ARTICLE IN PRESS. Control Engineering Practice

Size: px
Start display at page:

Download "ARTICLE IN PRESS. Control Engineering Practice"

Transcription

1 Control Engineering Practice 17 (29) Contents lists available at ScienceDirect Control Engineering Practice journal homepage: wwwelseviercom/locate/conengprac Set-point variation in learning schemes with applications to wafer scanners Marcel F Heertjes a,b,, René MJG van de Molengraft a a Department of Mechanical Engineering, Dynamics and Control Technology, Eindhoven University of Technology, Den Dolech 2, 56 MD Eindhoven, The Netherlands b ASML, Mechatronic Systems Development, De Run 651, 554 DR Veldhoven, The Netherlands article info Article history: Received 17 November 27 Accepted 18 August 28 Available online 5 November 28 Keywords: Finite impulse response modeling Multi-input multi-output feed-forward design Motion control systems (Nonlinear) Iterative learning control Wafer scanners abstract This paper presents a finite impulse response strategy to deal with set-point variation in learning schemes On the basis of converged learning forces obtained with learning control at a specific acceleration set-point profile, a finite impulse response mapping is derived to generalize the learned forces at a specific set-point toward arbitrary set-point profiles, thus relaxing the need for further learning The above strategy is applied to the motion control systems of a wafer scanner in a multi-input multi-output feed-forward setting, where a variety of set-point profiles is used Industrial potential is demonstrated via robustness to set-point variation and the improvements obtained in settling-time reduction & 28 Elsevier Ltd All rights reserved 1 Introduction In high-speed motion systems such as the reticle and the wafer stages of a wafer scanner (Van de Wal, Van Baars, Sperling, & Bosgra, 22) learning can significantly improve upon performance This is because of the repetitive nature of its scanning motion (Dijkstra & Bosgra, 22; Rotariu, Ellenbroek, & Steinbuch, 23; Rotariu, Dijkstra, & Steinbuch, 24) In learning, information of previous executions of a repeated motion is used to update a command (usually a force) needed to counteract the effect of such motion at future executions, see for example Bristow, Tharayil, and Alleyne (26), Cai, Freeman, Lewin, and Rogers (28), and Moore (1999) for learning algorithms and Gunnarsson and Norrlöf (21) and Tousain and Van der Meché (21) for learning designs Variation in the scanning motion, however, avoids the application of the resulting commands learned at a specific motion to be effective in achieving performance when applied during a different motion; see also Xu (1998), Xu and Tan (22), Dixon and Chen (23), Xu and Tan (23), and Rotariu, Ellenbroek, Van Baars, and Steinbuch (23) for a related problem statement For this purpose a finite impulse response mapping is used as proposed by Potsaid and Wen (24) The forces learned for a representative acceleration set-point are mapped onto a finite impulse response (FIR) model In the wafer scanning example, this is done prior to the process of wafer illumination whereas during this process the learned forces are replaced by generalized learned forces being the result of the finite impulse response model and the acceleration set-points at hand; this is different from a run-to-run control approach such as for example considered by Bode, Ko, and Edgar (24) which lacks in situ performance measurement (the wafer needs to be further processed) and in which all set-points are known In a general multi-input multi-output feed-forward setting, the advantages are twofold On one hand, learning during the process of wafer scanning is avoided This maintains the high standard of performance in terms of wafer throughput On the other hand, learning is based on a small sub-set of a generally large variation of wafer set-points This constitutes the efficiency of the method This paper has three contributions: (i) a generalized learning through FIR modeling for motion systems, (ii) a multi-input multioutput learning approach in case these motion systems are weakly coupled, and (iii) an application of learning in the field of industrial wafer scanners The paper is further organized as follows In Section 2, the wafer scanner application, which serves as an experimental benchmark, is discussed in terms of its relevant motion control sub-systems, in particular, the reticle and wafer stage Section 3 considers learning in the repetitive context of scanning motion Section 4 deals with FIR modeling for multiinput multi-output feed-forward design A performance assessment using examples from an industrial wafer stage is presented in Section 5 Section 6 summarizes the main findings of the paper Corresponding author at: Department of Mechanical Engineering, Dynamics and Control Technology, Eindhoven University of Technology, Den Dolech 2, 56 MD Eindhoven, The Netherlands Tel: ; fax: addresses: mfheertjes@tuenl, marcelheertjes@asmlcom (MF Heertjes), mjgvdmolengraft@tuenl (RMJG van de Molengraft) 2 Dynamics and control of wafer scanners During the lithographic manufacturing of integrated circuits (ICs) wafer scanners achieve performance by combining nano /$ - see front matter & 28 Elsevier Ltd All rights reserved doi:1116/jconengprac2884

2 346 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) scale resolution with optimized wafer throughput The wafer scanning process, see for example Groot-Wassink, Van de Wal, Scherer, and Bosgra (25) and Heertjes and Van de Wouw (26), can be described as follows Light from a laser passes from a reticle which contains an image, through a lens, which scales down the image, onto a wafer, see Fig 1 Both reticle and wafer are part of two separate sub-systems: the reticle stage and the wafer stage Each stage employs a dual stroke principle: a long stroke for large-range motion and a short stroke for accurate positioning The short-stroke modules can be modeled as floating masses which are controlled in six degrees-of-freedom on a single-input single-output basis A distinction is made between scanning directions and non-scanning directions The scanning directions, for example the x- and y-directions of the short-stroke wafer stage, are controlled on the basis of both feedback and feedforward The non-scanning directions, for example the x- and z-direction of the short-stroke reticle stage, are mainly controlled by feedback Each of the controlled short-stroke directions can be represented by the simplified block diagram representation of Fig 2 which can be considered in the more general context of motion control systems On the basis of an acceleration set-point a and resulting command r, a servo error signal e is constructed via e ¼ r y with y the actual position (or angle depending on the choice of axis) of the considered plant P, in this case a short-stroke stage The error signal e is fed into a feedback controller C fb that aims at disturbance rejection mainly induced by force disturbances f To obtain sufficient tracking accuracy, an SISO (and model-based) feed-forward controller C ff is added reticle stage lens The short-stroke (electro-)mechanics of both the reticle- and wafer stages in the individual directions are characterized by double integrator behavior along with the expression of higherorder dynamics In controlling such dynamics, the feedback controller C fb is chosen as a series connection of three filter blocks: a proportional-integrator-derivative (PID) filter, which aims at both disturbance rejection and robust stability, a secondorder low-pass filter to avoid high-frequency noise amplification, and several notch filters to deal with resonant behavior in the plant In the frequency-domain, the controlled electro-mechanics are characterized by the open-loop frequency response function O l ðjoþ ¼C fb ðjoþpðjoþ such as depicted in Fig 3 for the scanning y-direction of both a reticle and a wafer stage In Bode representation, this figure shows the characteristics derived from a closed-loop measurement (solid) along with the characteristics of a model (dashed) From this figure, it is concluded that robust stability is sufficiently guaranteed The reticle and wafer stage can be considered in the simplified MIMO context of Fig 4 Different from Fig 2, interaction between the feedback loops is modeled using cross-talk forces acting on the considered plants: f xy acting on P yy and f yx acting on P xx From stability point of view, these MIMO forces remain small enough to justify the SISO feedback/feed-forward design approach related to Fig 2 From performance standpoint, this is not the case Given the tight performance specifications this industry is faced with, it suffices to state that a zero settling control aim is not possible without at least including some MIMO characteristics in the feed-forward design This is the purpose of the proposed learning such as presented in the next sections Herein the databased MIMO feed-forward contributions are used atop the modelbased SISO feed-forwards The SISO feed-forwards, which strictly speaking become redundant, are kept as an initial estimate in achieving performance 3 Iterative learning control 25m wafer stage 45m Fig 1 Wafer scanner representation and its main components For controlled processes exhibiting repeated motion, iterative learning control provides a means to learn updated commands (or forces) from past servo information and apply these commands at future executions (or trials) of such motion, see also Chen and Hwang (25), Mishra, Coaplen, and Tomizuka (27), and Tayebi and Islam (26) The latter with the aim to improve upon servo performance The MIMO context in which iterative learning control is applied to the short-stroke stages of a wafer scanner is expressed by the simplified block diagram representation of Fig 5 Different from Fig 4, each axis contains a set of learning controllers C ilc 2 fc ilc;xx ; C ilc;xy ; C ilc;yx ; C ilc;yy g used to counteract the recurring error contributions e x and e y induced by the acceleration set-point profiles a x and a y For example, the x-axis contains a learning controller C ilc;xx which generates a force f ilc;xx This force is used to counteract the recurring part of the error response e x which is induced by the acceleration set-point a x ; note that this is the error C ff (s) a 1 r e C fb (s) - f ff f P (s) y Fig 2 Simplified block diagram representation of a motion control system

3 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) amplitude in db rs modelled ws modelled rs measured ws measured phase in degrees Fig 3 Bode representation of the measured open-loop frequency response functions of the short-stroke wafer (ws) and reticle (rs) stage dynamics in scanning direction along with the characteristics of two double integrator-based models C ff, xx f ff, xx C ilc, xx f ilc, xx a x 1 r x e x C fb, xx a y 1 r y e y C fb, yy f xy P xx P yy f yx y x y y C ilc, yx C ff, xx f ilc, yx f ff, xx C ff, yy f ff, yy Fig 4 Simplified block diagram representation of the MIMO controller structure such as encountered for both the short-stroke wafer and reticle stages a x 1 r x e x C fb, xx a y 1 r y e y C fb, yy f xy P xx P yy f yx y x y y response after application of the SISO model-based feed-forward controller C ff ;xx Additionally but applied in the y-axis C ilc;xy generates a force f ilc;xy used to counteract the recurring error response e y The latter being the result of cross-talk induced by the set-point a x The effectiveness of such a coupled system compensation stems from the assumption that the coupling is small enough to endanger SISO stability but large enough to improve upon MIMO performance Hence the wafer stage plant of about 225 kg, which is accelerated in the y-direction with 27:5ms 2, requires feed-forward forces near C ff ;yy 62 N Since f ilc;yy and f ilc;yx are typically in the order of 1 N, see Heertjes and Tso (27a), a sufficiently small coupling validates an essentially SISO-based stability approach However given a typical controller gain of Nm 1 such learning forces correspond to errors in the order of 5 nm which are large enough to allow for a significant MIMO performance enhancement The learned forces f ilc in Fig 5 comply with the schematics of Fig 6; see Heertjes and Tso (27a) for a detailed SISO C ff, yy C ilc, xy C ilc, yy f ff, yy f ilc, xy f ilc, yy Fig 5 Simplified block diagram representation of a simplified linear feedback connection of two coupled short-stroke axes having iterative learning control description That is, the error signals e related to a single execution of the acceleration set-point are stored in a buffer to form the error column eðkþ where k 2 N ers to the k-th execution (or trial) The buffer output is subjected to the nonlinear

4 348 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) weighting U, or UðeðkÞÞ ¼ diagðfðeðkþf1gþ; fðeðkþf2gþ; ; fðeðkþfngþþx, (1) with 8 >< fðxþ ¼ 1 d if jxj4d; jxj >: if jxjpd: All entries in eðkþ that are bounded in absolute value by a threshold level dx are assumed to be noise contributions and as such are excluded from learning; typically d is chosen equal to the L 1 -norm of the steady-state signals obtained after all transients have sufficiently damped out Though limited in fully capturing the noise characteristics, the choice for d along with the structure of (2) stems from both (nonlinear) stability and performance considerations such as given in Heertjes and Tso (27a, 27b) After this weighting, the error signals are subjected to a linear learning gain L which is given by L ¼ðS T p S p þ liþ 1 S T p, (3) with tuning parameter l4, see Ghosh and Paden (22) Basically, l is given the smallest possible value for which the learning process is stable, such that L closely resembles the e buffer C ilc buffer e (k) f ilc (k) f Φ ( ) ilc (k1) L z 1 I Fig 6 Block diagram representation of the iterative learning control f ilc (2) inverse process sensitivity S 1 p and a fast convergence is obtained S p 2 R nconn obs is given by 2 3 h 1 h 2 h 1 S p ¼, (4) h ncon h ncon 1 h ncon n obs þ1 where n con 2 N is the number of samples accessible for the learning force, the so-called controller window and n ob N is the number of samples used for error evaluation, named the observation window (Dijkstra & Bosgra, 22) S p has a Toeplitz structure where h 1 ; h 2 ; ; h n represent the Markov parameters h 1 represents the first error response sample e to a unitary force impulse f The Markov parameters are obtained from measurement and typically lect the average of 25 subsequent impulse response measurements as to reduce the effect of measurement noise, see also Ahn, Moore, and Chen (26) and Moore, Chen, and Bahl (25) for dealing with model uncertainty in this regard The learning filters C ilc;xx and C ilc;yx (see Fig 5) are both based on S p;xx, hence they relate to the error response e x induced by a unitary force impulse f x With S p, the updated learned forces f ilc ðk þ 1Þ are given by f ilc ðk þ 1Þ ¼f ilc ðkþþluðeðkþþeðkþ, (5) where f ilc ðkþ after buffering (see Fig 6) gives the learned forces f ilc For a detailed treatment of the stability and convergence properties of such a nonlinear learning controller, the reader is erred to Heertjes and Tso (27a, 27b) For a short-stroke reticle stage module, the need for MIMO learning in view of (settling) performance is illustrated in Fig 7 Given a representative acceleration profile a y, which in scaled form is depicted in the upper left part of the figure (dashed curve), the servo error signals in y- and z-directions show significant response to the variation included in this profile Note that this is the response encountered under SISO model-based 4 4 e y in nm e z in nm E y in nm 3 3 E z in nm 3 3 cpsd of E y in nm cpsd of E z in nm Fig 7 Time-series measurement of the error signals in y- and z-directions of a short stroke reticle stage (upper part) along with the non-recurring residuals (lower part) of 2 different scans

5 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) feed-forward conditions In the interval where performance is required, ie, the scanning interval of constant velocity, the responses e y and e z induce an undesired settling time needed for the error to become sufficiently small The reproducibility of the settling phenomenon is shown in the lower part of the figure By comparing 2 separately measured error traces obtained in subsequent trials it can be seen that during constant velocity (beyond t ¼ :35 s) the residual errors E roughly remain inside a noise bound of 3 nm for the y-axis (left part) and 3 nm for the z-axis (right part); E is defined as Efig ¼efig P n i¼1efig=n, with n ¼ 2 and i erring to an error trace realization A similar observation follows for the root-mean-square values such as obtained via cumulative power spectral density (cpsd) analysis, which is shown in the lower part of the figure In the MIMO context of Fig 5, the nonlinear learning control is applied to the reticle stage system with the aim of achieving zero settling times without transmitting too much noise through the learned forces The result of this is shown in Fig 8 For the nonlinear filter setting d y ¼ 3 nm and d z ¼ 3 nm, it can be seen that the servo errors after 1 trials (ilc) roughly remain below the threshold levels This gives a significant improvemenervo performance in comparison with the errors before learning () The amplification of noises through learning is kept small This is expressed by comparing 2 error traces With each trace being the result of a 1-trial learning applied under similar initial conditions, it can be seen that the noise level does not significantly differ from the noise level encountered before learning in Fig 7 Note that the idea of using ILC to compensate the recurrent part of the error but with a potential deterioration of the non-recurrent part is not restricted to the wafer scanner application, see, for example, Chen, Moore, Yu, and Zhang (28) in which hard disk drive servo performance is considered in terms of repeatable runouts (RRO) and non-repeatable run-outs (NRRO) In the wafer scanning process, the motivation for the application of learning techniques stems from the recurring acceleration set-point behavior during the process of wafer illumination At the same time, however, this process shows sufficient deviation in terms of set-points and servo errors to limit standard learning schemes from being effective This is illustrated in Fig 9 where a typical wafer scanning path during a job is depicted Starting from a fixed location outside the wafer (see the upper left part of the figure) an up-and-down scanning pattern in terms of third-order position set-points is commanded on a grid of equally distributed x- and y-locations along the wafer The resulting acceleration setpoints show much similarity which forms the basis of a recurring error response However variation occurs For example, at the crossings from a previous array of scans toward the next array (this is indicated by the dotted path) the acceleration set-points become different This mostly involves variation in the interval of y-position in mm wafer scanning path 3 mm wafer wafer stage top view x-position in mm Fig 9 Graphical representation of a wafer scanning path during a job e y in nm 4 ilc e z in nm E y in nm 3 3 E z in nm 3 3 cpsd of E y in nm cpsd of E z in nm Fig 8 Time-series measurement of the error signals in y- and z-directions of a short stroke reticle stage (upper part) before (gray) and after (black) learning (1 trials) along with the non-recurring residuals (lower part) of 2 different learned scans; l y ¼ 1 17, l z ¼ 1 15, d y ¼ 3 nm, d z ¼ 3 nm, n con ¼ 226, n obs ¼ 275

6 35 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) maximum acceleration where the time needed to reach the desired velocity varies, but also variation in the interval of maximum jerk, for example, when the desired velocity is reached prior to reaching the level of maximum acceleration All of this is related to the fact that the motion parameters such as maximum acceleration and jerk are generally fixed and chosen maximal as to reduce processing time As a result of set-point variation, however, the dynamics are excited differently inducing a different error response The previously learned forces based on earlier error behavior then become less effective or possibly even ineffective Since overall wafer performance is as good as the worst scan in a job such ineffectiveness is generally not acceptable and as such provides the motivation for including robustness against set-point variation in the learning process 4 FIR mapping In dealing with set-point variation, an FIR mapping is considered After convergence of the learning process, the ILC forces are mapped onto the corresponding acceleration set-point profiles using FIR modeling, see also Potsaid and Wen (24) For arbitrary scanning set-points, the resulting models are used to compute the generalized forces Essentially this reduces the learning strategy to a learned feed-forward design; see also Van der Meulen, Tousain, and Bosgra (28) and Baggen, Heertjes, and Kamidi (28) for learned feed-forward design based on optimization An MIMO learned feed-forward design is depicted in the simplified block diagram representation of Fig 1 Different from Fig 5 the FIR controllers C fir 2fC fir;xx ; C fir;xy ; C fir;yx ; C fir;yy g relate the acceleration set-points a x and a y with the generalized forces f fir 2ff fir;xx ; f fir;xy ; f fir;yx ; f fir;yy g For example, the acceleration setpoint a x serves as input to the FIR controller C fir;xx which generates a force f fir;xx used to counteract the error response e x induced by this set-point Additionally, C fir;xy is used to counteract the crosstalk effect of a x to the error response e y Note that C fir;xy is depicted in the upper part of the figure because a x serves as an input to the FIR mapping This is different from Fig 5 where C ilc;xy has e y as input and theore is located in the lower part of the figure The FIR controllers C fir are given in discrete time by f fir ðkþ ¼c 1 aðk þ n nc Þþþc n aðk n c þ 1Þ; k 2 Z, (6) with n nc the number of non-causal time samples, n c the number of causal time samples, and n ¼ n nc þ n c 2 N Herein it is important to note that the filter order n is often obtained by trial and error It is bound, however, by the following trade-off: choosing n too small limits the ability to describe the required feed-forward signals Choosing n too large involves the risk of over-fitting The coefficients c i with i 2f1; ; ng are obtained from a least-squares optimization That is, under the assumption that the learned (and converged) forces can be described by f ilc ¼ A ilc c, (7) with f ilc ¼½f ilc ðkþ f ilc ðk þ oþš T 2 R o1 a representative (training) set of ILC forces (oxn denoting the number of learned data points), the non-square matrix 2 a ilc ðk þ n nc Þ ::: 3 a ilc ðk n c þ 1Þ A ilc ¼ , (8) a ilc ðk þ n nc þ o 1Þ ::: a ilc ðk n c þ oþ with A ilc 2 R on, and c ¼½c 1 c n Š T 2 R n1, the FIR coefficients read c ¼ðA T ilc A ilcþ 1 A T ilc f ilc (9) a x 1 r x e x Note that although set-point variation occurs, during a wafer scanning job most of the scans are conducted under equal setpoint conditions This makes it intuitively clear to adopt these conditions as being most representative for training the set of ILC forces and obtaining the set of FIR coefficients Note moreover that the existence of a set of FIR coefficients c in (9) is based on the assumption that A T ilc A ilc is invertible The validity of this assumption and the related conditions on persistent excitation (PE) follow from the next result Theorem 1 Assume that A ilc in (8) is chosen such that a ilc ðk þ n nc Þa, oxn41, and A ilc does not contain columns having o equal entries, then A T ilc A ilc is invertible Proof See the Appendix The input output relation of the considered mapping has two important steady-state properties: first, a constant (acceleration) input gives a constant output and, second, an input with a constant slope gives an output with a constant slope For the class of third-order position set-points in which the acceleration profiles are composed of regions of constant output and regions of constant slope only, the first property follows from the fact that for a constant input aðmþ ¼¼aðm n þ 1Þ with mxn, the computed force f fir ðmþ in steady-state reads f fir ðmþ ¼c 1 aðmþþþc n aðm n þ 1Þ ¼ Xn i¼1 C fir, xx C fir, xy C ff, xx a y 1 r y e y C ff, yy C fir, yx C fir, yy f fir, xx f fir, xy f ff, xx C fb, xx C fb, yy f ff, yy f fir, yx f fir, yy Fig 1 Block diagram representation of a simplified feedback connection of a short-stroke x- and y-axes in a single direction having learned feed-forward control P xx c i aðmþ (1) f xy P yy f yx y x y y

7 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) The second property follows from re-writing (1), or f fir ðm þ 1Þ ¼f fir ðmþþ Xn i¼1 c i Da, (11) with Da ¼ aðm þ 1Þ aðmþ the fixed variation during one timesample For n ¼ n c ¼ 2 both properties can be distinguished from the FIR forces such as depicted in Fig 11 This figure also shows the corresponding ILC forces and acceleration profile; the latter being scaled with the sum of the filter coefficients, ie, P nc i¼1cðiþ In terms of FIR filter design, it can be seen that prior to the considered phases of constant slope, the ILC forces do not demonstrate the need for any non-causal filter contribution: the oscillations expressed by the ILC forces seem strictly noise related Theore, the mapping is studied in view of causal filter coefficients only, ie, n nc ¼ and n ¼ n c Note that the application of non-causal FIR filter coefficients can also be left to the least-squares optimization (by choosing n nc a) but with the possible effect of noise corruption by including non-relevant noise-, nonlinearity-, and non-set-point related contributions prior to the changes in the set-point profiles In the MIMO context of Figs 4, 5, and 1, Fig 12 shows the effect of: (i) no learning, (ii) an ILC such as discussed in the previous section, and (iii) a generalized learning control based on FIR modeling At a short-stroke wafer stage and in the absence of learning (), the effect of a single acceleration set-point in the x-axis (dashed curve) is clearly visible in terms of performance limiting error levels and settling times, see the upper part of the figure With learning (ilc), the error levels in both x and y-directions become significantly smaller In the middle part of the figure, it can be seen that the ILC forces f ilc;xx and f ilc;xy clearly relate to the set-point characteristics and show a good correspondence with the generalized FIR forces f fir;xx and f fir;xy (fir) This correspondence also applies to the error responses which can be seen both in time-domain (upper part) as well as in frequencydomain (lower part), the latter via cpsd analysis The ability to generalize the learned forces while maintaining servo performance is considered in Fig 13 In a cross validation experiment where two scanning velocities are considered: v ¼ 1:2ms 1 (left part) and v ¼ 2:4ms 1 (right part), the effect of learning is shown in terms of improved settling behavior At v ¼ 2:4ms 1, the learned forces (thick-gray) and corresponding set-point profile (dashed) are used to construct the FIR mapping f ilc, f fir in N 1 acc fir ilc 2 6 Fig 11 Time-series measurement of a learned force signal (thin) and the corresponding approximation (thick) along with the acceleration profile scaled with P nc i¼1cðiþ, nc ¼ 2 On the basis of this mapping, generalized FIR forces (thick-black) are applied to the coupled z-direction of a short-stroke reticle stage This is done at the set-point profile for which is learned but also for the set-point profile at v ¼ 1:2ms 1 for which in general is not learned Both in time-domain (upper part) as well as in frequency-domain representations (lower part), it can be seen that the generalizing properties of the FIR approach relate to a significant improvemenettling behavior, thereby showing a reasonable match with learning at each set-point separately In designing an FIR filter, the number of FIR coefficients n represents an important parameter in achieving servo performance To study its effect, different pairs of ILC forces and corresponding FIR models are compared, each pair related to a model based on a different number of FIR coefficients n ¼ n c 2f1; ; 55g The corresponding FIR forces are applied to a short-stroke wafer stage This is illustrated in Fig 14 where performance is assessed through the infinity norm of the nonscanning (coupled) error signal e rz Additionally, the effect is shown at a fixed number of coefficients n ¼ 3 by time-series measurement and cpsd analysis In a cross-validation context containing two scanning velocities v ¼ :6 and :3ms 1, it can be seen that beyond the large error reduction at n ¼ 5, only limited extra reduction is obtained by increasing the number of FIR coefficients The infinity norm is minimal at n ¼ 3, beyond which hardly any settling-induced behavior is found in the error signals The observation that a low-order feed-forward control design solves the performance problem to a large extent is widelyacknowledged in industrial motion control In this regard, the generalized FIR forces form no exception Apart from the number of FIR coefficients, the data interval used to construct the FIR mapping (this is indicated by n fir pn obs ) is an important design parameter This is because the linear mapping at hand demonstrates sensitivity to actuator/amplifier nonlinearity, 1 which is encountered when changing the direction of motion, the so-called direction dependency, but also when comparing the acceleration phase with the deceleration phase within a single motion Even when comparing the positive jerk (derivative of acceleration) phase with the negative jerk phase within a single acceleration (or deceleration) phase The effect is shown in Fig 15 at a reticle stage by considering two FIR controllers Each controller uses a different (but overlapping) interval of training data along the execution of the same acceleration set-point profile In the first interval, data is recorded from t ¼ 3:3 1 3 to 5:5 1 2 s(n fir ¼ 26) which covers the acceleration phase almost entirely In the second interval, data is recorded from t ¼ 3:2 1 2 to 5:5 1 2 s(n fir ¼ 118) which only contains data near the end of the acceleration phase From force time-series measurement, it becomes clear that the results related to interval 1 give a better approximation along the entire acceleration profile Since interval 2 only takes into account data near the end of the acceleration profile, the generalized forces at this part of the profile are very accurate accordingly From error time-series measurement this is the middle part of the figure it is shown that the forces based on interval 2 reduce the effect of settling behavior beyond the acceleration phase in a way similar to the ILC forces (indicated area) By evaluating the data (through the infinity norm) at the constant velocity phase, it can be seen that an optimum for n fir appears near the end of the acceleration phase (lower part) This is clear from the fact that earlier set-point excitation is largely excluded from the FIR mapping As a result, the FIR forces exactly match with the ILC forces beyond this phase The optimum is shown for the difference between the ILC and FIR 1 The amplifiers/actuators possess hysteresis as well as motor constant variation

8 352 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) e x in nm f ilc, xx, f fir, xx in N cpsd of e x in nm acc ilc fir e y in nm f ilc, xy, f fir, xy in N cpsd of e y in nm Fig 12 Time-series measurement of the servo error signals in scanning x-direction (left) and coupled y-direction (right) of a short stroke wafer stage before (gray), after learning (thin-black, 1 trials), and after the learned feed-forward control (thick-black); l ¼ 1 17, d ¼ 3 nm, n con ¼ 226, n obs ¼ 3, plus cumulative power spectral densities of the servo error signals e z in nm 4 v = 12 ms 1 acc ilc fir e z in nm 4 v = 24 ms 1 f ilc, yz, f fir, yz in N cpsd of e z in nm f ilc, yz, f fir, yz in N cpsd of e z in nm Fig 13 Time-series measurement of the servo error signals in the coupled z-direction of a short stroke reticle stage before (thin-black), after learning (thick-gray,1 trials), and after the learned feed-forward control (thick-black); l ¼ 1 15, d ¼ 1 nm, n con ¼ 22, n obs ¼ 325, plus cumulative power spectral densities of the servo error signals forces Df ¼ f ilc f fir (-symbols) but also for the servo errors e fir corresponding to the FIR forces (&-symbols), the latter being multiplied with the overall servo gain k p ¼ 2:3 1 7 Nm 1 Namely under the assumption that e ilc ¼, Df is related to e fir via the closed-loop process sensitivity function PðjoÞ e fir ðjoþ ¼ Df ðjoþ (12) 1 þ C fb ðjoþpðjoþ

9 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) e rz in μ rad 1 fir (6) fir (3) n c v = 6 ms 1 at n = 3 v = 3 ms 1 at n = 3 5 e rz in μ rad cpsd of e rz in μ rad acc fir e rz in μ rad cpsd of e rz in μ rad Fig 14 Time-series measurement of the error signal e rz evaluated (and cross-validated at two different scanning speeds) through the infinity norm and using FIR learning for a different number of FIR coefficients n 2f1; ; 55g in the non-scanning direction of a short-stroke wafer stage 1 interval 1 (n fir = 26) interval 2 (n fir = 118) 2 e y in nm f ilc, f fir in N ilc fir 4 6 in N interval 1 interval 2 4 Fig 15 Sensitivity of the finite impulse response mapping to expressions of nonlinear system behavior in the learned forces for the scanning y-direction of a short stroke reticle stage; l ¼ 1 17 m 2 N 2, d ¼ 5nm, n con ¼ 226, n obs ¼ 275, n ¼ n c ¼ 45 Below the bandwidth, it follows that ke fir ðjoþk kc fb ðjoþk 1 k Df ðjoþk For the considered PID design this can be simplified to ke fir ðjoþk kdf ðjoþk=k p From Fig 15, it is concluded that the presence of nonlinearity can severely limit the output of the mapping to resemble with the ILC forces and theore with the ILC error response

10 354 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) Performance assessment on a short-stroke wafer stage To demonstrate the potential of the learned feed-forward design in achieving robust performance, an industrial performance assessment is conducted on a short-stroke wafer stage In the analysis, robustness to set-point variation is tested by including different die sizes and scan velocities Performance is evaluated in terms of settling-time reduction In this regard, two industrial performance measures are considered: the moving average filter operation and the moving standard deviation filter operation (Heertjes & Van de Wouw, 26) The moving average filter operation expresses the level of position accuracy that can be obtained during the process of wafer scanning It has a strong relation with so-called scanning overlay (see also Bode et al, 24) and is defined as M a ðiþ ¼ 1 n win iþn win=2 1 X j¼i n win =2 eðjþ; 8i 2 Z, (13) with n win 2 N an application specific time frame This filter operation basically represents a low-pass filtering of the error signal e The moving standard deviation filter operation expresses the fading properties of the created image It is defined as vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u 1 iþn win=2 1 X M sd ðiþ ¼t ðeðjþ M a ðiþþ 2 ; 8i 2 Z, (14) n win j¼i n win =2 and as such has high-pass filter properties In the context of wafer scanner performance, Fig 16 shows the result of a robustness study to set-point variation At five different locations on the wafer (four corner points and the center point) eight similar scans at each location are executed and evaluated From these 4 measured time-series, the maximum absolute position error during scanning (denoted by kk 1 ) is depicted for both the short-stroke x- and y-axes, ie, a single value for each axis Nine of such values are obtained by repeating the experiment for three different scan velocities: :15; :3; :6ms 1, and three different die sizes: 8; 16; 32 mm, the latter lecting different position intervals centered about the considered wafer locations The experiments are executed without learning () and with a FIR controller (fir) thereby considering 72 different measured scans In terms of performance, Fig 16 shows that the infinity 6 6 fir e x in nm 3 e y in nm die size in mm 35 7 scan velocity in ms 1 4 die size in mm scan velocity in ms 1 Fig 16 Performance analysis from an unfiltered error viewpoint before (gray) and after (black) learned feed-forward control applied in scanning y-direction at a short stroke wafer stage at three different scan velocities: :15; :3; :6ms 1, and three different die sizes: 8; 16; 32 mm; l ¼ 1 17, d ¼ 3nm,n con ¼ 226, n obs ¼ 3, n ¼ n c ¼ 45 in nm 1 in nm 1 a (e x ) 5 a (e y ) 5 4 die size in mm scan velocity in ms die size in mm 35 7 scan velocity in ms 1 sd (e x ) in nm die size in mm 35 7 scan velocity in ms 1 in nm sd (e y ) 25 fir die size in mm 35 7 scan velocity in ms 1 Fig 17 Performance analysis from an M a- and M sd -filtered error viewpoints before (gray) and after (black) learned feed-forward control applied in scanning y-direction at a short stroke wafer stage at three different scan velocities: :15; :3; :6ms 1, and three different die sizes: 8; 16; 32 mm; l ¼ 1 17, d ¼ 3nm, n con ¼ 226, n obs ¼ 3, n ¼ n c ¼ 45

11 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) norm of the error, which is mainly induced by the settling phenomenon, is significantly reduced under the given MIMO learned feed-forward; the case without learning demonstrates what is obtained under model-based SISO feed-forward control In terms of scan velocity and die size variation, Fig 16 shows limited sensitivity, hence sufficient robustness, along the considered parameter plane Performance in terms of M a and M sd is considered in Fig 17 It shows that small scan velocities give rise to small error levels Apart from the reduced amount of excitation related to such velocities, this is clear from the fact that n win scan velocity 1 Decreasing the scan velocity increases n win which in Eqs (13) and (14) has the effect of averaging out the settling phenomenon Figs 16 and 17 clearly demonstrate the ability of the generalized learned forces to achieve robust performance 6 Conclusions For industrial wafer scanners, learning control provides a powerful means to improve upon settling performance The generalization of the learned forces through FIR modeling adds the necessary robustness to set-point variation which would otherwise severely limit the benefits of learning This is because the wafer scanning industry lacks an exact repetition of set-points and demands a firsttime-right strategy during scanning In an MIMO learned feed-forward context, the occurrence of settling behavior, one of the major servo limitations on wafer throughput, is shown to effectively disappear Contrarily, the FIR approach appears sensitive in the presence of nonlinear system behavior This avoids posing a general design rule regarding the data interval needed to construct the FIR mapping Regarding the number of FIR coefficients, a design argument other than keeping its number small cannot be deduced from the considered crossvalidation experiments The FIR approach shows a good robustness to set-point variation whereas servo performance is not severely compromised when compared to the application of the ILC forces at the set-point for which is learned This is the outcome of a cross validation experiment At the same time, the FIR approach demonstrates its ability to achieve performance in an industrial environment In this regard, the approach shows potential in the broader context of industrial motion control systems Appendix The proof of Theorem 1 is given as follows By adopting the notation a ilc ðk þ n nc Þ¼a n, it follows from (8) that 2 a 2 3 n þþa2 nþo 1 % ::: % a n 1 a n þþa nþo 2 a nþo 1 a 2 A T ilc A n 1 þþa2 nþo 2 ::: % ilc ¼, a 1 a n þþa oa nþo 1 a 1 a n 1 þþa oa nþo 2 ::: a 2 1 þ ::: þ a2 o (15) where % indicates the symmetric counterparts Invertibility of (15) implies that all columns are independent Consider the first and the second column, or 2 a 2 3 n þþa2 nþo 1 a n 1 a n þþa nþo 2 a nþo 1 a n 1 a n þþa nþo 2 a nþo 1 a 2 n 1 þ ::: þ 6 a2 nþo (16) In view of symmetry, both columns are independent if the two diagonal terms differ In fact, this holds true for each consecutive pair of columns, ie, the second and the third, the third and the fourth, and so on If both diagonal terms are equal, or a 2 n þþa2 nþo 1 ¼ a2 n 1 þþa2 nþo 2 ) a2 n 1 ¼ a2 nþo 1, (17) independency results from the fact that the symmetric offdiagonal terms differ from the diagonal terms Namely assume all four terms in (16) are equal, or a 2 n þþa2 nþo 1 ¼ a n 1a n þþa nþo 2 a nþo 1 ¼ 1 2 a2 n þþ1 2 a2 nþo 1 þ 1 2 a2 n 1 þþ1 2 a2 nþo 2 Then (18) combined with (17) gives 1 2 ða n a n 1 Þ ða nþo 1 a nþo 2 Þ 2 (18) 1 2 a2 n 1 þ 1 2 a2 nþo 1 ¼ 1 2 ða n a n 1 Þ ða nþo 1 a nþo 2 Þ 2 ¼, (19) or a n a n 1 ¼¼a nþo 1 a nþo 2 ¼, which under the assumption that a ilc ðk þ n nc Þ¼a n a, reduces to a n 1 ¼ a n ¼¼ a nþo 1 a But this contradicts the earlier assumption that A ilc contains no columns with o equal entries Hence the symmetric off-diagonal terms differ from the diagonal terms rendering the columns in (16) independent The remainder of the proof follows from repeating the previous steps for each consecutive pair of columns References Ahn, H-S, Moore, K L, & Chen, Y Q (26) Monotonic convergent iterative learning controller design based on interval model conversion IEEE Transactions on Automatic Control, 51(2), Baggen, M, Heertjes, M F, & Kamidi, R (28) Data-based feedforward control in MIMO motion systems In Proceedings of the American control conference, Seattle, WA (pp ) Bode, C A, Ko, B S, & Edgar, T F (24) Run-to-run control and performance monitoring of overlay in semiconductor manufacturing Control Engineering Practice, 12, Bristow, D A, Tharayil, M, & Alleyne, A G (26) A survey of iterative learning control A learning-based method for high-performance tracking IEEE Control Systems Magazine, Cai, Z, Freeman, C T, Lewin, P L, & Rogers, E (28) Iterative learning control for a non-minimum phase plant based on a erence shift algorithm Control Engineering Practice, 16, Chen, C K, & Hwang, J (25) Iterative learning control for position tracking of a pneumatic actuated X Y table Control Engineering Practice, 13(12), Chen, Y Q, Moore, K L, Yu, J, & Zhang, T (28) Iterative learning control and repetitive control in hard disk drive industry A tutorial International Journal of Adaptive Control and Signal Processing, 22(4), Dijkstra, B G, & Bosgra, O H (22) Extrapolation of optimal lifted system ILC solution, with application to a waferstage In Proceedings of the American control conference, Anchorage, AK (pp ) Dixon, W E, & Chen, J (23) Comments on A composite energy function-based learning control approach for nonlinear systems with time-varying parametric uncertainties IEEE Transactions on Automatic Control, 48(9), Ghosh, J, & Paden, B (22) A pseudoinverse-based iterative learning control IEEE Transactions on Automatic Control, 47(5), Groot-Wassink, M, Van de Wal, M, Scherer, C, & Bosgra, O (25) LPV control for a wafer stage: Beyond the theoretical solution Control Engineering Practice, 13, Gunnarsson, S, & Norrlöf, M (21) On the design of ILC algorithms using optimization Automatica, 37, Heertjes, M F, & Tso, T (27a) Nonlinear iterative learning control with applications to lithographic machinery Control Engineering Practice, 15, Heertjes, M F, & Tso, T (27b) Robustness, convergence, and Lyapunov stability of a nonlinear iterative learning control aplied at a wafer scanner In Proceedings of the American control conference, New York, WA (pp ) Heertjes, M F, & Van de Wouw, N (26) Variable control design and its application to wafer scanners In Proceedings of conference on decision and control, San Diego, CA (pp ) Mishra, S, Coaplen, J, & Tomizuka, M (27) Precision positioning of wafer scanners; segmented iterative learning control for nonrepetitive disturbances IEEE Control Systems Magazine, August, 2 25 Moore, K L (1999) An iterative learning control algorithm for systems with measurement noise In Proceedings of the conference on decision and control, Phoenix, AZ (pp )

12 356 MF Heertjes, RMJG van de Molengraft / Control Engineering Practice 17 (29) Moore, K L, Chen, Y Q, & Bahl, V (25) Monotonically convergent iterative learning control for linear discrete-time systems Automatica, 41, Potsaid, B, & Wen, J (24) High performance motion tracking control In Proceedings of the 24 IEEE international conference on control applications, Taipei, Taiwan (pp ) Rotariu, I, Dijkstra, B G, & Steinbuch, M (24) Comparison of standard and lifted ILC on a motion system In Proceedings of the third IFAC symposium on mechatronic systems, Sydney, Australia Rotariu, I, Ellenbroek, R, & Steinbuch, M (23) Time frequency analysis of a motion system with learning control In IEEE Proceedings of the American control conference, Denver, CO (pp ) Rotariu, I, Ellenbroek, R, Van Baars, G, & Steinbuch, M (23) Iterative learning control for variable setpoints, applied to a motion system In Proceedings of the European control conference, Cambridge, UK (pp ) Tayebi, A, & Islam, S (26) Adaptive iterative learning control for robot manipulators: Experimental results Control Engineering Practice, 14(7), Tousain, R L, & Van der Meché, E (21) Design strategies for iterative learning control based on optimal control In Proceedings of the 4th conference on decision and control, Orlando, FL (pp ) Van de Wal, M, Van Baars, G, Sperling, F, & Bosgra, O (22) Multivariable H 1=m feedback control design for high-precision wafer stage motion Control Engineering Practice, 1, Van der Meulen, S H, Tousain, R L, & Bosgra, O H (28) Fixed structure feedforward controller design exploiting iterative trials: Application to a wafer stage and a desktop printer Journal of Dynamic Systems, Measurement, and Control, 13(516), 1 16 Xu, J-X (1998) Direct learning of control efforts for trajectories with different time scales IEEE Transactions on Automatic Control, 43(7), Xu, J-X, & Tan, Y (22) A composite energy function-based learning control approach for nonlinear systems with time-varying parametric uncertainties IEEE Transactions on Automatic Control, 47(11), Xu, J-X, & Tan, Y (23) Author s reply IEEE Transactions on Automatic Control, 48(9),

HYBRID CONTROL FOR MOTION SYSTEMS WITH IMPROVED DISTURBANCE REJECTION

HYBRID CONTROL FOR MOTION SYSTEMS WITH IMPROVED DISTURBANCE REJECTION ENOC-8, Saint Petersburg, Russia, 3- June/July 8 HYBRID CONTROL FOR MOTION SYSTEMS WITH IMPROVED DISTURBANCE REJECTION Marcel Heertjes ASML Mechatronic Systems Development 56 MD Veldhoven, The Netherlands

More information

Jerk derivative feedforward control for motion systems

Jerk derivative feedforward control for motion systems Jerk derivative feedforward control for motion systems Matthijs Boerlage Rob Tousain Maarten Steinbuch Abstract This work discusses reference trajectory relevant model based feedforward design. For motion

More information

Circle Criterion in Linear Control Design

Circle Criterion in Linear Control Design 8 American Control Conference Westin Seattle Hotel Seattle Washington USA June -3 8 ThC8. Circle Criterion in Linear Control Design Marcel Heertjes and Maarten Steinbuch Abstract This paper presents a

More information

magnitude [db] phase [deg] frequency [Hz] feedforward motor load -

magnitude [db] phase [deg] frequency [Hz] feedforward motor load - ITERATIVE LEARNING CONTROL OF INDUSTRIAL MOTION SYSTEMS Maarten Steinbuch and René van de Molengraft Eindhoven University of Technology, Faculty of Mechanical Engineering, Systems and Control Group, P.O.

More information

Data-Based Feed-Forward Control in MIMO Motion Systems

Data-Based Feed-Forward Control in MIMO Motion Systems 28 American Control Conference Westin Seattle Hotel, Seattle, Washington, USA June 11-13, 28 ThB17.4 Data-Based Feed-Forward Control in MIMO Motion Sstems Mark Baggen, Marcel Heertjes and Ramidin Kamidi

More information

Multi-variable iterative tuning of a variable gain controller with application to a scanning stage system

Multi-variable iterative tuning of a variable gain controller with application to a scanning stage system 2 American Control Conference on O'Farrell Street, San Francisco, CA, USA June 29 - July, 2 Multi-variable iterative tuning of a variable gain controller with application to a scanning stage system Marcel

More information

ALMOST a century after the pioneering work of authors

ALMOST a century after the pioneering work of authors IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 56, NO. 5, MAY 2009 1347 Performance-Improved Design of N-PID Controlled Motion Systems With Applications to Wafer Stages Marcel François Heertjes, Xander

More information

SWITCHING CONTROL IN ACTIVE VIBRATION ISOLATION

SWITCHING CONTROL IN ACTIVE VIBRATION ISOLATION ENOC-28, Saint Petersburg, Russia, 3-4 June/July 28 SWITCHING CONTROL IN ACTIVE VIBRATION ISOLATION M.F. Heertjes Eindhoven University of Technology Department of Mechanical Engineering 56 MB Eindhoven,

More information

An Iteration-Domain Filter for Controlling Transient Growth in Iterative Learning Control

An Iteration-Domain Filter for Controlling Transient Growth in Iterative Learning Control 21 American Control Conference Marriott Waterfront, Baltimore, MD, USA June 3-July 2, 21 WeC14.1 An Iteration-Domain Filter for Controlling Transient Growth in Iterative Learning Control Qing Liu and Douglas

More information

Available online at ScienceDirect. IFAC-PapersOnLine (2016)

Available online at   ScienceDirect. IFAC-PapersOnLine (2016) Available online at www.sciencedirect.com ScienceDirect IFAC-PapersOnLine 49-13 (16) 93 98 Experimental Evaluation of Reset Control for Improved Stage Performance M.F. Heertjes, K.G.J. Gruntjens S.J.L.M.

More information

Iterative Learning Control Analysis and Design I

Iterative Learning Control Analysis and Design I Iterative Learning Control Analysis and Design I Electronics and Computer Science University of Southampton Southampton, SO17 1BJ, UK etar@ecs.soton.ac.uk http://www.ecs.soton.ac.uk/ Contents Basics Representations

More information

A Discrete Robust Adaptive Iterative Learning Control for a Class of Nonlinear Systems with Unknown Control Direction

A Discrete Robust Adaptive Iterative Learning Control for a Class of Nonlinear Systems with Unknown Control Direction Proceedings of the International MultiConference of Engineers and Computer Scientists 16 Vol I, IMECS 16, March 16-18, 16, Hong Kong A Discrete Robust Adaptive Iterative Learning Control for a Class of

More information

Design strategies for iterative learning control based on optimal control

Design strategies for iterative learning control based on optimal control Selected Topics in Signals, Systems and Control Vol. 2, September 2 Design strategies for iterative learning control based on optimal control Rob Tousain, Eduard van der Meché and Okko Bosgra Mechanical

More information

High Precision Control of Ball Screw Driven Stage Using Repetitive Control with Sharp Roll-off Learning Filter

High Precision Control of Ball Screw Driven Stage Using Repetitive Control with Sharp Roll-off Learning Filter High Precision Control of Ball Screw Driven Stage Using Repetitive Control with Sharp Roll-off Learning Filter Tadashi Takemura and Hiroshi Fujimoto The University of Tokyo --, Kashiwanoha, Kashiwa, Chiba,

More information

An Adaptive LQG Combined With the MRAS Based LFFC for Motion Control Systems

An Adaptive LQG Combined With the MRAS Based LFFC for Motion Control Systems Journal of Automation Control Engineering Vol 3 No 2 April 2015 An Adaptive LQG Combined With the MRAS Based LFFC for Motion Control Systems Nguyen Duy Cuong Nguyen Van Lanh Gia Thi Dinh Electronics Faculty

More information

ThM06-2. Coprime Factor Based Closed-Loop Model Validation Applied to a Flexible Structure

ThM06-2. Coprime Factor Based Closed-Loop Model Validation Applied to a Flexible Structure Proceedings of the 42nd IEEE Conference on Decision and Control Maui, Hawaii USA, December 2003 ThM06-2 Coprime Factor Based Closed-Loop Model Validation Applied to a Flexible Structure Marianne Crowder

More information

Variable gain motion control for transient performance improvement

Variable gain motion control for transient performance improvement 22 American Control Conference Fairmont Queen Elizabeth, Montréal, Canada June 27-June 29, 22 Variable gain motion control for transient performance improvement B.G.B. Hunnekens, N. van de Wouw and H.

More information

Steady-state performance optimization for variable-gain motion control using extremum seeking*

Steady-state performance optimization for variable-gain motion control using extremum seeking* 51st IEEE Conference on Decision and Control December 1-13, 1. Maui, Hawaii, USA Steady-state performance optimization for variable-gain motion control using extremum seeking* B.G.B. Hunnekens 1, M.A.M.

More information

Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control

Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control *Manuscript Click here to view linked References Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control Robbert van Herpen a, Tom Oomen a, Edward Kikken a, Marc van de Wal

More information

Nonlinear Adaptive Robust Control. Theory and Applications to the Integrated Design of Intelligent and Precision Mechatronic Systems.

Nonlinear Adaptive Robust Control. Theory and Applications to the Integrated Design of Intelligent and Precision Mechatronic Systems. A Short Course on Nonlinear Adaptive Robust Control Theory and Applications to the Integrated Design of Intelligent and Precision Mechatronic Systems Bin Yao Intelligent and Precision Control Laboratory

More information

Predictive Iterative Learning Control using Laguerre Functions

Predictive Iterative Learning Control using Laguerre Functions Milano (Italy) August 28 - September 2, 211 Predictive Iterative Learning Control using Laguerre Functions Liuping Wang Eric Rogers School of Electrical and Computer Engineering, RMIT University, Victoria

More information

AFAULT diagnosis procedure is typically divided into three

AFAULT diagnosis procedure is typically divided into three 576 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 47, NO. 4, APRIL 2002 A Robust Detection and Isolation Scheme for Abrupt and Incipient Faults in Nonlinear Systems Xiaodong Zhang, Marios M. Polycarpou,

More information

PERIODIC signals are commonly experienced in industrial

PERIODIC signals are commonly experienced in industrial IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 15, NO. 2, MARCH 2007 369 Repetitive Learning Control of Nonlinear Continuous-Time Systems Using Quasi-Sliding Mode Xiao-Dong Li, Tommy W. S. Chow,

More information

Performing Aggressive Maneuvers using Iterative Learning Control

Performing Aggressive Maneuvers using Iterative Learning Control 2009 IEEE International Conference on Robotics and Automation Kobe International Conference Center Kobe, Japan, May 12-17, 2009 Performing Aggressive Maneuvers using Iterative Learning Control Oliver Purwin

More information

Batch-to-Batch Rational Feedforward Control: from Iterative Learning to Identification Approaches, with Application to a Wafer Stage

Batch-to-Batch Rational Feedforward Control: from Iterative Learning to Identification Approaches, with Application to a Wafer Stage Batch-to-Batch Rational Feedforward Control: from Iterative Learning to Identification Approaches with Application to a Wafer Stage Lennart Blanken Frank Boeren Dennis Bruijnen Tom Oomen Abstract Feedforward

More information

On the Current Error Based Sampled-data Iterative Learning Control with Reduced Memory Capacity

On the Current Error Based Sampled-data Iterative Learning Control with Reduced Memory Capacity International Journal of Automation and Computing 12(3), June 2015, 307-315 DOI: 101007/s11633-015-0890-1 On the Current Error Based Sampled-data Iterative Learning Control with Reduced Memory Capacity

More information

Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control

Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control 204 American Control Conference (ACC) June 4-6, 204. Portland, Oregon, USA Exploiting Additional Actuators and Sensors for Nano-Positioning Robust Motion Control Robbert van Herpen, Tom Oomen, Edward Kikken,

More information

Part 1: Introduction to the Algebraic Approach to ILC

Part 1: Introduction to the Algebraic Approach to ILC IEEE ICMA 2006 Tutorial Workshop: Control Algebraic Analysis and Optimal Design Presenters: Contributor: Kevin L. Moore Colorado School of Mines YangQuan Chen Utah State University Hyo-Sung Ahn ETRI, Korea

More information

A 2D Systems Approach to Iterative Learning Control with Experimental Validation

A 2D Systems Approach to Iterative Learning Control with Experimental Validation Proceedings of the 17th World Congress The International Federation of Automatic Control Seoul, Korea, July 6-11, 28 A 2D Systems Approach to Iterative Learning Control with Experimental Validation Lukasz

More information

EVALUATION OF (UNSTABLE) NON-CAUSAL SYSTEMS APPLIED TO ITERATIVE LEARNING CONTROL

EVALUATION OF (UNSTABLE) NON-CAUSAL SYSTEMS APPLIED TO ITERATIVE LEARNING CONTROL EVALUATION OF (UNSTABLE) NON-CAUSAL SYSTEMS APPLIED TO ITERATIVE LEARNING CONTROL M.G.E. Schneiders, M.J.G. van de Molengraft and M. Steinbuch Eindhoven University of Technology, Control Systems Technology,

More information

Adaptive iterative learning control for robot manipulators: Experimental results $

Adaptive iterative learning control for robot manipulators: Experimental results $ Control Engineering Practice 4 (26) 843 85 www.elsevier.com/locate/conengprac Adaptive iterative learning control for robot manipulators: Experimental results $ A. Tayebi a,, S. Islam b, a Department of

More information

Feedback Control of Linear SISO systems. Process Dynamics and Control

Feedback Control of Linear SISO systems. Process Dynamics and Control Feedback Control of Linear SISO systems Process Dynamics and Control 1 Open-Loop Process The study of dynamics was limited to open-loop systems Observe process behavior as a result of specific input signals

More information

Simple Learning Control Made Practical by Zero-Phase Filtering: Applications to Robotics

Simple Learning Control Made Practical by Zero-Phase Filtering: Applications to Robotics IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I: FUNDAMENTAL THEORY AND APPLICATIONS, VOL 49, NO 6, JUNE 2002 753 Simple Learning Control Made Practical by Zero-Phase Filtering: Applications to Robotics Haluk

More information

LEARNING control is used in many motion systems.

LEARNING control is used in many motion systems. 1 Rational asis Functions in Iterative Learning Control - With Experimental Verification on a Motion System Joost older and Tom Oomen Abstract Iterative Learning Control (ILC) approaches often exhibit

More information

Rejection of fixed direction disturbances in multivariable electromechanical motion systems

Rejection of fixed direction disturbances in multivariable electromechanical motion systems Rejection of fixed direction disturbances in multivariable electromechanical motion systems Matthijs Boerlage Rick Middleton Maarten Steinbuch, Bram de Jager Technische Universiteit Eindhoven, Eindhoven,

More information

Iterative Feedforward Control: A Closed-loop Identification Problem and a Solution

Iterative Feedforward Control: A Closed-loop Identification Problem and a Solution 52nd IEEE Conference on Decision and Control December 10-13, 2013 Florence, Italy Iterative Feedforward Control: A Closed-loop Identification Problem and a Solution Frank Boeren and Tom Oomen Abstract

More information

Rational basis functions in iterative learning control - With experimental verification on a motion system Bolder, J.J.

Rational basis functions in iterative learning control - With experimental verification on a motion system Bolder, J.J. Rational basis functions in iterative learning control - With experimental verification on a motion system Bolder, J.J.; Oomen, Tom Published in: IEEE Transactions on Control Systems Technology DOI: 10.1109/TCST.2014.2327578

More information

Lifted approach to ILC/Repetitive Control

Lifted approach to ILC/Repetitive Control Lifted approach to ILC/Repetitive Control Okko H. Bosgra Maarten Steinbuch TUD Delft Centre for Systems and Control TU/e Control System Technology Dutch Institute of Systems and Control DISC winter semester

More information

Filtered-X LMS vs repetitive control for active structural acoustic control of periodic disturbances

Filtered-X LMS vs repetitive control for active structural acoustic control of periodic disturbances Filtered-X LMS vs repetitive control for active structural acoustic control of periodic disturbances B. Stallaert 1, G. Pinte 2, S. Devos 2, W. Symens 2, J. Swevers 1, P. Sas 1 1 K.U.Leuven, Department

More information

Nonlinear PD Controllers with Gravity Compensation for Robot Manipulators

Nonlinear PD Controllers with Gravity Compensation for Robot Manipulators BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 4, No Sofia 04 Print ISSN: 3-970; Online ISSN: 34-408 DOI: 0.478/cait-04-00 Nonlinear PD Controllers with Gravity Compensation

More information

Design of Decentralised PI Controller using Model Reference Adaptive Control for Quadruple Tank Process

Design of Decentralised PI Controller using Model Reference Adaptive Control for Quadruple Tank Process Design of Decentralised PI Controller using Model Reference Adaptive Control for Quadruple Tank Process D.Angeline Vijula #, Dr.N.Devarajan * # Electronics and Instrumentation Engineering Sri Ramakrishna

More information

Iterative Learning Control (ILC)

Iterative Learning Control (ILC) Department of Automatic Control LTH, Lund University ILC ILC - the main idea Time Domain ILC approaches Stability Analysis Example: The Milk Race Frequency Domain ILC Example: Marine Vibrator Material:

More information

MULTILOOP PI CONTROLLER FOR ACHIEVING SIMULTANEOUS TIME AND FREQUENCY DOMAIN SPECIFICATIONS

MULTILOOP PI CONTROLLER FOR ACHIEVING SIMULTANEOUS TIME AND FREQUENCY DOMAIN SPECIFICATIONS Journal of Engineering Science and Technology Vol. 1, No. 8 (215) 113-1115 School of Engineering, Taylor s University MULTILOOP PI CONTROLLER FOR ACHIEVING SIMULTANEOUS TIME AND FREQUENCY DOMAIN SPECIFICATIONS

More information

Trajectory planning and feedforward design for electromechanical motion systems version 2

Trajectory planning and feedforward design for electromechanical motion systems version 2 2 Trajectory planning and feedforward design for electromechanical motion systems version 2 Report nr. DCT 2003-8 Paul Lambrechts Email: P.F.Lambrechts@tue.nl April, 2003 Abstract This report considers

More information

Optimal algorithm and application for point to point iterative learning control via updating reference trajectory

Optimal algorithm and application for point to point iterative learning control via updating reference trajectory 33 9 2016 9 DOI: 10.7641/CTA.2016.50970 Control Theory & Applications Vol. 33 No. 9 Sep. 2016,, (, 214122) :,.,,.,,,.. : ; ; ; ; : TP273 : A Optimal algorithm and application for point to point iterative

More information

Simulation Study on Pressure Control using Nonlinear Input/Output Linearization Method and Classical PID Approach

Simulation Study on Pressure Control using Nonlinear Input/Output Linearization Method and Classical PID Approach Simulation Study on Pressure Control using Nonlinear Input/Output Linearization Method and Classical PID Approach Ufuk Bakirdogen*, Matthias Liermann** *Institute for Fluid Power Drives and Controls (IFAS),

More information

1 An Overview and Brief History of Feedback Control 1. 2 Dynamic Models 23. Contents. Preface. xiii

1 An Overview and Brief History of Feedback Control 1. 2 Dynamic Models 23. Contents. Preface. xiii Contents 1 An Overview and Brief History of Feedback Control 1 A Perspective on Feedback Control 1 Chapter Overview 2 1.1 A Simple Feedback System 3 1.2 A First Analysis of Feedback 6 1.3 Feedback System

More information

Survey of Methods of Combining Velocity Profiles with Position control

Survey of Methods of Combining Velocity Profiles with Position control Survey of Methods of Combining Profiles with control Petter Karlsson Mälardalen University P.O. Box 883 713 Västerås, Sweden pkn91@student.mdh.se ABSTRACT In many applications where some kind of motion

More information

Lyapunov Stability of Linear Predictor Feedback for Distributed Input Delays

Lyapunov Stability of Linear Predictor Feedback for Distributed Input Delays IEEE TRANSACTIONS ON AUTOMATIC CONTROL VOL. 56 NO. 3 MARCH 2011 655 Lyapunov Stability of Linear Predictor Feedback for Distributed Input Delays Nikolaos Bekiaris-Liberis Miroslav Krstic In this case system

More information

Fig. 1. Schematic illustration of a wafer scanner system, where ➀ light source, ➁ reticle, ➂ reticle stage, ➃ lens, ➄ wafer, and ➅ wafer stage.

Fig. 1. Schematic illustration of a wafer scanner system, where ➀ light source, ➁ reticle, ➂ reticle stage, ➃ lens, ➄ wafer, and ➅ wafer stage. 102 IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 22, NO. 1, JANUARY 2014 Connecting System Identification and Robust Control for Next-Generation Motion Control of a Wafer Stage Tom Oomen, Robbert

More information

THE paper deals with the application of ILC-methods to

THE paper deals with the application of ILC-methods to Application of Fourier Series Based Learning Control on Mechatronic Systems Sandra Baßler, Peter Dünow, Mathias Marquardt International Science Index, Mechanical and Mechatronics Engineering waset.org/publication/10005018

More information

Limit Cycles in High-Resolution Quantized Feedback Systems

Limit Cycles in High-Resolution Quantized Feedback Systems Limit Cycles in High-Resolution Quantized Feedback Systems Li Hong Idris Lim School of Engineering University of Glasgow Glasgow, United Kingdom LiHonIdris.Lim@glasgow.ac.uk Ai Poh Loh Department of Electrical

More information

FROM the early introduction of CD applications in the

FROM the early introduction of CD applications in the IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 14, NO. 3, MAY 2006 389 Experimental Frequency-Domain Analysis of Nonlinear Controlled Optical Storage Drives Marcel Heertjes, Erik Pastink, Nathan

More information

Iterative Controller Tuning Using Bode s Integrals

Iterative Controller Tuning Using Bode s Integrals Iterative Controller Tuning Using Bode s Integrals A. Karimi, D. Garcia and R. Longchamp Laboratoire d automatique, École Polytechnique Fédérale de Lausanne (EPFL), 05 Lausanne, Switzerland. email: alireza.karimi@epfl.ch

More information

D(s) G(s) A control system design definition

D(s) G(s) A control system design definition R E Compensation D(s) U Plant G(s) Y Figure 7. A control system design definition x x x 2 x 2 U 2 s s 7 2 Y Figure 7.2 A block diagram representing Eq. (7.) in control form z U 2 s z Y 4 z 2 s z 2 3 Figure

More information

General procedure for formulation of robot dynamics STEP 1 STEP 3. Module 9 : Robot Dynamics & controls

General procedure for formulation of robot dynamics STEP 1 STEP 3. Module 9 : Robot Dynamics & controls Module 9 : Robot Dynamics & controls Lecture 32 : General procedure for dynamics equation forming and introduction to control Objectives In this course you will learn the following Lagrangian Formulation

More information

Application of Neuro Fuzzy Reduced Order Observer in Magnetic Bearing Systems

Application of Neuro Fuzzy Reduced Order Observer in Magnetic Bearing Systems Application of Neuro Fuzzy Reduced Order Observer in Magnetic Bearing Systems M. A., Eltantawie, Member, IAENG Abstract Adaptive Neuro-Fuzzy Inference System (ANFIS) is used to design fuzzy reduced order

More information

Learning Model Predictive Control for Iterative Tasks: A Computationally Efficient Approach for Linear System

Learning Model Predictive Control for Iterative Tasks: A Computationally Efficient Approach for Linear System Learning Model Predictive Control for Iterative Tasks: A Computationally Efficient Approach for Linear System Ugo Rosolia Francesco Borrelli University of California at Berkeley, Berkeley, CA 94701, USA

More information

Adaptive Rejection of Periodic Disturbances Acting on Linear Systems with Unknown Dynamics

Adaptive Rejection of Periodic Disturbances Acting on Linear Systems with Unknown Dynamics 206 IEEE 55th Conference on Decision and Control (CDC) ARIA Resort & Casino December 2-4, 206, Las Vegas, USA Adaptive Rejection of Periodic Disturbances Acting on Linear Systems with Unknown Dynamics

More information

Position and Velocity Profile Tracking Control for New Generation Servo Track Writing

Position and Velocity Profile Tracking Control for New Generation Servo Track Writing Preprints of the 9th World Congress The International Federation of Automatic Control Cape Town, South Africa. August 24-29, 24 Position and Velocity Profile Tracking Control for New Generation Servo Track

More information

Chapter 7 Interconnected Systems and Feedback: Well-Posedness, Stability, and Performance 7. Introduction Feedback control is a powerful approach to o

Chapter 7 Interconnected Systems and Feedback: Well-Posedness, Stability, and Performance 7. Introduction Feedback control is a powerful approach to o Lectures on Dynamic Systems and Control Mohammed Dahleh Munther A. Dahleh George Verghese Department of Electrical Engineering and Computer Science Massachuasetts Institute of Technology c Chapter 7 Interconnected

More information

Recent Advances in Positive Systems: The Servomechanism Problem

Recent Advances in Positive Systems: The Servomechanism Problem Recent Advances in Positive Systems: The Servomechanism Problem 47 th IEEE Conference on Decision and Control December 28. Bartek Roszak and Edward J. Davison Systems Control Group, University of Toronto

More information

vehicle velocity (m/s) relative velocity (m/s) 22 relative velocity (m/s) 1.5 vehicle velocity (m/s) time (s)

vehicle velocity (m/s) relative velocity (m/s) 22 relative velocity (m/s) 1.5 vehicle velocity (m/s) time (s) Proceedings of the 4th IEEE Conference on Decision and Control, New Orleans, LA, December 99, pp. 477{48. Variable Time Headway for String Stability of Automated HeavyDuty Vehicles Diana Yanakiev and Ioannis

More information

An Iterative Learning Controller for Reduction of Repeatable Runout in Hard Disk Drives

An Iterative Learning Controller for Reduction of Repeatable Runout in Hard Disk Drives Proceedings of the 26 American Control Conference Minneapolis, Minnesota, USA, June 14-16, 26 WeA18.1 An Iterative Learning Controller for Reduction of Repeatable Runout in Hard Disk Drives M.R. Graham

More information

CBE495 LECTURE IV MODEL PREDICTIVE CONTROL

CBE495 LECTURE IV MODEL PREDICTIVE CONTROL What is Model Predictive Control (MPC)? CBE495 LECTURE IV MODEL PREDICTIVE CONTROL Professor Dae Ryook Yang Fall 2013 Dept. of Chemical and Biological Engineering Korea University * Some parts are from

More information

Control for. Maarten Steinbuch Dept. Mechanical Engineering Control Systems Technology Group TU/e

Control for. Maarten Steinbuch Dept. Mechanical Engineering Control Systems Technology Group TU/e Control for Maarten Steinbuch Dept. Mechanical Engineering Control Systems Technology Group TU/e Motion Systems m F Introduction Timedomain tuning Frequency domain & stability Filters Feedforward Servo-oriented

More information

Recurrent neural networks with trainable amplitude of activation functions

Recurrent neural networks with trainable amplitude of activation functions Neural Networks 16 (2003) 1095 1100 www.elsevier.com/locate/neunet Neural Networks letter Recurrent neural networks with trainable amplitude of activation functions Su Lee Goh*, Danilo P. Mandic Imperial

More information

Disturbance Compensation for DC Motor Mechanism Low Speed Regulation : A Feedforward and Feedback Implementation

Disturbance Compensation for DC Motor Mechanism Low Speed Regulation : A Feedforward and Feedback Implementation 211 5th IEEE Conference on Decision and Control and European Control Conference (CDC-ECC) Orlando, FL, USA, December 12-15, 211 Disturbance Compensation for DC Motor Mechanism Low Speed Regulation : A

More information

Decoupled Feedforward Control for an Air-Conditioning and Refrigeration System

Decoupled Feedforward Control for an Air-Conditioning and Refrigeration System American Control Conference Marriott Waterfront, Baltimore, MD, USA June 3-July, FrB1.4 Decoupled Feedforward Control for an Air-Conditioning and Refrigeration System Neera Jain, Member, IEEE, Richard

More information

Analysis of Discrete-Time Systems

Analysis of Discrete-Time Systems TU Berlin Discrete-Time Control Systems 1 Analysis of Discrete-Time Systems Overview Stability Sensitivity and Robustness Controllability, Reachability, Observability, and Detectabiliy TU Berlin Discrete-Time

More information

Enhancing Transient Response of Asymptotic Regulation with Disturbance Onset

Enhancing Transient Response of Asymptotic Regulation with Disturbance Onset 211 American Control Conference on O'Farrell Street, San Francisco, CA, USA June 29 - July 1, 211 Enhancing Transient Response of Asymptotic Regulation with Disturbance Onset Kevin C. Chu and Tsu-Chin

More information

PID continues to be popular in the industry due to its wide

PID continues to be popular in the industry due to its wide Constant in gain Lead in phase element - Application in precision motion control Niranjan Saikumar, Rahul Kumar Sinha, S. Hassan HosseinNia Precision and Microsystems Engineering, Faculty of Mechanical

More information

Iterative Control for Periodic Tasks with Robustness Considerations, Applied to a Nanopositioning Stage

Iterative Control for Periodic Tasks with Robustness Considerations, Applied to a Nanopositioning Stage Proceedings of the 7th IFAC Symposium on Mechatronic Systems, Loughborough University, UK, September 5-8, 216 ThP6T11 Iterative Control for Periodic Tasks with Robustness Considerations, Applied to a anopositioning

More information

Batch-to-batch strategies for cooling crystallization

Batch-to-batch strategies for cooling crystallization Batch-to-batch strategies for cooling crystallization Marco Forgione 1, Ali Mesbah 1, Xavier Bombois 1, Paul Van den Hof 2 1 Delft University of echnology Delft Center for Systems and Control 2 Eindhoven

More information

Design and Tuning of Fractional-order PID Controllers for Time-delayed Processes

Design and Tuning of Fractional-order PID Controllers for Time-delayed Processes Design and Tuning of Fractional-order PID Controllers for Time-delayed Processes Emmanuel Edet Technology and Innovation Centre University of Strathclyde 99 George Street Glasgow, United Kingdom emmanuel.edet@strath.ac.uk

More information

REPETITIVE LEARNING OF BACKSTEPPING CONTROLLED NONLINEAR ELECTROHYDRAULIC MATERIAL TESTING SYSTEM 1. Seunghyeokk James Lee 2, Tsu-Chin Tsao

REPETITIVE LEARNING OF BACKSTEPPING CONTROLLED NONLINEAR ELECTROHYDRAULIC MATERIAL TESTING SYSTEM 1. Seunghyeokk James Lee 2, Tsu-Chin Tsao REPETITIVE LEARNING OF BACKSTEPPING CONTROLLED NONLINEAR ELECTROHYDRAULIC MATERIAL TESTING SYSTEM Seunghyeokk James Lee, Tsu-Chin Tsao Mechanical and Aerospace Engineering Department University of California

More information

Control of MIMO processes. 1. Introduction. Control of MIMO processes. Control of Multiple-Input, Multiple Output (MIMO) Processes

Control of MIMO processes. 1. Introduction. Control of MIMO processes. Control of Multiple-Input, Multiple Output (MIMO) Processes Control of MIMO processes Control of Multiple-Input, Multiple Output (MIMO) Processes Statistical Process Control Feedforward and ratio control Cascade control Split range and selective control Control

More information

Phase correction for Learning Feedforward Control

Phase correction for Learning Feedforward Control hase correction for Learning Feedforward Control Bas J. de Kruif and Theo J. A. de Vries Drebbel Institute of Mechatronics, University of Twente, The Netherlands. email:b.j.dekruif@utwente.nl ABSTRACT

More information

Iterative Learning Control for Tailor Rolled Blanks Zhengyang Lu DCT

Iterative Learning Control for Tailor Rolled Blanks Zhengyang Lu DCT Iterative Learning Control for Tailor Rolled Blans Zhengyang Lu DCT 2007.066 Master Internship Report Supervisors: Prof. Maarten Steinbuch (TU/e) Camile Hol (Corus) Eindhoven University of Technology Department

More information

Robust control for a multi-stage evaporation plant in the presence of uncertainties

Robust control for a multi-stage evaporation plant in the presence of uncertainties Preprint 11th IFAC Symposium on Dynamics and Control of Process Systems including Biosystems June 6-8 16. NTNU Trondheim Norway Robust control for a multi-stage evaporation plant in the presence of uncertainties

More information

Control of industrial robots. Centralized control

Control of industrial robots. Centralized control Control of industrial robots Centralized control Prof. Paolo Rocco (paolo.rocco@polimi.it) Politecnico di Milano ipartimento di Elettronica, Informazione e Bioingegneria Introduction Centralized control

More information

Iterative Feedback Tuning

Iterative Feedback Tuning Iterative Feedback Tuning Michel Gevers CESAME - UCL Louvain-la-Neuve Belgium Collaboration : H. Hjalmarsson, S. Gunnarsson, O. Lequin, E. Bosmans, L. Triest, M. Mossberg Outline Problem formulation Iterative

More information

arxiv: v2 [cs.ro] 26 Sep 2016

arxiv: v2 [cs.ro] 26 Sep 2016 Distributed Iterative Learning Control for a Team of Quadrotors Andreas Hock and Angela P Schoellig arxiv:1635933v [csro] 6 Sep 16 Abstract The goal of this work is to enable a team of quadrotors to learn

More information

APPROXIMATE REALIZATION OF VALVE DYNAMICS WITH TIME DELAY

APPROXIMATE REALIZATION OF VALVE DYNAMICS WITH TIME DELAY APPROXIMATE REALIZATION OF VALVE DYNAMICS WITH TIME DELAY Jan van Helvoirt,,1 Okko Bosgra, Bram de Jager Maarten Steinbuch Control Systems Technology Group, Mechanical Engineering Department, Technische

More information

Observer Based Friction Cancellation in Mechanical Systems

Observer Based Friction Cancellation in Mechanical Systems 2014 14th International Conference on Control, Automation and Systems (ICCAS 2014) Oct. 22 25, 2014 in KINTEX, Gyeonggi-do, Korea Observer Based Friction Cancellation in Mechanical Systems Caner Odabaş

More information

Flexible ILC: Towards a Convex Approach for Non-Causal Rational Basis Functions

Flexible ILC: Towards a Convex Approach for Non-Causal Rational Basis Functions Preprints of the 20th World Congress The International Federation of Automatic Control Flexible ILC: Towards a Convex Approach for Non-Causal Rational Basis Functions Lennart Blanken Goksan Isil Sjirk

More information

OPTIMAL H CONTROL FOR LINEAR PERIODICALLY TIME-VARYING SYSTEMS IN HARD DISK DRIVES

OPTIMAL H CONTROL FOR LINEAR PERIODICALLY TIME-VARYING SYSTEMS IN HARD DISK DRIVES OPIMAL H CONROL FOR LINEAR PERIODICALLY IME-VARYING SYSEMS IN HARD DISK DRIVES Jianbin Nie Computer Mechanics Laboratory Department of Mechanical Engineering University of California, Berkeley Berkeley,

More information

Model Based Fault Detection and Diagnosis Using Structured Residual Approach in a Multi-Input Multi-Output System

Model Based Fault Detection and Diagnosis Using Structured Residual Approach in a Multi-Input Multi-Output System SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol. 4, No. 2, November 2007, 133-145 Model Based Fault Detection and Diagnosis Using Structured Residual Approach in a Multi-Input Multi-Output System A. Asokan

More information

Outline. Classical Control. Lecture 1

Outline. Classical Control. Lecture 1 Outline Outline Outline 1 Introduction 2 Prerequisites Block diagram for system modeling Modeling Mechanical Electrical Outline Introduction Background Basic Systems Models/Transfers functions 1 Introduction

More information

GAIN SCHEDULING CONTROL WITH MULTI-LOOP PID FOR 2- DOF ARM ROBOT TRAJECTORY CONTROL

GAIN SCHEDULING CONTROL WITH MULTI-LOOP PID FOR 2- DOF ARM ROBOT TRAJECTORY CONTROL GAIN SCHEDULING CONTROL WITH MULTI-LOOP PID FOR 2- DOF ARM ROBOT TRAJECTORY CONTROL 1 KHALED M. HELAL, 2 MOSTAFA R.A. ATIA, 3 MOHAMED I. ABU EL-SEBAH 1, 2 Mechanical Engineering Department ARAB ACADEMY

More information

A Model-Free Control System Based on the Sliding Mode Control Method with Applications to Multi-Input-Multi-Output Systems

A Model-Free Control System Based on the Sliding Mode Control Method with Applications to Multi-Input-Multi-Output Systems Proceedings of the 4 th International Conference of Control, Dynamic Systems, and Robotics (CDSR'17) Toronto, Canada August 21 23, 2017 Paper No. 119 DOI: 10.11159/cdsr17.119 A Model-Free Control System

More information

Control System Design

Control System Design ELEC ENG 4CL4: Control System Design Notes for Lecture #24 Wednesday, March 10, 2004 Dr. Ian C. Bruce Room: CRL-229 Phone ext.: 26984 Email: ibruce@mail.ece.mcmaster.ca Remedies We next turn to the question

More information

Suppression of the primary resonance vibrations of a forced nonlinear system using a dynamic vibration absorber

Suppression of the primary resonance vibrations of a forced nonlinear system using a dynamic vibration absorber Suppression of the primary resonance vibrations of a forced nonlinear system using a dynamic vibration absorber J.C. Ji, N. Zhang Faculty of Engineering, University of Technology, Sydney PO Box, Broadway,

More information

Optimal Plant Shaping for High Bandwidth Disturbance Rejection in Discrete Disturbance Observers

Optimal Plant Shaping for High Bandwidth Disturbance Rejection in Discrete Disturbance Observers Optimal Plant Shaping for High Bandwidth Disturbance Rejection in Discrete Disturbance Observers Xu Chen and Masayoshi Tomiuka Abstract The Qfilter cutoff frequency in a Disturbance Observer DOB) is restricted

More information

IDENTIFICATION FOR CONTROL

IDENTIFICATION FOR CONTROL IDENTIFICATION FOR CONTROL Raymond A. de Callafon, University of California San Diego, USA Paul M.J. Van den Hof, Delft University of Technology, the Netherlands Keywords: Controller, Closed loop model,

More information

A New Approach to Control of Robot

A New Approach to Control of Robot A New Approach to Control of Robot Ali Akbarzadeh Tootoonchi, Mohammad Reza Gharib, Yadollah Farzaneh Department of Mechanical Engineering Ferdowsi University of Mashhad Mashhad, IRAN ali_akbarzadeh_t@yahoo.com,

More information

Real-Time Feasibility of Nonlinear Predictive Control for Semi-batch Reactors

Real-Time Feasibility of Nonlinear Predictive Control for Semi-batch Reactors European Symposium on Computer Arded Aided Process Engineering 15 L. Puigjaner and A. Espuña (Editors) 2005 Elsevier Science B.V. All rights reserved. Real-Time Feasibility of Nonlinear Predictive Control

More information

Adaptive Robust Precision Control of Piezoelectric Positioning Stages

Adaptive Robust Precision Control of Piezoelectric Positioning Stages Proceedings of the 5 IEEE/ASME International Conference on Advanced Intelligent Mechatronics Monterey, California, USA, 4-8 July, 5 MB3-3 Adaptive Robust Precision Control of Piezoelectric Positioning

More information

and Mixed / Control of Dual-Actuator Hard Disk Drive via LMIs

and Mixed / Control of Dual-Actuator Hard Disk Drive via LMIs and Mixed / Control of Dual-Actuator Hard Disk Drive via LMIs Nasser Mohamad Zadeh Electrical Engineering Department Tarbiat Modares University Tehran, Iran mohamadzadeh@ieee.org Ramin Amirifar Electrical

More information

HYDRAULIC LINEAR ACTUATOR VELOCITY CONTROL USING A FEEDFORWARD-PLUS-PID CONTROL

HYDRAULIC LINEAR ACTUATOR VELOCITY CONTROL USING A FEEDFORWARD-PLUS-PID CONTROL HYDRAULIC LINEAR ACTUATOR VELOCITY CONTROL UING A FEEDFORWARD-PLU-PID CONTROL Qin Zhang Department of Agricultural Engineering University of Illinois at Urbana-Champaign, Urbana, IL 68 ABTRACT: A practical

More information