RL1. If 1) (p,, W ) olates the rank constrant (4g) on a certan lne (s t) E; 2) p s > p s, p t > p t ; then there exsts another feasble soluton (p,, W ) that 1) satsfes (4a) (4f); 2) satsfes h(p ) < h(p). The proof of Lemma 3 can be found n Appendx D. Lemma 3 says that f a feasble pont (p,, W ) olates the rank constrant (4g) for a certan lne s t, whle both p s and p t are not bndng, then we can always fnd another feasble pont (p,, W ) wth a better objecte alue. It mples that f the optmal soluton (p,, W ) to RL1 olates the rank constrant (4g) for a certan lne s t, then at least one of p s and p t must hae reached ts lower bound. Lemma 4: Assume Assumptons 1 and 2 hold for OPF1 and let (p,, W ) be a feasble soluton to RL1. If 1) (p,, W ) olates the rank constrant (4g) on a certan lne (s t) E; 2) ether (p s = p s & p t > p t ) or (p s > p s & p t = p t ); then there exsts another feasble soluton (p,, W ) that 1) satsfes (4a) (4f); 2) satsfes h(p ) < h(p). The proof of Lemma 4 can be found n Appendx E. Lemma 4 means that f a feasble pont (p,, W ) olates the rank constrant (4g) for a certan lne s t, whle ether p s or p t s bndng, then we can always fnd another feasble pont (p,, W ) wth a better objecte alue. Lemma 3 and 4 mply that f the optmal soluton (p,, W ) to RL1 olates the rank constrant (4g) for a certan lne s t, then t must satsfy p s = p s and p t = p t. Otherwse, we can always fnd a better soluton. Lemma 5: Assume Assumptons 1 and 2 hold for OPF1 and let (p,, W ) be a feasble soluton to RL1. If 1) (p,, W ) olates the rank constrant (4g) on a certan lne (s t) E; 2) p s = p s and p t = p t ; then there always exsts (p,, W ) that 1) satsfes (4a) (4f); 2) olates rank constrant (4g) for all the neghbourng lnes of s t,.e. j wth }, j}} s}, t}}. The proof of Lemma 5 can be found n Appendx F. Lemma 5 says that f a feasble pont (p,, W ) olates the rank constrant (4g) for a certan lne s t, whle the correspondng power njectons lower bounds are bndng, we can always fnd another feasble pont (p,, W ) whch olates the rank constrant (4g) for s t and all ts neghborng lnes. It should be noted that n Lemma 5, the constructon of feasble pont (p,, W ) does not change p. Snce the network G s connected, Lemma 5 mples that we can contnue such propagaton to obtan a feasble pont that olates the rank constrant (4g) for all the lnes, wthout changng p. As mentoned before, f the optmal soluton (p,, W ) to RL1 olates the rank constrant (4g) for a certan lne s t, Lemma 3 and 4 ensure that t must satsfy p s = p s and p t = p t. In ths case, accordng to Lemma 5, we can always fnd a feasble (p,, W ) whch satsfes p = p and olates the rank constrant (4g) for all the lnes. Followng ths dea, we can proe Theorem 2 based on Lemma 2 5. The proof of Theorem 2 can be found n Appendx G.

4 4 D. Topologcal Independence Recall the SOCP relaxaton n AC networks [10]. The approach conssts of two relaxaton steps: angle relaxaton and conc relaxaton. Smlarly, our method also contans two steps: equalent transformaton and conc relaxaton. Dfferng from AC networks, DC networks do not nole oltage angles. The frst step n our method (.e., equalent transformaton) s exact, as Theorem 1 states. Howeer, n the second step, drectly remong the rank constrants may result n nexactness. Thus, addtonal assumptons (Assumptons 1, 2) are made to ensure the exactness of conc relaxaton. By notng that none of the two steps depends on specfc network topologes, we drectly hae the followng theorem: Theorem 3: Assume Assumptons 1 and 2 hold. Then the exactness of RL1 s ndependent of network topologes. Remark 1: In terms of a grd-connected DC-MG, t has also been demonstrated that the SOCP relaxaton s topologyndependent [14]. Actually, when a DC-MG works n grdconnected state, one can smply assgn a substaton bus n RL1 by lettng p 0 =, p 0 = + and 0 = [V ref 0 ]2. Hence, combnng the results n [14] and Theorem 3 mmedately concludes that the SOCP relaxaton of the OPF problem of DC-MGs does not rely on the topology of network, whether the DC-MG works n grd-connected or stand-alone mode. E. Unqueness of the Optmal Soluton Next we show the the optmal soluton to RL1 s unque. Theorem 4: Assume Assumptons 1 and 2 hold for RL1. Then the optmal soluton to RL1 s unque. The proof of Theorem 4 can be found n Appendx H. In a grd-connected DC-MG, t has also been proen that the SOCP problem has at most one optmal soluton [14]. Thus, the unqueness of optmal soluton to the SOCP problem does not depend on the operatng mode of DC-MG. F. Branch Flow Model In an optmal soluton to RL1, and W j may be numercally close to each other, snce the range of nodal oltage s small (usually p.u.) and Rank(R j ) = 1 s satsfed, whch mples that j = W j W j. Thus, RL1 s ll-condtoned snce equaton (4a) requres the subtractons of and W j. Howeer, such subtractons can be aoded by conertng RL1 nto a branch flow model, so that the numercal stablty s mproed. In lght of [14], by defnng z j := 1/y j and adoptng alternate arables P, l, RL1 can be conerted nto a branch flow model a the map g : (, W ) (, P, l) defned as below. =, N ;(5a) g := P j = ( W j ) y j, j; (5b) l j = yj 2 ( W j W j + j ), j. (5c) Wth g, RL1 s conerted nto the followng optmzaton problem wth branch flow model (BFM): RLS1: mn h(p) oer: p, P,, l; s.t. p P j, N ; p p p, N ; V 2 V 2, N ; P j + P j = z j l j, j; j = z j (P j P j ), j; (6a) (6b) (6c) (6d) (6e) l j P j 2, j (6f) where P j denote the power flow through lne j, and l j denote the magntude square of the current through lne j. Theorem 5: RL1 and RLS1 are equalent. The proof of Theorem 5 can be found n Appendx I. Snce RL1 s a conex relaxaton of the non-conex OPF2 by remong the rank constrant (4g), Theorem 5 mples that RLS1 s the conex relaxaton of another non-conex problem, namely, OPF3, whch s obtaned by conertng OPF2 usng the one-to-one map g (5). OPF3: mn h(p) oer: p, P,, l s.t. (6a) (6e) l j = P 2 j, j Under Assumptons 1 and 2, let F denote a feasble set of a certan optmzaton problem, whch s ndcated by the subscrpt. Then the relatonshp between dfferent feasble sets s shown as below: F OP F 2 = f(f OP F 1 ) F RL1 F RLS1 = g(f RL1 ) where f s the one-to-one map gen n Theorem 1, and g s the one-to-one map gen n Theorem 5. The relatonshp between dfferent feasble sets s depcted n Fg. 3. F OP F 1 s a non-conex regon and (p, V ) denotes the optmal soluton to OPF1. F OP F 1 s transformed equalently nto another non-conex regon F OP F 2 by the map f. F OP F 2 s conexfed by remong the rank constrant (4g), yeldng F RL1, whch s larger than F OP F 2. The exactness of RL1 ensures that the optmal soluton to RL1,.e., (p,, W ) s not n the shaded regon of F RL1. Snce the objecte functon of OPF2 and RL1 are the same, (p,, W ) s also the optmal soluton to OPF2. Applyng the one-to-one map f, (p,, W ) s transformed nto (p, V ). The one-to-one map g conerts F RL1 nto an equalent conex regon F RLS1, and conerts F OP F 2 nto an equalent non-conex regon F OP F 3. (p, P,, l ) s the optmal soluton to RLS1, whch s also the optmal soluton to OPF3, snce RLS1 s exact. Moreoer, (p, P,, l ) can be conerted equalently nto (p,, W ) usng the map g. Extendng the work n [14], we hae shown that under the proposed assumptons, the OPF problem n DC-MGs can be conerted equalently nto a conex SOCP problem, regardless of topologes or operatng modes. Thereby solng the relaxed conex SOCP can obtan the global optmum wth theoretc guarantee. Furthermore, we hae shown that the SOCP problem has at most one optmal soluton, whether the

5 5 F OPF1 (p*,v*) f F OPF2 (p*,*,w*) g F OPF3 (p*,p*,*,l*) The oltage lower and upper bounds are set as 0.95 p.u. and 1.05 p.u., respectely. The objecte s to mnmze total network loss and the problems are soled usng MOSEK. The results are lsted n Table I. F RL1 Fg. 3. Relatonshp between dfferent feasble sets. DC-MG s workng n grd-connected or stand-alone mode. Remark 2: In OPF1 OPF3, lne flow constrants are not consdered. Whereas we hae not found any olaton of lne constrants n all our tests, we cannot theoretcally exclude the possblty of such olaton. In such a crcumstance, approxmate solutons can be heurstcally constructed. We wll dscuss the stuaton ths ssue later on n Secton V. A. 16-bus System IV. CASE STUDIES Tests are conducted on the modfed 16-bus system [15] to demonstrate the effcacy of the proposed method when the system works n dfferent operatng modes and network topologes. The three-feeder system s shown n Fg. 4 where 6 dstrbuted generators (DGs) are added wth the same capacty of 5MW. The dashed lnes represent te lnes wth te breakers. The system s able to work n two modes: F RLS1 Grd-connected: the system s connected to a power grd a all the three feeders (Feeders A, B and C). Stand-alone: all the three feeders are dsconnected from the grd. Addtonally, the network can swtch between tree and mesh topologes by openng or closng the te breakers. Tree topology: all the te breakers are swtched off. Mesh topology: all the te breakers are swtched on. Thus, the followng four cases are studed. 1) GT: grd-connected mode wth a tree topology. 2) GM: grd-connected mode wth a mesh topology. 3) ST: stand-alone mode wth a tree topology. 4) SM: stand-alone mode wth a mesh topology. G Feeder A Feeder B Feeder C G 5 G G 11 Fg bus system wth DGs G G TABLE I RESULTS OF 16-BUS SYSTEM GT GM ST SM Exactness 5.73E E E E-10 Objecte Value (p.u.) Computaton Tme (s) At a numercal RLS1 soluton (p,, W ), R j can be obtaned for each j. If RLS1 s exact, then for all j, we should hae rank(r j ) = 1,.e., the dfference D j := j W j W j = 0. Hence the smaller D j ndcates, the closer R j s to rank one. The row Exactness lsts the maxmum alue of D j for all j, ndcatng RLS1 s exact for all the test networks. Moreoer, t ndcates that the proposed SOCP relaxaton does not depend on the operatng mode or network topology. The objecte alues ndcate that n ths system, mesh topology reduces network loss n both grd-connected and stand-alone modes, snce mesh topology s able to support hgher nodal oltages. Addtonally, workng n the same network topology, grd-connected mode experences less network loss than stand-alone mode, snce more power s njected from the feeders to support hgher nodal oltages. The results of power njectons and nodal oltages are lsted n Table II. TABLE II RESULTS OF POWER INJECTIONS AND NODAL VOLTAGES Power Injecton (p.u.) Nodal Voltage (p.u.) Bus GT GM ST SM GT GM ST SM B. Exactness and Comparson More tests are conducted on both mesh and tree networks to check the exactness of the SOCP relaxaton. In addton, the objecte alue and computaton tme of the relaxed model (.e., RLS1) and the orgnal model (.e., OPF1) are compared to show the effcacy of the relaxaton. The objecte of both models s to mnmze total network loss. The mesh networks [14] are modfed from MATPOWER by gnorng lne reactances. All the lne resstances are reduced to 10% of the orgnal alues to smulate the DC-MG condton. Partcularly, the zero resstances are reset as 10 3 p.u.. Addtonally, the IEEE 118-bus system s appled to show the scalablty of

6 6 the proposed method. Four radal dstrbuton networks n the lterature are also used to erfy the topologcal ndependence of the relaxaton. In these systems, all the lne resstances are also reduced to 10% of the orgnal alues to smulate the DC- MG condton. In all the test systems, the oltage lower and upper bounds are set as 0.95 p.u. and 1.05 p.u., respectely. RLS1 s soled usng MOSEK, whle OPF1 s soled usng IPOPT. The results are lsted n Table III. TABLE III EXACTNESS OF RLS1 AND COMPARISON OF TWO MODELS Exactness Objecte Value (p.u.) Tme (s) Topology System of RLS1 RLS1 OPF1 RLS1 OPF1 case6ww 1.24E E E case9 7.17E E E Mesh case eee E E E case E E E case E E E bus [16] a 1.28E E E Tree 70-bus [17] b 5.35E E E bus [18] c 3.09E E E a 6 DGs are added at bus 5, 10, 15, 20, 25, 30, wth the capacty of 50kW. b 13 DGs are added at bus 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, wth the capacty of 50kW. c 13 DGs are added at bus 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, wth the capacty of 50kW. It s obsered that RLS1 s exact for both mesh and tree topologes. The scale of a DC-MG s small, howeer, more and more DC-MGs may be bult n the future to ntegrate dstrbuted renewable energy generaton and electrc ehcles. The results of computaton tme mples that the proposed method may be appled n large scale systems, for example, the cluster of DC-MGs. We also tested the model wth lne constrants (.e., RLS2), on the aboe systems. Howeer, we hae not found any case where the optmal soluton olates the rank constrants (4g). In ths regard, we conjecture that the RLS2 s almost always exact n practce. Snce we cannot theoretcally exclude the possblty of nexactness of OPF n ths case, we ge some dscussons and nsghts n the next secton. V. DISCUSSION ON LINE CONSTRAINTS A. OPF wth Lne Constrants Let I j denote the threshold of current through lne j, then the lne constrant can be formulated as whch s correspondng to y 2 j( W j W j + j ) I 2 j (7) y 2 j(v V j ) 2 I 2 j (8) n the orgnal problem. By addng (7) nto RL1, we hae the followng model. RL2: mn h(p) oer: p,, W s.t. (4a) (4f), (7) Usng the transformaton n Theorem 5, the lne constrant (7) can be be added nto RLS1, yeldng the followng model. RLS2: mn h(p) oer: p, P,, l s.t. (6a) (6f) l j I 2 j (9) B. Exactness Condtons When Consderng Lne Constrants We frst ntroduce the followng theorem to ge the condtons that guarantee the exactness of RL2. Theorem 6: Assume Assumptons 1 and 2 hold for RL2. Then RL2 s exact under ether of the followng condtons. Condton 1: ( j) E, constrant (7) s not bndng ; Condton 2: For each lne (s t) E, f constrant (7) s bndng for ths lne, then the correspondng lower bounds on p s and p t are not bndng. The proof of Theorem 6 can be found n Appendx J. Furthermore, f the soluton to RL2 s not exact, t s possble to fnd an approxmate soluton by relaxng the bounds on the power njectons, as the followng theorem ndcates. Theorem 7: Assume Assumptons 1 and 2 hold for RL2 and let (p,, W ) be a feasble soluton to RL2. If 1) (p,, W ) olates the rank constrant (4g) for a certan lne (s t) E where the lne constrant (7) s bndng; 2) at least one of p s = p s and p t = p t s satsfed; then there must exst another soluton (p,, W ) such that 1) satsfes all the constrants of RL2 except for (4b) (.e., (4a), (4c) (4f) and (7)); 2) satsfes the rank constrant (4g). 3) has a lower objecte alue than (p,, W ). The proof of Theorem 7 can be found n Appendx K. Theorem 7 ndcates that f the adjustment of power njecton ɛ s allowed at bus s and bus t, then a soluton whch satsfes the rank constrant (4g) can be constructed. In a DC- MG, such adjustment may be acheed by employng demand response or energorage. C. Constructng Approxmate Optmal Solutons Snce RL2 s not always exact, we can check the soluton after solng RL2. If the soluton satsfes the rank constrant (4g), then t s global optmal for the orgnal problem OPF1 (1). Otherwse, nspred by the recoery methods of semdefnte program (SDP) relaxaton [19], we propose two heurstc methods to construct a nearly optmal soluton. Theorem 6 ndcates that f (p,, W ) s the optmal soluton to RL2, and olates the rank constrant (4g) for a certan lne (s t) E where the lne constrant (7) s bndng, then at least one of p s and p t must reach ts lower bound. Otherwse, we can always fnd another feasble pont, whch has a smaller objecte alue than (p,, W ). Therefore, we only need to dscuss the cases that at least one of p s and p t reaches ts lower bound. 1) Drect constructon method: In [19], a drect recoery method s proposed for AC networks, usng the frst column of the optmal soluton matrx W to recoer a nearly optmal soluton. It nspres a drect constructon method for DC-MGs. Assume (p,, W ) s the optmal soluton to RL2, whch olates the rank constrant (4g) for some lnes. Instead of usng the frst column of W (see n [19]), we use to construct an approxmate soluton (p, V ) to the orgnal problem

7 7 OPF1. Frst, V ( N ) s dered by lettng V =. Then, p ( N ) s dered by substtutng V nto the power balance equaton (1a). Next we show the approxmate soluton (p, V ): 1) satsfes power balance equaton (1a), the oltage constrant (1c), lne constrant (8); 2) may olate (1b) for some N, but the olatons hae lmted bounds. Accordng to the constructon of (p, V ), t s straghtforward to check that (p, V ) satsfes (1a) and (1c). Hence we only need to examne constrant (8) to justfy the frst asserton. For any N, let C denote the set of buses mmedately connected to bus. Let B C denote the subset such that for any j B, lne j olates the rank constrant (4g). If all the neghborng lnes of bus do not olate the rank constrant (4g), we hae B =. It follows that for all ( j) E, j > W j for j B, whle j = W j for j / B. Snce (p,, W ) satsfes (7), for any j(j B ), t follows that yj(v 2 V j ) 2 = yj(v 2 2 2V V j + Vj 2 ) = yj( 2 2 j + j ) and for any j(j / B ) < y 2 j( 2W j + j ) I 2 j yj(v 2 V j ) 2 = yj(v 2 2 2V V j + Vj 2 ) = yj( 2 2 j + j ) = y 2 j( 2W j + j ) I 2 j. Therefore, (p, V ) satsfes (8). Next, we check constrant (1b). It follows from (1a) that p V (V V j ) y j ( ) j y j,j / B +,j B,j / B = p. ( ) j y j ( ) j y j ( ) j y j ( W j) yj +,j B ( Wj ) yj Hence, p may olate (1b). And the olaton s bounded by p p ( ) j + W j y j.,j B 2) Slack arable method: Accordng to Theorem 6, RL2 s exact f the lower bounds of power njectons are not bndng. Thus, after solng RL2, f the soluton olates the rank constrant (4g) for a certan lne (s t) E, and any lower bound of p s and p t s bndng, for example, p s = p s, then a correspondng slack arable ε s (ε s > 0) can be added nto (4b) to reformulate the constrant as p s p s ε s p s (10) so that p s wll not reach ts lower bound n the next teraton to sole RL2. Addtonally, n order to mnmze ε s, t s also added nto the objecte functon as mn h(p) + ˆN where ˆN s the set of buses where the rank constrant (4g) s olated and the power njecton lower bound s bndng at the same tme. The procedure s shown n Fg. 5. No Sole RL2 Rank(R j )=1 for all the lnes? No e has been added for all the buses? Yes Add e for the lne where Rank(R j ) 1 Add e n the objecte functon Sole RL2 Rank(R j )=1 for all the lnes? No Yes Feasble Soluton Fg. 5. Flowchart of slack arable method. Yes ε VI. CONCLUSIONS Optmal Soluton Feasble Soluton In ths paper, we hae proposed an SOCP relaxaton of the OPF problem n stand-alone DC-MGs, whch do not consst of any substaton wth an unconstraned power njecton and a fxed nodal oltage. We hae also proposed two mld assumptons whch only requre unform oltage upper bounds and poste network losses. Under such assumptons, we hae proen the exactness of the SOCP relaxaton by extendng the results n [14]. Combnng the results n [14] and those n ths paper, we hae a more comprehense understandng on the OPF problem n DC-MGs: 1) Under the proposed assumptons, the exactness of SOCP relaxaton n DC-MGs does not rely on the operatng mode of grd. Ths property facltates the optmal operaton of DC-MG, snce t allows the operator to achee optmal operaton whether the DC-MG s workng n grdconnected or stand-alone mode. 2) Unlke AC systems, the exactness of SOCP relaxaton n DC-MGs does not rely on the topology of networks. It mples that when the topology s changed due to

8 8 lne swtchng, the OPF problem can stll be conerted equalently nto a conex SOCP problem. 3) The unqueness of the global optmal soluton to the SOCP problem s ndependent of the topology and operatng mode of DC-MGs. Ths property s especally useful for settng a target n optmal control of DC-MGs. REFERENCES [1] T. Dragčeć, J. C. Vasquez, J. M. Guerrero, and D. Škrlec, Adanced LVDC electrcal power archtectures and mcrogrds, IEEE Electrf. Mag., ol. 2, no. 1, pp , [2] T. Dragčeć, X. Lu, J. C. Vasquez, and J. M. Guerrero, DC Mcrogrds - Part I: A Reew of Control Strateges and Stablzaton Technques, IEEE Trans. Power Electron., ol. 31, no. 7, pp , [3] G. F. Reed, B. M. Granger, A. R. Sparacno, and Z.-H. Mao, Shp to grd: Medum-oltage dc concepts n theory and practce, IEEE Power Energy Mag., ol. 10, no. 6, pp , [4] Z. Jn, G. Sullgo, R. Cuzner, L. Meng, J. C. Vasquez, and J. M. Guerrero, Next-generaton shpboard dc power system: Introducton smart grd and dc mcrogrd technologes nto martme electrcal netowrks, IEEE Electrf. Mag., ol. 4, no. 2, pp , [5] P. Magne, B. Nahd-Mobarakeh, and S. Perfederc, General acte global stablzaton of multloads DC-power networks, IEEE Trans. Power Electron., ol. 27, no. 4, pp , [6], Acte stablzaton of dc mcrogrds wthout remote sensors for more electrc arcraft, IEEE Trans. Ind. Appl., ol. 49, no. 5, pp , [7] A. Emad, A. Khalgh, C. H. Retta, and G. A. Wllamson, Constant power loads and negate mpedance nstablty n automote systems: Defnton, modelng, stablty, and control of power electronc conerters and motor dres, IEEE Trans. Veh. Technol., ol. 55, no. 4, pp , [8] R. A. Jabr, Radal dstrbuton load flow usng conc programmng, IEEE Trans. Power Syst., ol. 21, no. 3, pp , [9] S. Sojoud and J. Laae, Physcs of power networks makes hard optmzaton problems easy to sole, IEEE Power Energy Soc. Gen. Meet., pp. 1 8, [10] M. Farar and S. H. Low, Branch flow model: Relaxatons and conexfcaton Part I, IEEE Trans. Power Syst., ol. 28, no. 3, pp , [11] S. H. Low, Conex relaxaton of optmal power flow Part I: Formulaton and equalence, IEEE Trans. Control Netw. Syst., ol. 1, no. 1, pp , [12], Conex relaxaton of optmal power flow Part II: Exactness, IEEE Trans. Control Netw. Syst., ol. 1, no. 2, pp , [13] S. Huang, Q. Wu, J. Wang, and H. Zhao, A suffcent condton on conex relaxaton of AC optmal power flow n dstrbuton networks, IEEE Trans. Power Syst., ol. 32, no. 2, pp , [14] L. Gan and S. H. Low, Optmal power flow n drect current networks, IEEE Trans. Power Syst., ol. 29, no. 6, pp , [15] S. Canlar, J. J. Granger, H. Yn, and S. S. H. Lee, Dstrbuton feeder reconfguraton for loss reducton, IEEE Trans. Power Del., ol. 3, no. 3, pp , [16] M. E. Baran and F. F. Wu, Network reconfguraton n dstrbuton systems for loss reducton and load balancng, IEEE Trans. Power Del., ol. 4, no. 2, pp , [17] D. Das, A fuzzy multobjecte approach for network reconfguraton of dstrbuton systems, IEEE Trans. Power Del., ol. 21, no. 1, pp , [18] C.-T. Su and C.-S. Lee, Network reconfguraton of dstrbuton systems usng mproed mxed-nteger hybrd dfferental eoluton, IEEE Trans. Power Del., ol. 18, no. 3, pp , [19] G. Fazelna, R. Madan, A. Kalbat, and J. Laae, Conex relaxaton for optmal dstrbuted control problems, IEEE Trans. Automat. Contr., ol. 62, no. 1, pp , 2017.

9 9 A. Proof of Lemma 1 APPENDIX Proof: Exstence: Let V = for N. It suffces to show that V satsfes V > 0 for N and (2). Snce > 0 for N, t follows that V = > 0 for N. It s straghtforward to check that V satsfes (2a). Snce R j s not full rank, we hae j W j W j = 0, j Snce W j 0, we hae W j = Wj 2 = W j W j = j = V V j for j,.e., V, satsfes (2b). Ths completes the proof of exstence. Unqueness: Let Ṽ denote an arbtrary soluton to Ṽ > 0 for N and (2). It suffces to show that Ṽ = for N. Assume Ṽ for some N, then t follows from (2a) that Ṽ = < 0, whch contradcts wth Ṽ > 0 for N. Thus, Ṽ = for N. Ths completes the proof of unqueness. B. Proof of Theorem 1 Proof: Let F OP F 1 and F OP F 2 denote the feasble sets of OPF1 and OPF2, respectely. Snce OPF1 and OPF2 hae the same objecte functon, t suffces to show that there exsts a one-to-one map between F OP F 1 and F OP F 2. Specfcally, we show the map f : (V ) (, W ) gen by (2) s one-to-one, snce p s unquely determned by V or (, W ). To ths end, t suffces to show the map s both nto and onto. On the one hand, t s straghtforward that for any V 1 V 2 F OP F 1, f(v 1 ) f(v 2 ) F OP F 2. On the other hand, as Lemma 1 says, for each (, W ) F OP F 2, there exsts a unque f 1 (, W ) = V F OP F 1. Ths completes the proof. C. Proof of Lemma 2 Proof: For brety, we only proe the case of bus s as the proof of bus t s the same. Wth regard to the alue of p s, we only need to dscuss two cases: p s = p s < 0 and p s = p s = 0. In the frst case, we hae p s ( s W s ) y s < 0 : s accordng to (4a). Therefore, ( s W s ) y s < 0 for some N. It follows from (4d) (4f) that W s s. Accordng to Assumpton 1, V s = V. Thus, we hae s < W s 2 2 s V s V = V 2 s. In the second case, we hae p s ( s W s ) y s = 0 (11) : s accordng to (4a). Then we dscuss the followng two cases: 1) When s W s = 0 for all s, we hae s = W st for = t. Snce (p,, W ) olates the rank constrant (4g) for s t, R st s non-sngular. Hence, W st s t W st < s t due to (4d) (4f). Accordng to Assumpton 1, V s = V t. It follows that s = W st < 2 2 s t V s V t = V 2 s. 2) When s W s = 0 does not hold for some of s, there must exst at least one s such that s W s < 0 (and accordngly there exsts at least another s such that s W s > 0 n order to satsfy (11) ). Hence we hae s < W s 2 2 s V s V = V 2 s due to V s = V. Ths completes the proof. D. Proof of Lemma 3 Proof: Snce (p,, W ) satsfes (4d) (4f), we hae 0 W j j for j. Snce (p,, W ) olates the rank constrant (4g) for s t, we hae W st s t. Thus, W st < s t. We can always choose a small enough number ɛ > 0 such that ɛ < mn ps p s, p t p t, } s t W st. Followng the lne of the proof of Lemma 7 n [14], we can use ɛ to construct W as W j W j + ɛ f, j} = s, t}; = otherwse; W j and construct p as p ( W j ) yj, N. The pont (p,, W ) satsfes (4a) due to the constructon of p. When s, t, we hae p ( W j) yj ( W j ) y j = p. When = s, t, we hae p ( W j)y j ( W j )y j ɛ = p ɛ (p, p ). Hence, (p,, W ) satsfes (4b) and p < p f = s, t; = p otherwse. It follows that h(p ) < h(p) snce f s strctly ncreasng n p for N. The pont (p,, W ) satsfes (4c) snce remans the same as the feasble pont (p,, W ). The pont (p,, W ) satsfes (4d) due to the constructon of W. The pont (p,, W ) satsfes (4e) snce W j W j = W j W j = 0 for j. Addtonally, the pont (p,, W ) satsfes (4f) snce when, j} s, t} and W j = W j [0, j ] W j = W j + ɛ [0, j ) when, j} = s, t}. Ths completes the proof.

10 10 E. Proof of Lemma 4 Proof: We present the proof for the case (p s = p s & p t > p t ). The proof for the case (p s > p s & p t = p t ) s smlar and omtted for brety. Smlar to the proof of Lemma 3, we hae W st < s t. Addtonally, t follows from Lemma 2 that s < V 2 s, so we can always choose a small enough number ɛ > 0 such that pt p ɛ < mn t, j:j s s t W st, y ( ) } sj V 2 s s, then p t ɛ > p t, W st + ɛ < s t, s + j:j s y sj ɛ < V 2 s. Followng the lne of the proof of Lemma 8 n [14], we can use ɛ to construct W as W j W j + ɛ f, j} = s, t}; = otherwse; construct as = W j + ɛ f = s; yj otherwse; and construct p as p ( W j)y j, N. The pont (p,, W ) satsfes (4a) accordng to the constructon of p. When s, t, we hae p ( W j) yj ( W j ) y j = p. When = s, we hae p s ( s W sj j:j s ) ysj ( s W sj ) y sj + ( s s ) y sj ɛ j:j s ( s W sj ) y sj = p s. j:j s When = t, we hae p t ( t W tj)y tj j:j t j:j s ( t W tj )y tj ɛ j:j t = p t ɛ (p t, p t ). Thus, (p,, W ) satsfes (4b) and p < p f = t; = p otherwse. It follows that h(p ) < h(p). Snce = f s and < < V 2 f = s, the pont (p,, W ) satsfes (4c). Smlar to the proof of Lemma 3, t s straghtforward to check that the pont (p,, W ) satsfes (4d), (4e) and (4f). Ths completes the proof. F. Proof of Lemma 5 Proof: Smlar to the proof of Lemma 3, we hae W st < s t. Accordng to Lemma 2, we hae s < V 2 s, t < V 2 t. Therefore, we can always choose a small enough number ɛ > 0 so that s j:j s ɛ < mn t W st, y ( ) sj V 2 s s, y st j:j t y ( ) } tj V 2 t t, then W st + ɛ < s t, s + j:j s y sj ɛ < V 2 s, t + j:j t y ɛ < V 2 t. tj Followng the lne of the proof of Lemma 9 n [14], we can use ɛ to construct W as W j = W j + ɛ W j f, j} = s, t}; otherwse; and construct as + ɛ = s, t; = yj otherwse. The pont (p,, W ) satsfes (4a) snce when = s, t, ( W j ) yj ( W j ) y j + ( W j ) y j = p. y j ɛy j ɛ When s, t, we hae ( W j) yj ( W j ) y j = p. (12) Snce p does not change, the pont (p,, W ) satsfes (4b). It follows from (12) that = f s, t and < < V 2 f = s, t. Hence, the pont (p,, W ) satsfes (4c). Smlar to the proof of Lemma 3, t s straghtforward to check that the pont (p,, W ) satsfes (4d), (4e) and (4f). Snce G s connected, there always exsts some j such that, j} s, t} and }, j}} s}, t}}. Then we hae W j = W j j < j. Partcularly, when, j} = s, t}, W j = W j + ɛ < j < j. It means (p,, W ) olates the rank constrant (4g) for s t and all ts neghborng lnes. Ths completes the proof. G. Proof of Theorem 2 Proof: To proe that RL1 s exact, t suffces to show that any optmal soluton to RL2 satsfes the rank constrant (4g). We frst assume RL1 s not exact for the sake of

11 11 contradcton. Then there must exst at least one optmal soluton (p,, W ) of RL, whch olates the rank constrant (4g) for a certan lne (s t) E. We dscuss from the followng three aspects of the power njectons p s and p t correspondng to s t. (1) p s > p s & p t > p t. In ths case, as Lemma 3 ponts out, there must exst a feasble (p,, W ) whch has a smaller objecte alue than (p,, W ). In ths stuaton, (p,, W ) cannot be optmal for RL. (2) (p s = p s & p t > p t ) or (p s > p s & p t = p t ). In ths case, Lemma 4 states that there must exst a feasble (p,, W ) that has a smaller objecte alue than (p,, W ). Hence, (p,, W ) cannot be optmal for RL. (3) p s = p s & p t = p t. In ths case, we show such an optmal soluton (p,, W ) does not exst. Accordng to Lemma 5, there always exsts a feasble (p,, W ) whch olates the rank constrant (4g) for s t and all ts neghborng lnes. Snce G s connected and p = p, we can further propagate such constructon to obtan another feasble (p,, W ) of RL, whch olates the rank constrant (4g) for all the lnes. Hence, p satsfes p = p for all N. It follows that p p 0 N N Ths contradcts Assumpton 2 whch states N p > 0. Thus, (p,, W ) does not exst. Thus, eery optmal soluton must satsfy the rank constrant (4g),.e., RL1 s exact. Ths completes the proof. H. Proof of Theorem 4 Proof: Let x 1 := (p 1, 1, W 1 ) and x 2 := (p 2, 2, W 2 ) be two optmal solutons of RL1, then we hae f (p 1 ) f (p 2 ) (13) N N Followng the lne of the proof of Theorem 3 n [14], we know 1 2 = 1 j j 2 W 1 j = = η, j; 1 1 j It follows from (4a) that p 1 = η ( 1 ( η j Wj 1 ) yj ηwj 2 ) yj = ηw 2 j. = ηp 2. Snce f (p ) s strctly ncreasng, we must hae η = 1. Otherwse, t contradcts (13). Thus, x 1 = x 2,.e., RL1 has at most one optmal soluton. Ths completes the proof. I. Proof of Theorem 5 Proof: Let F RL1 and F RLS1 denote the feasble sets of RL1 and RLS1, respectely. It suffces to show that the map g (5) s one-to-one between F RL1 and F RLS1, snce p s determned by (, W ) n F RL1 and determned by P n F RLS1. On the one hand, t s straghtforward that for any ( 1, W 1 ) ( 2, W 2 ) F RL1, g( 1, W 1 ) g( 2, W 2 ) F RLS1. On the other hand, for any (, P, l) F RLS1, t can be erfed that there exsts a g 1 (, P, l) = (, W ) F RL1. Ths completes the proof. J. Proof of Theorem 6 Proof: When the lne constrant (7) s not bndng, RL2 s equalent to RL1, whch s exact under Assumptons 1 and 2 accordng to Theorem 2. Howeer, when constrant (7) s bndng, Theorem 2 does not hold. In ths stuaton, assume (p,, W ) s the optmal soluton to RL2. Assume (p,, W ) olates the rank constrant (4g) for a certan lne (s t) E where constrant (7) s bndng, and the lower bounds of p s and p t are not bndng. Snce (p,, W ) satsfes (4d) (4f), we hae 0 Wj j for j. Snce (p,, W ) olates the rank constrant (4g) for s t, we hae Wst s t. Thus, Wst < s t. We can always choose a small enough number ɛ > 0 such that ɛ < mn p s p s, p t p t, y s t Wst st Then we can use ɛ to construct (p,, W ) where W j Wj := + ɛ f, j} = s, t}; Wj otherwse; p : ( W j ) yj, N. }. It s easy to erfy that (p,, W ) satsfes (4a) (4f) and h(p ) < h(p ). Addtonally, when, j} s, t}, we hae y 2 j( W j W j + j ) = y 2 j( W j W j + j ) I 2 j. When, j} = s, t}, snce the lne constrant (7) s bndng, y 2 j( W j W j+ j ) = y 2 j( W j W j+ j 2ɛ) < I 2 j. Thus, the pont (p,, W ) satsfes (7). It means that the pont (p,, W ) s feasble for RL2 and has a smaller objecte alue than (p,, W ). Thus, (p,, W ) cannot be the optmal soluton. In contrast, the optmal soluton to RL2 must satsfy the rank constrant (4g). Ths completes the proof. K. Proof of Theorem 7 Proof: Notng that (p,, W ) satsfes (4d) (4f), we hae 0 W j j for j. Snce (p,, W ) olates the rank constrant (4g) for s t, W st s t. Hence W st < s t. Then we can always choose ɛ := s t W st > 0 to construct another soluton (p,, W ) as below: W j W j + ɛ f, j} = s, t}; := W j otherwse; p : ( W j ) yj, N. It s easy to erfy that (p,, W ) satsfes (4a), (4c) (4f) and (7). Hence the frst asserton s proen.

12 12 In terms of the second asserton, (p,, W ) satsfes (4g) by notng and W j = W j = j for, j} s, t} W j = W j + ɛ = j for, j} = s, t}. Next we consder the thrd asserton. Accordng to the constructon of (p,, W ), for any s, t, p satsfes p ( W j) yj ( W j ) y j = p and for any when = s, t, p ( W j)y j ( W j )y j ɛ = p ɛ < p Then (p,, W ) has a lower objecte alue than (p,, W ) because the objecte functon s strctly ncreasng n p. Ths completes the proof.

Assortment Optimization under MNL

Assortment Optimization under MNL Assortment Optmzaton under MNL Haotan Song Aprl 30, 2017 1 Introducton The assortment optmzaton problem ams to fnd the revenue-maxmzng assortment of products to offer when the prces of products are fxed.

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

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

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

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

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

ELE B7 Power Systems Engineering. Power Flow- Introduction

ELE B7 Power Systems Engineering. Power Flow- Introduction ELE B7 Power Systems Engneerng Power Flow- Introducton Introducton to Load Flow Analyss The power flow s the backbone of the power system operaton, analyss and desgn. It s necessary for plannng, operaton,

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

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

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

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

Graph Reconstruction by Permutations

Graph Reconstruction by Permutations Graph Reconstructon by Permutatons Perre Ille and Wllam Kocay* Insttut de Mathémathques de Lumny CNRS UMR 6206 163 avenue de Lumny, Case 907 13288 Marselle Cedex 9, France e-mal: lle@ml.unv-mrs.fr Computer

More information

NP-Completeness : Proofs

NP-Completeness : Proofs NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem

More information

The Study of Teaching-learning-based Optimization Algorithm

The Study of Teaching-learning-based Optimization Algorithm Advanced Scence and Technology Letters Vol. (AST 06), pp.05- http://dx.do.org/0.57/astl.06. The Study of Teachng-learnng-based Optmzaton Algorthm u Sun, Yan fu, Lele Kong, Haolang Q,, Helongang Insttute

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

Lecture 20: Lift and Project, SDP Duality. Today we will study the Lift and Project method. Then we will prove the SDP duality theorem.

Lecture 20: Lift and Project, SDP Duality. Today we will study the Lift and Project method. Then we will prove the SDP duality theorem. prnceton u. sp 02 cos 598B: algorthms and complexty Lecture 20: Lft and Project, SDP Dualty Lecturer: Sanjeev Arora Scrbe:Yury Makarychev Today we wll study the Lft and Project method. Then we wll prove

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

Optimization Methods for Engineering Design. Logic-Based. John Hooker. Turkish Operational Research Society. Carnegie Mellon University

Optimization Methods for Engineering Design. Logic-Based. John Hooker. Turkish Operational Research Society. Carnegie Mellon University Logc-Based Optmzaton Methods for Engneerng Desgn John Hooker Carnege Mellon Unerst Turksh Operatonal Research Socet Ankara June 1999 Jont work wth: Srnas Bollapragada General Electrc R&D Omar Ghattas Cl

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

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

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

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

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

On the Multicriteria Integer Network Flow Problem

On the Multicriteria Integer Network Flow Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of

More information

Energy Storage Elements: Capacitors and Inductors

Energy Storage Elements: Capacitors and Inductors CHAPTER 6 Energy Storage Elements: Capactors and Inductors To ths pont n our study of electronc crcuts, tme has not been mportant. The analyss and desgns we hae performed so far hae been statc, and all

More information

Copyright 2004 by Oxford University Press, Inc.

Copyright 2004 by Oxford University Press, Inc. JT as an Amplfer &a Swtch, Large Sgnal Operaton, Graphcal Analyss, JT at D, asng JT, Small Sgnal Operaton Model, Hybrd P-Model, TModel. Lecture # 7 1 Drecton of urrent Flow & Operaton for Amplfer Applcaton

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

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

Week 11: Differential Amplifiers

Week 11: Differential Amplifiers ELE 0A Electronc rcuts Week : Dfferental Amplfers Lecture - Large sgnal analyss Topcs to coer A analyss Half-crcut analyss eadng Assgnment: hap 5.-5.8 of Jaeger and Blalock or hap 7. - 7.3, of Sedra and

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

FEEDBACK AMPLIFIERS. v i or v s v 0

FEEDBACK AMPLIFIERS. v i or v s v 0 FEEDBCK MPLIFIERS Feedback n mplers FEEDBCK IS THE PROCESS OF FEEDING FRCTION OF OUTPUT ENERGY (VOLTGE OR CURRENT) BCK TO THE INPUT CIRCUIT. THE CIRCUIT EMPLOYED FOR THIS PURPOSE IS CLLED FEEDBCK NETWORK.

More information

Affine transformations and convexity

Affine transformations and convexity Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that

More information

matter consists, measured in coulombs (C) 1 C of charge requires electrons Law of conservation of charge: charge cannot be created or

matter consists, measured in coulombs (C) 1 C of charge requires electrons Law of conservation of charge: charge cannot be created or Basc Concepts Oerew SI Prefxes Defntons: Current, Voltage, Power, & Energy Passe sgn conenton Crcut elements Ideal s Portland State Unersty ECE 221 Basc Concepts Ver. 1.24 1 Crcut Analyss: Introducton

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

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 Decibel and its Usage

The Decibel and its Usage The Decbel and ts Usage Consder a two-stage amlfer system, as shown n Fg.. Each amlfer rodes an ncrease of the sgnal ower. Ths effect s referred to as the ower gan,, of the amlfer. Ths means that the sgnal

More information

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES BÂRZĂ, Slvu Faculty of Mathematcs-Informatcs Spru Haret Unversty barza_slvu@yahoo.com Abstract Ths paper wants to contnue

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

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

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

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

New Turnpike Theorems for the Unbounded Knapsack Problem

New Turnpike Theorems for the Unbounded Knapsack Problem New Turnpke Theorems for the Unbounded Knapsack Problem Png Hed Huang¹* and Thomas L. Morn² ¹Krannert School of Management, ²School of Industral Engneerng Purdue Unersty, West Lafayette, Indana 47907 Abstract:

More information

6.01: Introduction to EECS 1 Week 6 October 15, 2009

6.01: Introduction to EECS 1 Week 6 October 15, 2009 6.0: ntroducton to EECS Week 6 October 5, 2009 6.0: ntroducton to EECS Crcuts The Crcut Abstracton Crcuts represent systems as connectons of component through whch currents (through arables) flow and across

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

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

EEE 241: Linear Systems

EEE 241: Linear Systems EEE : Lnear Systems Summary #: Backpropagaton BACKPROPAGATION The perceptron rule as well as the Wdrow Hoff learnng were desgned to tran sngle layer networks. They suffer from the same dsadvantage: they

More information

R. W. Erickson. Department of Electrical, Computer, and Energy Engineering University of Colorado, Boulder

R. W. Erickson. Department of Electrical, Computer, and Energy Engineering University of Colorado, Boulder R. W. Erckson Department of Electrcal, Computer, and Energy Engneerng Unersty of Colorado, Boulder 3.5. Example: ncluson of semconductor conducton losses n the boost conerter model Boost conerter example

More information

Lecture 21: Numerical methods for pricing American type derivatives

Lecture 21: Numerical methods for pricing American type derivatives Lecture 21: Numercal methods for prcng Amercan type dervatves Xaoguang Wang STAT 598W Aprl 10th, 2014 (STAT 598W) Lecture 21 1 / 26 Outlne 1 Fnte Dfference Method Explct Method Penalty Method (STAT 598W)

More information

Fundamental loop-current method using virtual voltage sources technique for special cases

Fundamental loop-current method using virtual voltage sources technique for special cases Fundamental loop-current method usng vrtual voltage sources technque for specal cases George E. Chatzaraks, 1 Marna D. Tortorel 1 and Anastasos D. Tzolas 1 Electrcal and Electroncs Engneerng Departments,

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

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

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

On a direct solver for linear least squares problems

On a direct solver for linear least squares problems ISSN 2066-6594 Ann. Acad. Rom. Sc. Ser. Math. Appl. Vol. 8, No. 2/2016 On a drect solver for lnear least squares problems Constantn Popa Abstract The Null Space (NS) algorthm s a drect solver for lnear

More information

6.854J / J Advanced Algorithms Fall 2008

6.854J / J Advanced Algorithms Fall 2008 MIT OpenCourseWare http://ocw.mt.edu 6.854J / 18.415J Advanced Algorthms Fall 2008 For nformaton about ctng these materals or our Terms of Use, vst: http://ocw.mt.edu/terms. 18.415/6.854 Advanced Algorthms

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

3.2 Terminal Characteristics of Junction Diodes (pp )

3.2 Terminal Characteristics of Junction Diodes (pp ) /9/008 secton3_termnal_characterstcs_of_juncton_odes.doc /6 3. Termnal Characterstcs of Juncton odes (pp.47-53) A Juncton ode I.E., A real dode! Smlar to an deal dode, ts crcut symbol s: HO: The Juncton

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

The Second Anti-Mathima on Game Theory

The Second Anti-Mathima on Game Theory The Second Ant-Mathma on Game Theory Ath. Kehagas December 1 2006 1 Introducton In ths note we wll examne the noton of game equlbrum for three types of games 1. 2-player 2-acton zero-sum games 2. 2-player

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

Research on the Method of Calculating Node Injected Reactive Power Based on L Indicator

Research on the Method of Calculating Node Injected Reactive Power Based on L Indicator Journal of Power and Energy Engneerng, 2014, 2, 361-367 Publshed Onlne Aprl 2014 n ScRes. http://www.scrp.org/ournal/pee http://dx.do.org/10.4236/pee.2014.24048 Research on the Method of Calculatng Node

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

Boise State University Department of Electrical and Computer Engineering ECE 212L Circuit Analysis and Design Lab

Boise State University Department of Electrical and Computer Engineering ECE 212L Circuit Analysis and Design Lab Bose State Unersty Department of Electrcal and omputer Engneerng EE 1L rcut Analyss and Desgn Lab Experment #8: The Integratng and Dfferentatng Op-Amp rcuts 1 Objectes The objectes of ths laboratory experment

More information

Systems of Equations (SUR, GMM, and 3SLS)

Systems of Equations (SUR, GMM, and 3SLS) Lecture otes on Advanced Econometrcs Takash Yamano Fall Semester 4 Lecture 4: Sstems of Equatons (SUR, MM, and 3SLS) Seemngl Unrelated Regresson (SUR) Model Consder a set of lnear equatons: $ + ɛ $ + ɛ

More information

(1 ) (1 ) 0 (1 ) (1 ) 0

(1 ) (1 ) 0 (1 ) (1 ) 0 Appendx A Appendx A contans proofs for resubmsson "Contractng Informaton Securty n the Presence of Double oral Hazard" Proof of Lemma 1: Assume that, to the contrary, BS efforts are achevable under a blateral

More information

The optimal delay of the second test is therefore approximately 210 hours earlier than =2.

The optimal delay of the second test is therefore approximately 210 hours earlier than =2. THE IEC 61508 FORMULAS 223 The optmal delay of the second test s therefore approxmately 210 hours earler than =2. 8.4 The IEC 61508 Formulas IEC 61508-6 provdes approxmaton formulas for the PF for smple

More information

CHAPTER III Neural Networks as Associative Memory

CHAPTER III Neural Networks as Associative Memory CHAPTER III Neural Networs as Assocatve Memory Introducton One of the prmary functons of the bran s assocatve memory. We assocate the faces wth names, letters wth sounds, or we can recognze the people

More information

Linearity. If kx is applied to the element, the output must be ky. kx ky. 2. additivity property. x 1 y 1, x 2 y 2

Linearity. If kx is applied to the element, the output must be ky. kx ky. 2. additivity property. x 1 y 1, x 2 y 2 Lnearty An element s sad to be lnear f t satsfes homogenety (scalng) property and addte (superposton) property. 1. homogenety property Let x be the nput and y be the output of an element. x y If kx s appled

More information

Supplementary Notes for Chapter 9 Mixture Thermodynamics

Supplementary Notes for Chapter 9 Mixture Thermodynamics Supplementary Notes for Chapter 9 Mxture Thermodynamcs Key ponts Nne major topcs of Chapter 9 are revewed below: 1. Notaton and operatonal equatons for mxtures 2. PVTN EOSs for mxtures 3. General effects

More information

CHAPTER 13. Exercises. E13.1 The emitter current is given by the Shockley equation:

CHAPTER 13. Exercises. E13.1 The emitter current is given by the Shockley equation: HPT 3 xercses 3. The emtter current s gen by the Shockley equaton: S exp VT For operaton wth, we hae exp >> S >>, and we can wrte VT S exp VT Solng for, we hae 3. 0 6ln 78.4 mv 0 0.784 5 4.86 V VT ln 4

More information

Second Order Analysis

Second Order Analysis Second Order Analyss In the prevous classes we looked at a method that determnes the load correspondng to a state of bfurcaton equlbrum of a perfect frame by egenvalye analyss The system was assumed to

More information

Graphical Analysis of a BJT Amplifier

Graphical Analysis of a BJT Amplifier 4/6/2011 A Graphcal Analyss of a BJT Amplfer lecture 1/18 Graphcal Analyss of a BJT Amplfer onsder agan ths smple BJT amplfer: ( t) = + ( t) O O o B + We note that for ths amplfer, the output oltage s

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

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

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur Module Random Processes Lesson 6 Functons of Random Varables After readng ths lesson, ou wll learn about cdf of functon of a random varable. Formula for determnng the pdf of a random varable. Let, X be

More information

Boise State University Department of Electrical and Computer Engineering ECE 212L Circuit Analysis and Design Lab

Boise State University Department of Electrical and Computer Engineering ECE 212L Circuit Analysis and Design Lab Bose State Unersty Department of Electrcal and omputer Engneerng EE 1L rcut Analyss and Desgn Lab Experment #8: The Integratng and Dfferentatng Op-Amp rcuts 1 Objectes The objectes of ths laboratory experment

More information

12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA. 4. Tensor product

12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA. 4. Tensor product 12 MATH 101A: ALGEBRA I, PART C: MULTILINEAR ALGEBRA Here s an outlne of what I dd: (1) categorcal defnton (2) constructon (3) lst of basc propertes (4) dstrbutve property (5) rght exactness (6) localzaton

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

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

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

Circuit Variables. Unit: volt (V = J/C)

Circuit Variables. Unit: volt (V = J/C) Crcut Varables Scentfc nestgaton of statc electrcty was done n late 700 s and Coulomb s credted wth most of the dscoeres. He found that electrc charges hae two attrbutes: amount and polarty. There are

More information

Chapter 7 Partially Balanced Incomplete Block Design (PBIBD)

Chapter 7 Partially Balanced Incomplete Block Design (PBIBD) Chapter 7 Partally Balanced Incomplete Block Desgn (PBIBD) The balanced ncomplete block desgns hae seeral adantages. They are connected desgns as well as the block szes are also equal. A restrcton on usng

More information

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso Supplement: Proofs and Techncal Detals for The Soluton Path of the Generalzed Lasso Ryan J. Tbshran Jonathan Taylor In ths document we gve supplementary detals to the paper The Soluton Path of the Generalzed

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

A New Refinement of Jacobi Method for Solution of Linear System Equations AX=b

A New Refinement of Jacobi Method for Solution of Linear System Equations AX=b Int J Contemp Math Scences, Vol 3, 28, no 17, 819-827 A New Refnement of Jacob Method for Soluton of Lnear System Equatons AX=b F Naem Dafchah Department of Mathematcs, Faculty of Scences Unversty of Gulan,

More information

e - c o m p a n i o n

e - c o m p a n i o n OPERATIONS RESEARCH http://dxdoorg/0287/opre007ec e - c o m p a n o n ONLY AVAILABLE IN ELECTRONIC FORM 202 INFORMS Electronc Companon Generalzed Quantty Competton for Multple Products and Loss of Effcency

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

Volume 18 Figure 1. Notation 1. Notation 2. Observation 1. Remark 1. Remark 2. Remark 3. Remark 4. Remark 5. Remark 6. Theorem A [2]. Theorem B [2].

Volume 18 Figure 1. Notation 1. Notation 2. Observation 1. Remark 1. Remark 2. Remark 3. Remark 4. Remark 5. Remark 6. Theorem A [2]. Theorem B [2]. Bulletn of Mathematcal Scences and Applcatons Submtted: 016-04-07 ISSN: 78-9634, Vol. 18, pp 1-10 Revsed: 016-09-08 do:10.1805/www.scpress.com/bmsa.18.1 Accepted: 016-10-13 017 ScPress Ltd., Swtzerland

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

EE215 FUNDAMENTALS OF ELECTRICAL ENGINEERING

EE215 FUNDAMENTALS OF ELECTRICAL ENGINEERING EE215 FUNDAMENTALS OF ELECTRICAL ENGINEERING TaChang Chen Unersty of Washngton, Bothell Sprng 2010 EE215 1 WEEK 8 FIRST ORDER CIRCUIT RESPONSE May 21 st, 2010 EE215 2 1 QUESTIONS TO ANSWER Frst order crcuts

More information

Spectral Graph Theory and its Applications September 16, Lecture 5

Spectral Graph Theory and its Applications September 16, Lecture 5 Spectral Graph Theory and ts Applcatons September 16, 2004 Lecturer: Danel A. Spelman Lecture 5 5.1 Introducton In ths lecture, we wll prove the followng theorem: Theorem 5.1.1. Let G be a planar graph

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

Singular Value Decomposition: Theory and Applications

Singular Value Decomposition: Theory and Applications Sngular Value Decomposton: Theory and Applcatons Danel Khashab Sprng 2015 Last Update: March 2, 2015 1 Introducton A = UDV where columns of U and V are orthonormal and matrx D s dagonal wth postve real

More information

An Interactive Optimisation Tool for Allocation Problems

An Interactive Optimisation Tool for Allocation Problems An Interactve Optmsaton ool for Allocaton Problems Fredr Bonäs, Joam Westerlund and apo Westerlund Process Desgn Laboratory, Faculty of echnology, Åbo Aadem Unversty, uru 20500, Fnland hs paper presents

More information

COMPLEX NUMBERS AND QUADRATIC EQUATIONS

COMPLEX NUMBERS AND QUADRATIC EQUATIONS COMPLEX NUMBERS AND QUADRATIC EQUATIONS INTRODUCTION We know that x 0 for all x R e the square of a real number (whether postve, negatve or ero) s non-negatve Hence the equatons x, x, x + 7 0 etc are not

More information

55:041 Electronic Circuits

55:041 Electronic Circuits 55:04 Electronc Crcuts Feedback & Stablty Sectons of Chapter 2. Kruger Feedback & Stablty Confguraton of Feedback mplfer Negate feedback β s the feedback transfer functon S o S S o o S S o f S S S S fb

More information

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens THE CHINESE REMAINDER THEOREM KEITH CONRAD We should thank the Chnese for ther wonderful remander theorem. Glenn Stevens 1. Introducton The Chnese remander theorem says we can unquely solve any par of

More information

Chapter 2 A Class of Robust Solution for Linear Bilevel Programming

Chapter 2 A Class of Robust Solution for Linear Bilevel Programming Chapter 2 A Class of Robust Soluton for Lnear Blevel Programmng Bo Lu, Bo L and Yan L Abstract Under the way of the centralzed decson-makng, the lnear b-level programmng (BLP) whose coeffcents are supposed

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