28th Benelux Meeting on Systems and Control

March 16 18, 2009 Spa, Belgium

Book of Abstracts

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Welcomest is plesure to weloming you to the 28th Benelux Meet the homin ol gress in pD felgiumFting on Systems and ControlD

Scientic Committeehe ienti( gommittee of the PVth fenelux weeting onsists of FEeF esil @nivF tholique de vouvinAD hF eeyels @nivF qentAD F flondel @nivF tholique de vouvinAD yF fosgr @ehE nishe nivF helftAD qF he goomn @nivF qentAD wF qevers @nivF tholique de vouE vinAD wF uinnert @nivF lire de fruxellesAD F vn den rof @ehnishe nivF helftAD qF weinsm @nivF wenteAD fF de woor @utholieke nivF veuvenAD rF xijmeijer @ehnishe nivF iindhovenAD F intelon @rije nivF frusselAD eF vn der hft @ijksunivF qroningenAD tF herpen @ijksuE nivF qroningenAD tF houkens @rije nivF frusselAD wF teinuh @ehnishe nivF iindhovenAD F trmigioli @nivF wenteAD eF toorvogel @nivF wenteAD F n hooren @nivF tholique de vouvinAD qF vn trten @geningen nivFAD F epulhre @nivF de vigeAD rF tigter @geningen nivFAD eF nde ouwer @pult olytehnique de wonsAD vF ehenkel @nivF de vigeAD tF inkin @pults nivFxotreEhme de l ixD xmurAD tF illems @utholieke nivF veuvenAD F eilnd @ehnishe nivF iindhovenAF

he seretrit of the onferene hs een ensured y gE line hizier nd wihle helville from the eFsFwF @essoiE tion des sngnieurs de wonte(oreAF hey re grtefully kE nowledged for their (rstElss supportF he sien(ti orgniztion of the onferene would not hve een possile without the hrd work of wihel tourE ne @onferene wesiteD progrmD nd ook of strtsA nd elin rlette @progrmAF hey re grtefully knowE ledged for full ommitment in the midst of writing their hh thesisF he meeting is (nnilly supported y the following orgE niztions X ! pEpx @felgin xtionl pund for ienti( eE serhA ! felgin rogrmme on snteruniversity ettrtion oles hgy @hynmil ystemsD gontrolD nd yptimizE tionAD initited y the felgin tteD iene oliy y0eF odolphe epulhre nd wihel qevers gonferene orgnizers wrh PHHW

Aimhe im of the fenelux weeting is to promote reserh tivities nd to enhne oopertion etween reserhers in ystems nd gontrolF his is the twentyEeighth in series of nnul onferenes tht re held lterntely in felgium nd he xetherlndsF

Instructions for speakerspor ontriuted leture the ville time is PS minutesF lese leve few minutes of this period for disussion nd room hnges nd dhere to the indited sheduleF sn eh room emers will e villeF gomputers re not providedF

Overview of the Scientic ProgramIF lenry letures y invited spekersNavin Khaneja

Best junior presentation awardgontinuing trdition egun in IWWTD the fenelux meeE ting will lose with the nnounement of the winner of the fest tunior resenttion ewrdF his wrd is given for the est presenttion t the meeting given y junior reserE her @i.e.D someone working towrds hh degreeAF he wrd is spei(lly given for qulity of presenttion rE ther thn qulity of reserhD whih is judged in di'erent wyF et the meetingD the hirs of sessions will sk three volunteers in the udiene to (ll out n evlution formF efter the sessionD the evlution forms will e olleted y the rize gommissioners who will then ompute rnkingF he winner will e nnouned on ednesdyD wrh IVD immeditely fter the (nl letures of the meeting nd he or she will e presented with the wrdD whih onsists of Q

Control of ensemblesRen Vidal

@rrvrd niversityD eA

Binet-Cauchy Kernels for the Recognition of Visual DynamicsPF wini ourseJorge Corts

@tohns ropkins niversityD eA

nd Sonia Martnez @niversity of gliforni t n hiegoD eA

Distributed control of Robotic Networks

QF gontriuted short leturesD see the list of sessions for the titles nd uthors of these leturesF

fook of estrts trophy tht my e kept for one yer nd erti(teF he evlution forms of eh presenttion will e returE ned to the junior reserher who gve the presenttionF he rize gommissioners re toseph inkin @niversity of xmurEpxhAD nd eter reuerger @helft niversity of ehnologyA F he orgnizing ommittee is ounting on the oopertion of the prtiipnts to mke this ontest suessF

PVth fenelux weeting on ystems nd gontrol

Conference locationhe onferene tkes ple in the olress seminr enterD t wlking distne from the enter of the ity of p loted in the felgin erdennesF he homin ol gress is situted t the edge of IQHH hetr forestF he town of p is only IH minutes wy y foot or I minute y funiulr @rilwyAF he p9s new herml fths re just up the rodF homin ol gress ploumontD S fERWHH p felgique

Websiteen electronic version of the fook of estrts n e downloded from the fenelux weeting we site X http XGGwwwFmonte(oreFulgFFeGeneluxmeetingHWG

R

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Part 1 Programmatic Table of Contents

S

fook of estrts

PVth fenelux weeting on ystems nd gontrol

T

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Monday, March 16, 2009 Pierre le Grand Welcome and opening Chair: Rodolphe Sepulchre 11.2511.30

MoA01-6Ampliers

16.0516.30

Estimating a Power-Scalable Linearized Model for

F F F F F F F F F F F F F F F F F F F F F F 26 uoen ndermot rije niversiteit frussel ves olin rije niversiteit frussel qerd ndersteen rije niversiteit frussel ik intelon

MoA02 Plenary Pierre le Grand Binet-Cauchy Kernels for the Recognition of Visual Dynamics Ren Vidal (Johns Hopkins University, USA) Chair: Paul Van Dooren 11.3012.30Binet-Cauchy Kernels for the Recognition of Visual Dynamics

Chair: MoA02-1

Source de la Reine Synchronization and clustering 14.0016.30 14.0014.25F F F F F F F F F F F F F 27 okyo wetropolitn niversity iindhoven niversity of ehnology okyo wetropolitn niversity

A synchronization criterion for delay-coupled systems based on absolute stability

F F F F F F F F F F F F F F F F F F F F F F en idl @tohns ropkins niversityD eA

179

oshiki yguhi renk xijmeijer xoriko nk

MoA01

Identication 1 Chair: Paul Van den Hof MoA01-1purposes F F F F F F F F F fF ruyk pF vogist fF he woor tF he frnterD tn n

Pierre le Grand 14.0016.30 14.0014.25

MoA02-2del for coupled oscillators

14.2514.50F F F F F F F F F F F F F F 28 niversity of vige niversity of vige

Recent advances and open questions on Peskin mo-

elexndre wuroy odolphe epulhre

Identication of a distillation column for PLC control

MoA02-3clustering model

14.5015.15F F F F F F F F F F F F F F F F F F F 29 qhent niversity qhent niversity

F F F F F F F F F F F F F F 21 uro intEvieven utholieke niversiteit veuven utholieke niversiteit veuven smpe

A rst-order phase transition in a multi-dimensional

pilip he met hirk eeyels

MoA01-2

14.2514.50

MoA02-4pled neuronal oscillators

15.1515.40F F F F F F F F F F F F F F F 30 iindhoven niversity of ehnology niversity of veiester iindhoven niversity of ehnology

Semi-passivity and synchronization of diusively cou-

Eects of Overlapping and Windowing on the Estimation of the Frequency Response of a System using White Noise F F F F F F F F F F F F F F F F F F F F F 22 hr F hF idnge rije niversiteit of frussels rofessor tF vF houe niversity of rwik rofessor uF F qodfrey niversity of rwik

irik teur svn yukin renk xijmeijer

MoA02-5cillators.

15.4016.05

MoA01-3dels : an advection-reaction case study

14.5015.15F F F F F F F 23 geningen niversity geningen niversity

On the inuence of positive and negative feedback loops on the phase response curve of biological os-

Linear regressive realizations of LTI state space mo-

urel ueesmn xurulhud uhirudin

F F F F F F F F F F F F F F F F F F F F F F F 31 ierre r niversity of vige odolphe epulhre niversity of vige

MoA01-4

15.1515.40

MoA02-6qF hrion Fepulhre Feutin

Bursting modeling in dopaminergic neurons

System Identication of a Spindle with Active Magnetic Bearings F F F F F F F F F F F F F F F F F F F 24 srF FF flom helft niversity of ehnology rofF drF irF FwFtF n den rof helft niversity of ehnology hrFirF rFrF vngen helft niversity of ehnology rofF irF FrF wunnig hmidt

F F F F 32 niversity of vige niversity of vige niversity of vige

16.0516.30

MoA03

Wellington Control for teleoperation Chair: Thomas Delwiche 14.0016.30 MoA03-1hF vn ij wF teinuh U F F F F F F F F F F F 33 iindhoven niversity of ehnology iindhoven niversity of ehnology

MoA01-5sues

15.4016.05

Towards Automatic Control of Scanning Transmission Electron Microscopes : System Identication Is-

14.0014.25

Haptic Feedback in Telesurgery

F F F F F F F F F F F F F F F F F F F F F F F F F 25 hr erturo ejd helft niversity of ehnology

fook of estrts

PVth fenelux weeting on ystems nd gontrol

MoA03-2tors : Evaluation and Robustness Analysis

14.2514.50F F niversity of niversity of niversity of F F F 34 trsourg trsourg trsourg

MoA04-4accurate identication of microbial kinetics

15.1515.40F F F F 42 uveuven uveuven uveuven

User Adapted Control of Force Feedback Teleopera-

On the comparison of OED/PE strategies for the

vF fr fF fyle iF vrohe wF de wthelin

iv n herlinden uristel fernerts tn pF n smpe

MoA03-3

14.5015.15

MoA04-5Crystallization Processes

15.4016.05F F F F F F F F F F F F F F 43 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology F F F 44 utholieke niversiteit veuven utholieke niversiteit veuven utholieke niversiteit veuven

Stochastic Observers for Industrial Seeded Batch

A Passivity Study of the Classical Position-Force Teleoperation controller F F F F F F F F F F F F F F F F 35 fF illert uFFveuvenD felgium fF gorteville hF eynerts rF n frusselD iFfF nder oorten

eF wesh sF wor woreno eF ruesmn F n den rof

MoA04-6

16.0516.30

MoA03-4

F 36 iF fF nder oorten utholieke niversiteit veuven hF eynerts utholieke niversiteit veuven rF n frussel utholieke niversiteit veuven F unno nd F okokohjiRobust 4-channel Teleoperation Controller Design

15.1515.40

Model analysis for individual-based modelling

eFtF erhulst uF fernerts tFpFwF n smpe

MoA05

MoA03-5for teleoperation

15.4016.05

Groesbeeck Computational methods Chair: Wim Michiels 14.0016.30 MoA05-1itor ynlinx wihel erleysen inent ertzNonlinear projection on manifolds

Optimization of a static output feedback controller

F F F F F F F F F F F F F F F F F F F 37 homs helwihe niversit vire de fruxelles mir eerkne niversit renri oinr wihel uinnert niversit vire de fruxelles vurent gtoireD erge orfs

F F F F F F niversit tholique niversit tholique niversit tholiqueManifolds

14.0014.25F F F 45 de vouvin de vouvin de vouvin

MoA03-6Passivity and Transparency in manipulation

16.0516.30Bilateral Tele-

MoA05-2Fitting Curves on Riemannian Energy Minimization

14.2514.50Using

F F F F F F F F F F F F F F F F F F F F F 38 srF wFgFtF prnken niversity of wente rofFdrFirF F trmigioli niversity of wente

MoA04 Pouhon Pia Estimation for biochemical processes Chair: Alain Vande Wouwer 14.0016.30 MoA04-1 14.0014.25

gh(k mir ierreEentoine esil enuj rivstv irik ulssen

F F F F F F F F F F F F F F F F 46 niversit tholique de vouvin niversit tholique de vouvin

MoA05-3

Algorithms for nonsmooth optimization on manifolds

gF vgemn F epulhre

47 niversite de vige niversite de vige

14.5015.15

MoA05-4cissa

15.1515.40

Stabilizing the baseline current of a microbial fuel cell-based biosensor F F F F F F F F F F F F F F F F F 39 xienke tein etsusD entre of exellene for sustinle wter tehnology ruertus FwF rmelers geningen niversity rns tigter geningen niversity gees fuismn

Optimal H2-design with the smoothed spectral abs-

F F F F F F F F F F F F F F F F F F F F F F F F F 48 toris niervliet uveuven im wihiels uveuven tefn ndewlle uveuven

MoA05-5

15.4016.05F 49 uF F veuven uF F veuven

Computing H-innity norm of time-delay systems

MoA04-2a Dissipative Observer

14.2514.50F F F F F F F F F F F F F F F 40 niversity of trthlyde niversity of trthlyde

Identication of Biochemical Reaction Systems using

ut qumussoy im wihiels

hirk pey iri fullinger

MoA05-6

MoA04-3using a linear quasi unknown input observer

14.5015.15F F F F 41 pult olytehnique de wons pult olytehnique de wons V

Input and state estimation of a microalgae culture

F 50 xiels vn hijk iindhoven niversity of ehnology xthn vn de ouw iindhoven niversity of ehnology renk xijmeijer iindhoven niversity of ehnologyRobust stability assessment in high-speed milling

16.0516.30

MoB01

iF ohEgztl eF nde ouwer

Model reduction Chair: Pierre-Antoine Absil

Pierre le Grand 17.0018.40

PVth fenelux weeting on ystems nd gontrol

fook of estrts

MoB01-1Systems

17.0017.25

MoB03-2tef nler qeert qins tn pFwF n smpe

17.2517.5059 uFFveuven uFFveuven uFFveuven

Model Reduction and Controller Synthesis for L2

Fault detection in a biochemical fed-batch process

F F F F F F F F F F F F F F F F F F F F F F F 51 wrk wutsers iindhoven niversity of ehnology iep eilnd iindhoven niversity of ehnology

MoB01-2

F F F F F F 52 frt fesselink iindhoven niversity of ehnology xthn vn de ouw iindhoven niversity of ehnology renk xijmeijer iindhoven niversity of ehnologyModel reduction for Lur'e type systems

17.2517.50

MoB03-3in a fusion plasma

17.5018.15

A control system for suppression of magnetic islands

MoB01-3red models

17.5018.15

Nonlinear model approximation using block structu-

F F F F F F F F F F F F F F F F F F F F F F 53 ymr xeem ehnishe niversiteit helft eFiFwF ruesmn ehnishe niversiteit helft yF rF fosgr ehnishe niversiteit helft

F F F F F F F F F F F F F F F F F F 60 frt rennenpywEsnstitute for lsm hysis ijnhuizen hrF iF esterhof pywEsnstitute for lsm hysis ijnhuizen hrF srF FFtFwF xuij ehnishe niversiteit iindhoven hrF wFF de frD rofF hrF srF wF teinuh

MoB03-4machines

18.1518.40

Detection and isolation of sensor faults in induction

MoB02

Source de la Reine Model-free control Chair: Vincent Wertz 17.0018.40 MoB02-1phel ponteneu usn wurphy hmien irnst vouis ehenkel

F F F F F F F F F F F F F F F F F F F F F F 61 wnuel qlvez niversit vire de fruxelles wihel uinnert niversit vire de fruxelles

MoB04

17.0017.25F 54 niversity of vige niversity of wihign niversity of vige

Inferring bounds on the performance of a control policy from a sample of one-step system transitions

Humanoid robotics Chair: Philippe Lefvre MoB04-1vFgF isser F grloni F trmigioli

Pouhon Pia 17.0018.40 17.0017.25

Motion control of the Twente Humanoid Head

MoB02-2

17.2517.50F F F F F F F F F 55 niversiteit iindhoven niversiteit iindhoven niversiteit iindhoven

F F 62 niversity of wente niversity of wente niversity of wente

Fixed Structure Controller Synthesis Using Non-

F F F F F F F F F srmk eldgli ehnishe erjen den rmer ehnishe wrten teinuh ehnishe qeorgo engelisParametric Plant

MoB04-2gaze controllers.

17.2517.50

A gaze saccade model based on separate head and

MoB02-3Adaptive tabolism extremum-seeking control of

17.5018.15fed-batch

F F F F F F F F F F F F F F F F F F F 63 ierre hye gvouvin vne yptin xtionl snstitut of relth hilippe vefvre gvouvin qunnr flohmD ueens niversityGgompxeurosi vD lohmdiomedFqueensuF

cultures of micro-organisms exhibiting overow me-

F F F F F F F F F F F F F F F F F F F F F F F 56 vF hewsme pult olytehnique de wons eF nde ouwer pult olytehnique de wons fF rinivsn iole olytehnique de wontrl wF errier

MoB04-3prdri greveoeur tenEvouis honnrd hilippe vefvre

Motor commands are optimized in the gravity eld

64 niversit tholique de vouvin niversit tholique de vouvin niversit tholique de vouvin

17.5018.15

MoB02-4

Model-free Optimal Control Synthesis

erjen den rmer iep eilnd wrten teinuh

F F F F F F F 57 iindhoven niversity of ehnology iindhoven niversity of ehnology iindhoven niversity of ehnology

18.1518.40

MoB04-4qijs vn yort tefno trmigioli

Control of walking robots using virtual springs

F F F 65 niversity of wente niversity of wente

18.1518.40

MoB03

Fault detection applications Chair: Jan Van Impe 17.0018.40 MoB03-1more ? Lazy online batch-end quality estimation : is less

Wellington

MoB05

Distributed control Chair: Dragan Kostic MoB05-1

Groesbeeck 17.0018.40 17.0017.25

17.0017.25

A Jacobi algorithm for distributed model predictive control of dynamically coupled systems F F F F F F 66 hng hon helft niversity of ehnology ms uevizky helft niversity of ehnology frt he hutter helft niversity of ehnology son xeorD woritz hiehl

F F F F F F F F F F F F F F F F F F F F F F F F 58 qeert qins utholieke niversiteit veuven tef nler utholieke niversiteit veuven tn pFwF n smpe utholieke niversiteit veuven W

fook of estrts

PVth fenelux weeting on ystems nd gontrol

MoB05-2

17.2517.50

An application of distributed control : the segmented primary mirror for the European Extremely Large Telescope F F F F F F F F F F F F F F F F F F F F F F 67 ghristin fstin niversity of vige elin rlette niversity of vige odolphe epulhre niversity of vige wrtin himmlerD ooms irmD fk edghiD fertrnd fuvir

Tuesday, March 17, 2009 TuA01 Pierre le Grand Stochastic estimation and decision making Chair: 8.3010.35 TuA01-1 8.308.55

MoB05-3

F F F F F 68 wohmmed izzeldin ehnishe niversiteit iindhoven hrF eF toki ehnishe niversiteit iindhoven rofFdrFirF FFtF vn den fosh ehnishe niversiteit iindhovenModeling and Control of Inkjet Printhead

17.5018.15

Risk-aware sequential decision making and dynamic programming F F F F F F F F F F F F F F F F F F F F 70 foris hefourny niversity of vige hmien irnst niversity of vige vouis ehenkel niversity of vige

MoB05-4

18.1518.40

TuA01-2assessment

8.559.20

A comparison between dierent state estimation methods in a nonlinear distributed parameter system

Estimating distributions from censored microbiological contamination data for use in quantitative risk

gF etml eF nde ouwer wF uinnert gF ils

F 69 niversit vire de fruxelles pult olytehnique de wons niversit vire de fruxelles

F F F F F F F F F F F F F F F F F F F F F F 71 F fusshert utholieke niversiteit veuven eFrF qeererd utholieke niversiteit veuven tFpFwF n smpe utholieke niversiteit veuven wF yttendele F 72 irF hougls lz niversity of qhent rofFhrFirF oin he ueyser qhent niversity E pulty of ingineering hrFirF qrille he vnnoy niversity of qhent rofFhrFirF lentijn uwelsThe Kalman Filter applied to Hydrologic Systems

TuA01-3

9.209.45

TuA01-4nitive radio

9.4510.10

Multi-armed bandit based decision making for cog-

F F F F F F F F F F F F F F F F F F F F F 73 ssim touini ivig hmien irnst niversity of vige ghristophe woy ivig tques liot

TuA01-5Agent-Based Simulations

10.1010.35F F F F F F F F F F F F F F 74 helft niversity of ehnology

Formal Verication and Improved Analysis Tools for

hrF erturo ejd uiz

TuA02

Source de la Reine Nonlinear control 1 Chair: Alain Sarlette 8.3010.35 TuA02-1hFeF hirksz tFwFeF herpen

8.308.5575 niversity of qroningen niversity of qroningen

Passivity-based tracking control of port-Hamiltonian mechanical systems with only position measurements

TuA02-2o vn qils wihel peetjens renk xijmeijer IH

Feedback stabilisation of a pool-boiling system

F F 76 ehnishe niversiteit iindhoven ehnishe niversiteit iindhoven ehnishe niversiteit iindhoven

8.559.20

PVth fenelux weeting on ystems nd gontrol

fook of estrts

TuA02-3control and its rate

9.209.45

TuA04-1hirk ries eter erheijen erjn den hekker

Stabilization via positive invariance for linear distributed parameter systems with constraints on both

Input design for PWA biological cell systems

F F F F F F F F F F F F F F F F F 77 fF eouzid giewiD niversit tholique de vouvinD vvxD felgium F ertz giewiD niversit tholique de vouvinD vvxD felgium wF iF ehh niversit ghoui houkkliD woroo

F F F F 85 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

8.308.55

TuA04-2

8.559.20

A systematic procedure to develop dynamic models of bioprocesses F F F F F F F F F F F F F F F F F F F 86 tF wilier pult olytehnique de wons eF nde ouwer pult olytehnique de wons

TuA02-4approach

9.4510.10

Control of the exothermic CSTR : the power-shaping

F F F F F F F F F F F F F F F F F F F F F F F 78 eudrey pvhe niversit tholique de vouvin henis hohin niversit tholique de vouvin

TuA04-3in AI-2 dynamics of Salmonella Typhimurium

9.209.45

Macroscopic modeling as a tool for gaining insight

TuA02-5sF olt rofFhrF gFF herer

A Study of Stability Conditions for Haptics

F F F F 79 ehnishe niversiteit helft ehnishe niversiteit helft

10.1010.35

F F F 87 estrid wF gppuyns uFFveuven uristel fernerts uFFveuven tn pF n smpe uFFveuven igrid gF he ueersmekerD tos nderleyden

TuA04-4on Bivalve-proxy information

9.4510.10

Three ways to do temperature reconstruction based

TuA03

Wellington Advanced actuation and sensing Chair: Herman Ramon 8.3010.35F F F F 80 hr sr F F vn der roeven ehnishe niversiteit helftAutomatic control of microscope alignment

TuA03-1

8.308.55

F F F F F F F F F F F F 88 wite fuwens rij niversiteit frussel renrik yhlsson vinkping niversity uurt fr rij niversiteit frussel tohn houkensD prnk hehirs

TuA04-5dans batch fermentations

10.1010.35F F F F F F F F F F F F F utholieke niversiteit utholieke niversiteit utholieke niversiteit F 89 veuven veuven veuven

Quantitative characterization of Streptomyces livi-

TuA03-2tron lenses

8.559.20

Characterization of hysteresis within magnetic elec-

F F F F F F F F F F F F F F F F F F F F F F 81 FtF vn free iindhoven niversity of ehnology gFwFwF vn vierop iindhoven niversity of ehnology FFtF vn den fosh iindhoven niversity of ehnology

FEtF h9ruys uF fernerts tFpFwF n smpe tF enn

TuA05

TuA03-3Muscle applications

9.209.45

Control in networks Chair: Bram de Jager TuA05-1F vin fF he hutter F i

Groesbeeck 8.3010.35 8.308.55

Pneumatic circuit assessment in Pneumatic Articial

F F F F F F F F F F F F F F F F F 82 ri oEwinh uveuven rermn mon uveuven rendrik n frussel uveuven

A simplied model for urban trac network control

TuA03-4tFF vn rulzen FwFtF n den rof qF hitter tF vn iijk

Load Dynamics in Piezoelectric Actuation

F F F F F 83 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

9.4510.10

90 helft niversity of ehnology helft niversity of ehnology hnghi tio ong niversity

TuA05-2hybrid models

8.559.20

Distributed control of urban trac networks using

F F F F F F F F F F F F F F F F F F F F 91 srF xiole wrini niversiteit qent hrF srF ene foel niversiteit qent

TuA03-5F F F F F F F tF ehtererg ehnishe gFwFwFvn vierop ehnishe FFtF vn den fosh ehnishedustrial specications

10.1010.35F F F F F F F F F 84 niversiteit iindhoven niversiteit iindhoven niversiteit iindhoven

TuA05-3mation

9.209.45

Optimal control in hybrid vehicles using route infor-

Control of a moving magnet planar actuator at in-

F F F F F F F F F F F F F F F F F F F F F F F F 92 hijs vn ueulen ehnishe niversiteit iindhoven frm de tger ehnishe niversiteit iindhoven wrten teinuh ehnishe niversiteit iindhoven

TuA05-4based measurements

9.4510.10F F F F F F F F F F F F F F F F 93 rije niversiteit frussel rije niversiteit frussel

TuA04

Biochemical processes Chair: Hans Stigter

Pouhon Pia 8.3010.35II

Tracking in WiMAX Networks depending on RSS-

wuss fshr veo n fiesen

fook of estrts

PVth fenelux weeting on ystems nd gontrol

TuA05-5

10.1010.35

TuB01-4processes

17.3017.55

Stability Analysis of Networked Control Systems using a Swichted Linear Systems Approach F F F F 94 wFgFpF honkers iindhoven niversity of ehnology vF retel iole gentrle de ville FFwFrF reemels iindhoven niversity of ehnology xF vFdF ouwD wF teinuh

Model based kinetics estimation for crystallization

F F F F F F F F F F F F F F F F F F F F F F 98 tFeFFissers ehnishe niversiteit iindhoven Feilnd ehnishe niversiteit iindhoven Ffkx ehnishe niversiteit iindhoven

TuB01-5F hjordjevi FwFtF n den rof yFrF fosgr hF teltsem

Mini Course Pierre le Grand Distributed control of Robotic Networks Jorge Corts and Sonia Martnez (University of California at San Diego, USA) Chair: Dirk Aeyels 11.0012.30Distributed control of Robotic Networks F F F F F 193 torge gorts nd oni wrtnez @niversity of gliforni t n hiegoD eA

Compartmental model of a bubble column

F F F F F 99 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

17.5518.20

TuB01-6Nuclear Fusion

18.2018.45

Modeling and simulating the sawtooth instability in

Pierre le Grand DISC PhD Thesis Award 2008 and DISC Certicates Chair: Paul Van den Hof 14.0014.15

F F F F F F F F F F F F F F F F F F F F 100 srF qF itvoet ehnishe niversiteit iindhoven hrF iF esterhof pyw rofFdrFirF wF teinuhehnishe niversiteit iindhoven hrFirF xFtF hoelmn

TuB02 Chair: TuB02-1hF i(mov F epulhre

Source de la Reine Nonlinear control 2 16.1518.45101 niversity of vige niversity of vige

Mini Course Pierre le Grand Distributed control of Robotic Networks Jorge Corts and Sonia Martnez (University of California at San Diego, USA) Chair: Dirk Aeyels 14.1515.45F F F F F 213 torge gorts nd oni wrtnez @niversity of gliforni t n hiegoD eADistributed control of Robotic Networks

16.1516.40

PRC-based phase resetting for nonlinear oscillators

TuB02-2regulation problem

16.4017.05F F F F F F F F F F F F F F F F F 102 helft niversity of ehnology helft niversity of ehnology

Dynamical positioning of ship motion ; a nonlinear

F wuhmmd tFF vn der oude

Pierre le Grand Modeling, simulation and control of distributed parameter systems Chair: Joseph Winkin 16.1518.45 TuB01-1 16.1516.40

TuB01

TuB02-3a mobile robot

17.0517.30

State predictor based on synchronization applied to

Spatial discretization of a 1D Euler-Bernoulli beam model F F F F F F F F F F F F F F F F F F F F F F F F 95 F oss ijksuniversiteit qroningen tFwFeF herpen ijksuniversiteit qroningen

F F F F F F F F F F F F F F F F F F F F 103 elejndro elvrezEeguirre iindhoven niversity of ehnology renk xijmeijer iindhoven niversity of ehnology oshiki yguhi okyo wetropolitn niversity

TuB02-4Inkjet Printhead

17.3017.55

Improving the performance of a Drop-on-Demand

TuB01-2production of low-density polyethylene.

16.4017.05

Model based optimisation of tubular reactors for the

F F F F F F F 96 eter wFwF n irdeghem utholieke niversiteit veuven pilip vogist utholieke niversiteit veuven tn pF n smpe utholieke niversiteit veuven

F F F F F F F F F F F F F F F F F F F 104 eF eF uhlte helft niversity of ehnology hrFsrF FtFeF fomois helft niversity of ehnology rofF hrF F fuk helft niversity of ehnology

TuB02-5tors

17.5518.20

On the robustness of feedback linearizing control schemes for multiple-input/multiple-output bioreac-

TuB01-3cable for DSL applications

17.0517.30F F F F F F F F F F F F rije niversiteit rije niversiteit rije niversiteit F 97 frussel frussel frussel IP

Modeling and Validation of the parameters of a quad

F F F F F F F F F F F F F F F F F F F F F F F F F 105 hniel goutinho pult olytehnique de wons elin nde ouwer pult olytehnique de wons

im pouert grine xeus veo n fiesen ves olin

TuB03

Aerospace Chair: Jonathan Rogge

Wellington 16.1518.45

PVth fenelux weeting on ystems nd gontrol

fook of estrts

TuB03-1iF vn umpen iF de eerdt FF ghu tFeF wulder

Global nonlinear optimization using interval analysis

106 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

16.1516.40

TuB04-5

17.5518.20

Novel Dexterous robotic nger concept with controlled stiness F F F F F F F F F F F F F F F F F F F F F 115 wrtin ssink niversity of wente 'ell grloni niversity of wente hnnis frouwer niversity of wente tefno trmigioli

TuB03-2craft formations using interval analysis

16.4017.05F F F F F F F 107 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

Fuel optimization for constrained rotations of space-

TuB05

iF de eerdt FF ghu tFeF wulder

Groesbeeck Consensus and cooperative control Chair: Ming Cao 16.1518.45 TuB05-1tEg helvenne wF frniky u r tohnsson F mpieriThe arithmetics of average consensus

TuB03-3craft

17.0517.30

Feasibility of thrust vectoring in large passenger air-

rF u

F F F F F F F F F F F F F F F F F F F F F F F F F 108 helft niversity of ehnology

F F F F F F F 116 niversit tholique de vouvin gse estern eserve niversity urD tokholm

16.1516.40

TuB03-4Design with Multivariate Splines

17.3017.55F F F F F F F F F F 109 helft niversity of ehnology helft niversity of ehnology F F F F F F F F F F F F rije niversiteit rije niversiteit rije niversiteit

TuB05-2enwu u wing go

Aerodynamic Model Identication and Flight Control

Second-order Consensus Algorithms

gFgF de isser tFeF wulder

F F F F F F F F 117 gity niversity of rong uong niversity of qroningen

16.4017.05

TuB05-3cessary Conditions for Pareto Optimality

17.0517.30F F F F F F 118 ilurg niversity ilurg niversity

TuB03-5Nonlinear analysis of utter

17.5518.20F 110 frussel frussel frussel

Innite Horizon Cooperative Dierential Games - Ne-

wttijs n de lle tohn houkens teve nlnduit

uduru iswndh eddy to ingwerd

TuB05-4bile robots

17.3017.55

TuB04

Biomedical robotics Chair: Maarten Steinbuch TuB04-1for minimally invasive thoracic surgery

Pouhon Pia 16.1518.45 16.1516.40

Collision-free coordination of a group of unicycle mo-

F F F F F F F F F F F F F F F F F F F F F F 119 hF uosti ehnishe niversiteit iindhoven F edinndr ehnishe niversiteit iindhoven tF grls ehnishe niversiteit iindhoven rF xijmeijer

Design and control of a teleoperated palpation device

engelo futtfuoo homs helwihe wihel uinnert

F F F F F niversit vire de niversit vire de niversit vire de

F F 111 fruxelles fruxelles fruxelles

TuB05-5

F F F F F F F F 120 qerrit xus iindhoven niversity of ehnology teroen loeg xyD fusiness nit eutomotive ene9 vFdF wolengrft iindhoven niversity of ehnologyCooperative Adaptive Cruise Control

17.5518.20

TuB04-2

112 ueesEtn ndsteeg hilips epplied ehnologies hennis fruijnen hilips epplied ehnologies foudewijn erhr hilips epplied ehnologies eter prissenD qeorge de pokertD hennis fosHaptic control of a tele-operated ultrasound probe

16.4017.05

TuB05-6gets

18.2018.45

Multi-robot coverage to locate xed and moving tar-

F F F F F F F F F F F F F F F F F F F F F F F F F 121 hr sr tF ogge qhent niversity hrF srF hF eeyels qhent niversity

TuB04-3static grip force during object manipulation

17.0517.30F F F F 113 niversit tholique de vouvin niversit tholique de vouvin niversit tholique de vouvin

Inuence of the skin moisture at ngertip on the

F endr F vefvre tEvF honnrd

TuB04-4surgery

17.3017.55F F 114 fruxelles fruxelles fruxelles IQ

Modeling of friction in a trocar for minimally invasive

F F F F F F F F F F F F F F F F F F F F F F tF erspehtD niversit vire de F helwihe niversit vire de wF uinnert niversit vire de

fook of estrts

PVth fenelux weeting on ystems nd gontrol

Wednesday, March 18, 2009 WeA01 Pierre le Grand 8.3011.00 8.308.55

WeA02-3dynamics

9.209.45

Energy equipartition and the second law of thermo-

F F F F F F F F F F F F F F F F F F F F F F 130 qerrd vn illigenurg geningen niversity

Identication 2 Chair: Rik Pintelon WeA01-1uentin entmeesters FEeF esil ul n hooren

WeA02-4Extension of the behavioral approach to parameter-varying systems

9.4510.10linear

Identication method for time-varying ARX models

122 niversit tholique de vouvin niversit tholique de vouvin niversit tholique de vouvin

F th tF gF illems F F gF reuerger F wF tF n den rof

F F F F F F F F F F F F F 131 helft niversity of ehnology utholieke niversiteit veuven helft niversity of ehnology

WeA02-5

10.1010.35F 132 ehnishe niversiteit helft ehnishe niversiteit helft ehnishe niversiteit helft

WeA01-2Extracting information on time-varying using multisines

8.559.20systems

An IQC Approach to Robust Estimation against Perturbations of Smoothly Time-Varying Parameters

tohn vtire ik intelon

F F F F F F F F F F F F F F F F F F F 123 rije niversiteit frussel rije niversiteit frussel

sr tF eenmn hr rF urolu rof hr gF F herer

WeA01-3systems with measurement noise

9.209.45F F F F F F F F F rije niversiteit rije niversiteit rije niversiteit F 124 frussel frussel frussel

WeA02-6

10.3511.00

The Optimal Linear Quadratic Feedback State Re-

Blind maximum likelihood identication of Wiener

vurent neylen ik intelon ieter de qroen

133 ingwerd tFgF ilurg niversity lmh qdjh wd niversityD ogykrtD sndonesi ijynti sFiF qdjh wd niversityD ogykrtD sndonesigulator Problem for Index One Descriptor Systems

WeA01-4for black box identication

9.4510.10

WeA03

Comparison of two nonlinear optimization methods

F F F F F F F F F F F F F 125 enne n wulders rije niversiteit frussel wrnix olkert utholieke niversiteit veuven rofF hrF srF tn wevers utholieke niversiteit veuven woritz hiehlD tohn houkens

Automotive applications Chair: Bayu Jayawardhana WeA03-1

Wellington 8.3011.00 8.308.55

Improving Pushbelt Continuously Variable Transmission Eciency via Extremum Seeking Control

WeA01-5idwin eynders profF quido he oek

10.1010.35

System realization of generalized model structures

126 uFFveuven uFFveuven

F F F 134 tn vn der weulen iindhoven niversity of ehnology frm de tger iindhoven niversity of ehnology irik vn der xoll fosh fusiness nit g

WeA03-2emissions

8.559.20

Model predictive control for the reduction of trac

WeA01-6tor for linear dynamic systems

10.3511.00F F F F F F F F F F rije niversiteit rije niversiteit rije niversiteit F 127 frussel frussel frussel

Finite record eects of the errors-in-variables estima-

uurt fr ik intelon qerd ndersteen

F F F F F F F F F F F F F F F F F F F F F F 135 F uF egeye helft niversity of ehnology fF he hutter helft niversity of ehnology tF rellendoorn helft niversity of ehnology

WeA03-3trains

9.209.45

Minimum-fuel control of combustion engine power-

WeA02

Source de la Reine System Theory Chair: Gerard van Willigenburg 8.3011.00 WeA02-1using the Cayley transform

F F F F F F F F F F F F F F F F F F F F F F F F 136 frt erens utholieke niversiteit veuven iri n den fulk utholieke niversiteit veuven

WeA03-4

9.4510.10

8.308.55F F F F F F F F F F F F F 128 niversity of wente niversity of wente

Controlling an active cabin suspension for commer-

Stability analysis in continuous- and discrete-time,

xiels fesseling rns wrt

F F F F F F F F F F F F F F F F F F F F F 137 illemEtn ivers ehnishe niversiteit iindhoven erjn eerhuis xy eutomotive sgo fesselink ehnishe niversiteit iindhoven renk xijmeijercial vehicles

WeA02-2and rational systems

8.559.20F F F F F F F F F F F F F F F F 129 gentrum iskunde 8 snformtiD IR

WeA03-5wthieu qerrd wihel erhegen idwrd rolweg

Identiability of the parametrizations of polynomial

Global chassis control based on load sensing

tn xemov emsterdm

F F F F 138 helft niversity of ehnology helft niversity of ehnology helft niversity of ehnology

10.1010.35

PVth fenelux weeting on ystems nd gontrol

fook of estrts

WeA03-6seat belt systems

10.3511.00F F F F F F F F F ehnishe ehnishe ehnishe F F F F F F F F F 139 niversiteit iindhoven niversiteit iindhoven niversiteit iindhoven

WeA05-1Predictive input shaping prelters

Development of a belt force actuator for controlled

iFF vn der vn eFqF de tger pFiF eldpus

vF n den froek wF hiehl tF wevers

F F F F F F F F F 146 uFFveuven uFFveuven uFFveuven

8.308.55

WeA05-2ilm her uen mits rermn fruyninkx toris he hutter

WeA04

Pouhon Pia Methods for computational biology Chair: Georges Bastin 8.3011.00 WeA04-1using a Bayesian approach

Extending iTaSC to support inequality constraints

F 147 utholieke niversiteit veuven utholieke niversiteit veuven utholieke niversiteit veuven

8.559.20

8.308.55F F F F F F F F F F F F rije niversiteit rije niversiteit rije niversiteit F 140 frussel frussel frussel

Estimating the parameters of a Rice distribution

WeA05-3

vieve vuwers uurt fr endy n woer ik intelon

148 wihel onde ehnishe niversiteit iindhoven oel werry ehnishe niversiteit iindhoven en vn de wolengrftehnishe niversiteit iindhoven wrten teinuhD ihrd uoopsD wrijn vn eghelDirectional repetitive control of a metrological AFM

9.209.45

WeA04-2knowledge

8.559.20

WeA05-4cesses Theory

9.4510.10

Reverse-engineering genetic networks without prior

Iterative Learning Control by Linear Repetitive Pro-

F F F F F F F F F F F F F F F F F F F F F F 141 timmy ymony geningen niversity hrFirF eFtFfF vn foxtel geningen niversity rofFdrFirF qF vn trten geningen niversity hrFirF vFrF de qr'

F F F F F F F F F F F F F F F F F F F F 149 ojieh szke iindhoven niversity of ehnology

WeA05-5version in iterative learning control

10.1010.35F F F F F F F F F 150 utholieke niversiteit veuven rije niversiteit frussel utholieke niversiteit veuven

A constrained Gauss-Newton method for model in-

WeA04-3nite matrices

9.209.45

Stochastic learning of xed-rank positive semide-

F F F F F F F F F F F F F F F F F F F F 142 qilles weyer niversity of vige ilvre fonnel niversity of vige odolphe epulhre niversity of vige

wF olkert eF n wulders wF hiehl tF wevers

WeA05-6for complex mechatronic systems

10.3511.00

Iterative optimization of parameterized trajectories

WeA04-4prnis morno elin nde ouwer qeorges fstin

Metabolic Flux Interval Analysis of CHO Cells

F F F 143 ervie d9eutomtique ervie d9eutomtique niversit tholique de vouvin

9.4510.10

F F F F F F F F F F 151 fruno hepretere utholieke niversiteit veuven tn wevers utholieke niversiteit veuven im ymens plnders wehtronis ehnology gentre qregory inteD lter erdonk

WeA04-5ponent analysis

10.1010.35F F F F F F F F F F F F F F F F F F F 144 niversity of vige niversit tholique de vouvin niversit tholique de vouvin

Plenary

Generalized power method for sparse principal com-

wF tourne F xesterov eter ihtrik F epulhre

Pierre le Grand Control of ensembles Navin Khaneja (Harvard University, USA) Chair: Rodolphe Sepulchre 11.3012.30233

Control of ensembles

F F F F F F F F F F F F F F F F xvin uhnej @rrvrd niversityD eA

WeA04-6tulin fonill woritz hiehl frt he woor tn n smpeD gsGfioeg

10.3511.00145 utholieke niversiteit veuven utholieke niversiteit veuven utholieke niversiteit veuven utholieke niversiteit veuvenD

An initialization procedure for parameter estimation problems using simultaneous Gauss-Newton method

WeB01

Pierre le Grand Advanced application modeling Chair: Yves Rolain 13.5015.55 WeB01-1 13.5014.15

Robust beyond-rigid-body control of next generation wafer stages F F F F F F F F F F F F F F F F F F F F F 152 oert vn rerpen iindhoven niversity of ehnology om yomen iindhoven niversity of ehnology wr vn de l hilips epplied ehnologies ykko fosgr

WeA05 Groesbeeck Optimization for learning and control Chair: 8.3011.00IS

fook of estrts

PVth fenelux weeting on ystems nd gontrol

WeB01-2duced anesthesia in ICU patients

14.1514.40F F F F F F F F F F 153 qhent niversity qhent niversity qhent niversity

Model development for propofol and remifentanil in-

WeB03 Chair: Ren Boel WeB03-1nodes

Power systems

Wellington 13.5015.55 13.5014.15

irF mon rodre irF fogdn xour hrdFirF glr sonesu rofFhrFirF oin he ueyser

Using multidimensional scaling to represent a power

WeB01-3hr fF tywrdhn

Dynamical Modeling of Micro-Assembly Systems

F 154 ijksuniversiteit qroningen

14.4015.05 15.0515.30

system according to the electric distances between

WeB01-4wF uuindersm iF qriEgnseo tFwFeF herpen

F F F F F F F F F F F F F F F F F F F F F F F F 162 plorene ponteneuEfelmudes niversity of vige hmien irnst niversity of vige vouis ehenkel niversity of vige

Modeling and Control of a Wobble Yoke Stirling Engine : Application to a Micro-Cogeneration System

155 ijkuniversiteit qroningen ehnishe niversiteit iindhoven ijkuniversiteit qroningen

WeB03-2wer plants

14.1514.40F F F F F F F F F F F F F F F F F F F F F F 163 iole olytehnique de vouvin iole olytehnique de vouvin

LMI formulation for optimal control of coal red po-

WeB01-5wave detector

15.3015.55

iF imon F ertz

Modeling the baseband output envelope of a micro-

WeB03-3F F F F F F F F F F F F F F eFgFFwF hmoiseux ehnishe FwF rermns ehnishe eF toki ehnishe wF vzrD FFtF n den foshnetworks

14.4015.05F F F F F F F F F 164 niversiteit iindhoven niversiteit iindhoven niversiteit iindhoven

F F F F F F F F F F F F F F F F F F F F 156 vieseth qomm rije niversiteit frussel ves olin rije niversiteit frussel tohn houkens rije niversiteit frussel ik intelon

Non-centralized model predictive control of power

WeB02 Source de la Reine Modeling and control of chemical processes Chair: Karel Keesman 13.5015.55 WeB02-1 13.5014.15Identication of Low Order Model for Large Scale Systems F F F F F F F F F F F F F F F F F F F F F F F 157 FuF ttmwr ehnil niversity of iindhoven iep eilnd ehnil niversity of iindhoven

WeB03-4tems Based on Cyclodissipativity

15.0515.30F F F F F F F F F F 165 ijksuniversiteit qroningen ijksuniversiteit qroningen ivig

Power Factor Compensation in Nonsinusoidal Sys-

hF del uertoEplores tF wF eF herpen F yrteg

WeB03-5ted systems design

15.3015.55F F F F F F F F F F F F F F F F F 166 pult olytehnique de wons

Using the event-driven Petri net in complex distribu-

WeB02-2Ecient multiple objective optimal (bio)chemical engineering applications

14.1514.40control :

elexndre krylnik

pF vogist FwFwF n irdeghem tFpFwF n smpe

F F F F F F utholieke niversiteit utholieke niversiteit utholieke niversiteitfor Model

F 158 veuven veuven veuven

WeB04

Pouhon Pia Articial and living vision systems Chair: Raaella Carloni 13.5015.30 WeB04-1Biological Motion boosts the oculomotor response

WeB02-3An Ecient Methodology and Reduced Order Models

14.4015.05Predictive

Control of SMB Plants Based on the Wave Theory

ils grlos nde ouwer elin

F F F F F F F F F F F F F 159 pult olytehnique de wons pult olytehnique de wons

167 estien goppe niversite tholique de vouvin tenEtques yrn de ivry tohns ropkins niversity wrus wissl niversite tholique de vouvin hilippe vefevre

13.5014.15

WeB02-4processes

15.0515.30

WeB04-2gaze emulation

14.1514.40

The eect of imperfect maintenance on deterioration

Controlling a moving camera setup for humanoid

F F F F F F F F F F F F F F F F F F F F F F 160 mir F prhni ehnishe niversiteit helft tF eF wF vn der eide ehnishe niversiteit helft wF tF ullen ru gonsultnts F F xioli

F F F F F F F F F F F F F F F F F F F 168 F eilink niversity of wente F trmigioli niversity of wente pF vn der reijden niversity of wente

WeB02-5tsper tolte on fkx ykko fosgr

Very fast temperature pulsing : catalytic reactor design

161 ehnishe niversiteit iindhoven ehnishe niversiteit iindhoven ehnishe niversiteit iindhoven IT

15.3015.55

WeB04-3visually guided arm movements

14.4015.05F F F F F F F F F F F 169 niversit tholique de vouvin ueen9s niversity niversit tholique de vouvin

The visuomotor transformation of velocity signals for

quillume velerq qunnr flohm hilippe vefvre

PVth fenelux weeting on ystems nd gontrol F 170 teroen de fest ehnishe niversiteit iindhoven en vn de wolengrftehnishe niversiteit iindhoven wrten teinuh ehnishe niversiteit iindhovenHow To Obtain a 1 kHz Visual Servoing Setup ?

fook of estrts

WeB04-4

15.0515.30

WeB05

Control of vibrations Chair: Jan Swevers WeB05-1-synthesis F F F F FeFgF ershuren xFtFwF vn hijk xF vn de ouw rF xijmeijer

Groesbeeck 13.5015.55 13.5014.15

Active chatter control for high-speed milling using

F F F F F F F F F F F F F F F F F 171 iindhoven niversity of ehnology iindhoven niversity of ehnology iindhoven niversity of ehnology

WeB05-2

172 hrF srF qregory inte plnders9 wehtronis ehnology genter hrF srF teven hevos plnders9 wehtronis ehnology genter hrF srF im ymens plnders9 wehtronis ehnology genter fert tllertD tn weversD ul sA modular bearing with internal piezoshunt damping

14.1514.40

WeB05-3

14.4015.05

Self-sensing actuation and damping of a Piezoelectric Tubescanner

F uuiper qF hitter

F F F F F F F F F F F F F F F F F F 173 helft niversity of ehnology helft niversity of ehnology

WeB05-4wrten itters tn wevers

15.0515.30

Black box identication of a semi-active damper

WeB05-5

15.3015.55F F F F F F F ehnishe ehnishe ehnishe

F 174 uFFveuven uFFveuven

Stabilization & Vibration Control of Gaussmount Suspension System

F gF hing eFeFrF hmen FFtF vn den fosh

F F F F F F F F F 175 niversiteit iindhoven niversiteit iindhoven niversiteit iindhoven

Pierre le Grand Best Junior Presentation Award ceremony 16.1516.30Part 1 : Programmatic Table of Contents

yverview of sienti( progrmPart 2 : Contributed Lectures

F F F F F

5 19 177 245 261

yneEpge strtsPart 3 : Plenary Lectures

F F F F F F F F F F F F

resenttion mterils elphetil list

F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F

Part 4 : List of Participants

Part 5 : Overview Program

IU

fook of estrts

PVth fenelux weeting on ystems nd gontrol

IV

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Part 2 Contributed Lectures

IW

fook of estrts

PVth fenelux weeting on ystems nd gontrol

PH

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Identication of a Distillation Column for PLC Control PurposesBart Huyck, Jos De Brabanter KaHo Sint Lieven - Department Industrieel Ingenieur Email: bart.huyck@kahosl.be jos.debrabanter@kahosl.be Filip Logist, Jan Van Impe K.U.Leuven - Department of Chemical Engineering (CIT) Email: filip.logist@cit.kuleuven.be jan.vanimpe@cit.kuleuven.be Bart De Moor K.U.Leuven - Department of Electrical Engineering (ESAT - SCD) Email: bart.demoor@esat.kuleuven.be1 Introduction In a world where economic and environmental issues become more and more important, efcient control systems have become indispensable. When dealing with complex processes, Model Predictive Control (MPC) is one of the possible control strategies[1]. In practice, current linear and non-linear MPC algorithms require powerful computers. However, since Programmable Logic Controllers (PLCs) with less computational power are used a lot in industry for control, it might be interesting to explore the possibilities and limitations of these devices for MPC. For this purpose, a 6 m high pilot scale binary distillation column, is selected as an industrial example. 2 Goal The column is currently controlled by PI controllers, but the goal is to upgrade the control system with a linear MPC running on a PLC. However, before a model based controller can be used on a PLC, an accurate (but simple) process model has to be constructed. Therefore linear parametric MIMO black-box models (e.g., ARX, ARMAX, and output error) are adopted. 3 Experimental set-up In this set-up, four variables can be manipulated: the reboiler duty Qr, the feed rate Fv, the duty of the feed heater Qv and the distillate ow rate Fd. Measurements are available for the distillate ow rate Fd , the feed ow rate Fv and nine temperatures, i.e., the temperature at the top of the column T t, the temperatures in the center of every packing section (T s1, T s2 and T s3, respectively), the temperature between section 1 and 2 T v1, the ambient temperature Tamb, the temperature in the reboiler of the column T b, and the temperatures of the feed before and after heating (T v0 and T v, respectively).PI

4 Model identication procedure A parametric model is tted following the Box-Jenkins modelling procedure using the Matlab System Identication Toolbox [2, 3]. An experiment with PRBN input signals is performed for 20000 seconds. From these recorded signals, 5 inputs (Qr, Fv, Qv, Fd and Tamb) and 5 outputs (T s1, T s2, T s3 , T t and T b) are selected to create a model. Two datasets are prepared: one with sampling rate of 5 seconds and an other with sampling rate of 60 seconds. Both sets are split up in an identication and validation part. MIMO ARX, ARMAX and OE models are tted and validated. The AIC criterium is adopted to select the correct model order. Additional model reduction is performed with the help of Hankel Singular Values. 5 Results Only ARX and ARMAX models predict the output accurately, but the best performing models are ARMAX models. After model reduction and conversion, these models result in a 6th order state space model for both datasets. The authors believe that these models (despite their low complexity) will predict the output accurately enough to be employed in an MPC algorithm which can be implemented on a PLC. 6 AcknowledgementsWork supported in part by Projects OT/03/30 and EF/05/006 (Center-ofExcellence Optimization in Engineering) of the Research Council of the Katholieke Universiteit Leuven, and by the Belgian Program on Interuniversity Poles of Attraction, initiated by the Belgian Federal Science Policy Ofce. The scientic responsibility is assumed by its authors. References

[1] S. J. Qin and T. A. Badgwell. A survey of industrial model predictive control technology, Contr Eng Pract, 11:733764, 2003. [2] L. Ljung. System Identication: Theory for the User, Second Edition. Prentice Hall, Upper Saddle River, New Jersey, 1999. [3] L. Ljung. System Identication Toolbox Users Guide. The MathWorks, Inc, Natick, 2008.

fook of estrts

PVth fenelux weeting on ystems nd gontrol

Effects of Overlapping and Windowing on the Estimation of the Frequency Response of a System using White NoiseW. D. Widanage Department ELEC Vrije Universiteit of Brussels Pleinlaan 2. 1050. Belgium Email: wwidanag@vub.ac.be J. L. Douce and K. R. Godfrey School of Engineering University of Warwick Coventry. CV4 7AL U.K. Email: John.Douce@btinternet.com Email: K.R.Godfrey@warwick.ac.uking gain of the window. With any such window, the gain from input to output varies with time over the duration of the block. It leads to the suggestion that distortion occurs due to windowing, and the position of this distortion relative to other blocks affects the performance of the overlapping procedure. 3 Simulation Results The effect of overlap on the variance of an estimated frequency response is illustrated, with a pure time delay of 10% of the block length and Hanning windowing. The input is a zero mean white noise process and the statistics of the estimate are obtained over 1000 repeated simulations. With

1 Abstract Errors are introduced when estimating the frequency response of a system, when using random inputs with block overlap and time windowing. The sources of these errors are studied in the absence of external noise. The sources of errors for the bias and variance of the frequency response are identied to be due to the Fourier transforms of the end effects that arise due to nonperiodic signals, and a time varying error over the period of a block that is dependent on the the type of window. Using a pure time delay as an example for the system, it is shown that there is a limit to the improvement achievable in reducing the variance as the block overlap is increased. 2 Introduction Consider an input and output pair of blocks with the system being a pure time delay. The rst D samples (this will be termed the header of the output) in each measured output block yl (n) is uncorrelated with the entire measured input xl (n). Similarly, the last D samples of xl (n) (this will be termed the tail of the input) is reproduced at the start of the next output block, yl+1 (n). Therefore for a given pair of xl (n) and yl (n), the block yl (n) does not capture the full response to xl (n) and contains a term that is correlated with xl (n) and a further term uncorrelated with it. This uncorrelated term, which can be treated as a noise source, leads to the variance of the estimate. The composition of the error between the measured output and the modelled output from an estimated frequency response is considered to explain the effects of block overlap. There are two factors involved. The rst factor arises from the periodic assumption of the Discrete Fourier Transform. The modeled output includes in the header the response to the tail of the input signal, while the true output includes the response to the input over the preceding block xl1 (n). This factor introduces errors in the estimated frequency response function (FRF) due to the header signal diverging from the signal produced by the nonperiodic input block. The second factor is due to the distortion introduced by the time varyPP

Figure 1: The variance of the FRF against percentage overlapwith Hanning window.

Hanning windowing no signicant improvement in variance is achieved for an overlap of more than 65%. This phenomenon is also observed with other types of systems. 4 Conclusions The sources of errors and their effects on the estimated frequency response using overlapping blocks are discussed and illustrated. The errors are due to 1) the response of the tail of the input, 2) the response from the preceding input block, and 3) the time-varying gain of the window function. By considering the distribution of these errors over the length of a block, it leads to the conclusion that the predominant factor in the source of variance and, hence, the performance of the overlapping operation is the characteristic of the window function and not of the dynamics of the system.

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Linear regressive realizations of LTI state space models: an advection-reaction case studyK.J. Keesman and N. Khairudin Systems and Control Group, Wageningen University THE NETHERLANDS karel.keesman@wur.nl1 P ROBLEM FORMULATION AND SOLUTION Estimation of physical parameters in continuous-time linear time-invariant state-space models generally leads to nonlinear estimation problems. It is well known that nonlinear estimation problems frequently lead to local minima solutions. Furthermore, solving these problems can be very (computer) time consuming, especially when multi-start procedures are used. Since the global solution is not known beforehand, no characterization of the systematic error in the estimates can be given. In this presentation, our aim is to uniquely estimate model parameters in state-space model structures while preserving the (original) physical model structure. Our approach, for continuous-time systems, is to handle the parameter estimation problem via discretization and a linear regressive parametric realization of the dynamical system. Unlike data based methods such as subspace identication, we will conserve the physical model structure. In particular, we will consider the class of nite LTI state space systems S(). By the properties of this class, we are able to nd another realization of which is suited for linear estimation and prediction. It will be shown later on, that the resolvent of the system matrix (R(A)) of the discretetime system, plays a key role in this. In specic cases, the discrete-time system matrix (I +A()), with a vector with physical parameters, becomes a bidiagonal matrix or a symmetric tridiagonal matrix. For the latter case, explicit solutions to R(A) are known (see [1]). The goal here is to unravel the structure of R(A), in such a way that we may write the system as a linear regressive set of equations: T = . Herein, i = i () are known reparametrization functions that are not confounded with coefcients that originate from either the discretization step or from constants in A. From here, it is rather straightforward to arrive at an ordinary least squares estimate and at an explicit expression for the output at time instant k. The key objective of the presentation is twofold: to show the derivation of linear regressive model structures from LTI state space models, while conserving the physical model structure, and to illustrate this to an advection-reaction system.PQ

2 C ONCLUDING REMARKS The proposed procedure allows the conservation of the underlying physical model structure in combination with linear regressive parameter estimation. The realization of a linear regressive system from a state space system is based on linearity of the system and linearity in the parameter of A() and B(). References [1] Vries, D. Estimation and Prediction of ConvectionDiffusion-Reaction Systems from Point Measurements. PhD Thesis, Wageningen University. pp. 170, 2008.

fook of estrts

PVth fenelux weeting on ystems nd gontrol

System Identication of a Spindle with Active Magnetic BearingsRogier S. Blom a,b, , Paul M.J. Van den Hof a , H.H. Langen b , R.H. Munnig Schmidt ba

Delft Center for Systems and Control, b Precision and Microsystems Engineering Delft University of Technology

Corresponding author, email: r.s.blom@tudelft.nl

1 Introduction The problem addressed in this presentation has its roots in the area of micro-manufacturing. In particular micromilling is considered, which entails the scaling of conventional milling into the microdomain. Active Magnetic Bearing (AMB) spindle technology is a promising technology for the micro-milling process. Not only are high rotational speeds attainable, but the active nature of these spindles can be used for monitoring and control purposes, resulting in a more stable cutting process and better manufacturing results.(y direction out of plane)

The dynamics are unstable, hence experiments need to be performed in closed-loop; The exural modes of the rotor are very lightly damped and modes split due to gyroscopy; The AMB spindle exhibits nonlinear behavior, caused by the electromagnetic actuators. 3 Approach A frequency domain approach is used to identify the dynamics of the AMB spindle. This approach consists of two stages. In the rst step, a non-parametric estimate of the multivariable frequency response function (FRF) is made. Orthogonal odd random phase multisine excitation signals are selected for their favorable effect on the variance of the FRF estimate [1]. To ensure that for the chosen excitation the effect of the nonlinearities present in the system is negligible, tests are performed to measure the level of nonlinear distortion. The second step involves estimation of a parametric model of the plant dynamics from the frequency response data obtained in the rst step. Using the approach in [2], models in matrix fraction description are t to the estimated FRF. Results will be presented for a high-speed micro-milling AMB spindle (120,000 rpm). Estimated models will be compared for different rotational speeds. 4 Acknowledgement This research is supported by MicroNed and the Delft Center for Mechatronics and Microsystems.

x

z

Rear bearing

Axial bearing

Motor Front bearing Magnetic coil Gap sensor + controller Toolholder Cutting tool

Figure 1: Schematic of an AMB spindle

2 Problem statement Availability of an accurate model of the dynamics of the AMB spindle is of crucial importance to develop the process monitoring and control techniques mentioned in the introduction. Here, identication of the AMB spindle system from measured data sequences is considered. This problem can be characterized by the following aspects (see g. 1): The dynamics are multivariable. The coupling between the inputs (the currents through the coils of the actuators) and outputs (the position of the rotor shaft at the bearings) varies with the rotational speed;PR

References [1] T. Dobrowiecki, J. Schoukens, P. Guillaume. Optimized Excitation Signals for MIMO Frequency Response Function Measurements. IEEE Transactions of Instrumentation and Measurement, Vol. 55, No. 6, Dec. 2006, 20722079. [2] R.A. de Callafon, D. de Roover, P.M.J. Van den Hof. Multivariable least squares frequency domain identication using polynomial matrix fraction descriptions. Proc. 35th IEEE Conference on Decision and Control, Kobe, Japan, 11 - 13 December 1996, 2030-2035.

PVth fenelux weeting on ystems nd gontrol

fook of estrts

Towards Automatic Control of Electron Microscopes: System Identication IssuesDr. Arturo Tejada Ruiz Delft Center for Systems and Controls Delft University of Technology Mekelweg 2, 2628 CD, Delft, The Netherlands Email: a.tejadaruiz@tudelft.nl1 Introduction Scanning transmission electron microscopes (STEMs) are the tools of choice for material science research, since they provide information on the internal structure of a wide range of specimens. These complex machines are operated by skilled technicians, who execute repetitive tasks (e.g., alignment, particle counting, etc.) following long scripted manual procedures and using mainly visual feedback. Automating such procedures is a crucial step towards transforming the STEMs from qualitative tools into exible quantitative nano-measuring tools. This is the goal of the CONDOR project, which is managed by the Embedded Systems Institute (www.esi.nl). To enable this automation, STEM dynamical models of enough delity are needed. Unfortunately, to the best of our knowledge, such models are not available in the literature. The rst steps toward developing such dynamical models were reported in recently [1]. In here, we outline our most recent efforts towards deriving those models. 2 STEM Modeling Figure 1 shows a simplied STEM dynamical model. As explained in [1], a STEM can be divided into two section: Electronics: This represents the STEMs electronics. The value of an operator-controlled knob, u(t), sets the value of an optical parameter, p(t). For instance, u(t) could represent the voltage applied to an electromagnetic lens, while p(t) could represent the corresponding focal distance Optics + Algorithm: This represents the STEM image formation process and a feature extraction algorithm. Note that the image formation process on bright eld mode is a time independent process, while the algorithm is time dependent. The dynamics of the electronics are given by the (possibly nonlinear) function f : R2 R. That is, p(t) = f (p(t), u(t)). The dynamics of the optics + algorithm block are more difcult to establish. Using bright eld images (BFI), these dynamics are given by I(r) = | (r) h(r, p)|2 , where (r) isu(t )

G(r )f (,) p(t )

h(r, p)

I (r )

c(t )

Figure 1: Simplied STEM model. a two-dimensional complex signal that represents the sample electrical potential, h(r, p) represents the microscopes two-dimensional transfer function, which is parameterized by p, and represents the convolution operator (over r). The algorithms used to measure p vary depending on the nature of p. For focal distance measurement, algorithms are readily available based on BFI analysis. Their outputs, c, are static functions of the parameter p. That is, c = g(p), where g : R R is usually a polynomial function. Hence, the optics + algorithm block can be modeled as follows: c(t) = g(p(t )) + (t), where denotes the delay introduced by the algorithm and (t) represents measurement noise. The overall system model is the p(t) = f (p(t), u(t)) c(t) = g(p(t )) + (t), Thus, c(t) is a nonlinear observation of the state variable p(t). The function g has been recently characterized via simulated images and will soon be validated through experiments with STEM microscopes. Acknowledgements This research was sponsored by the Condor project at FEI company, under the responsibilities of the Embedded Systems Institute (ESI). This project is partially supported by the Dutch Ministry of Economic Affairs under the BSIK program. References [1] A. Tejada, S. van der Hoeven, A. den Dekker, and P. van den Hof, Towards automatic control of scanning transmission electron microscopes, in Proc. of the Conference of Control Applications, Saint Pertersburg, Russia, 2009, p. under review.

PS

fook of estrts

PVth fenelux weeting on ystems nd gontrol

Estimating a Power-Scalable Linearized Model for AmplifiersKoen Vandermot, Yves Rolain, Gerd Vandersteen, Rik PintelonVrije Universiteit Brussel, dep. ELEC, Pleinlaan 2, 1050 Brussels, BELGIUMe-mail: koen.vandermot@vub.ac.be

Abstract - The input-output behavior of RF amplifiers is not only function of the frequency but also shows an input power dependency. Here a linearized 2-dimensional model will be estimated by a Maximum Likelihood Estimator.

estimator the parameters am , n a a parametric model are found:( M b, N b )

and b ma, n a

of the

I. INTRODUCTION For the design of telecommunication systems, designers rely on models of the different components of their system. For amplifiers there exists already a linearized and nonparametric model that is easy to create out of input and output signals: the Best Linear Approximation (BLA) [1]. However, this approximation is only valid for the chosen class of signals: the class of the gaussian noise signals with a fixed power spectrum (rms value and coloring are fixed) [2]. A consequence of this approach is that the BLA will be different for an other RMS value of the input signal. Since the input power is one of the tuning parameters in the design of a communication system, the model for the amplifier has to predict the behavior also for a variation of the input power. In this abstract, a 2D rational model will be identified by a maximum likelihood estimator (MLE). The reason for this choice of estimator are its properties: consistency, asymptotic normality and asymptotic efficiency. Due to these statistical properties, a measure for the quality of the obtained model as the uncertainty bounds on the estimates are known. II. ESTIMATING THE MODEL FROM MEASUREMENTS By measuring the S 21 parameter for every chosen couple of the frequency f k and the power of the input signal P l , one gets the complex value of the S 21 transfer function. Since this is a linearization of the input-output behavior of the RF amplifier under the considered operation condition, this gives a nonparametric BLA for each power level separately:Y ( j k, P l ) G m ( j k, P l ) = -------------------------- = S 21 ( j k, P l ) U ( j k, P l )

( m b, n b ) = ( 0, 0 ) G ( j, P ) = -----------------------------------------------------------------------------( M a, N a ) ( m a, n a ) = ( 0, 0 )

mb nb b m , n ( j ) P b b ma na

(2)

a m , n ( j ) a a

P

With ( M a , N a ) and ( M b , N b ) the orders of respectively the denominator and the numerator of this model. When a low order is chosen, it is possible that the model is not capable to explain all the measured dynamics. For an high order, the estimated model will also capture the noise disturbations. Furthermore, choosing higher orders results in models that are not stable over the power and frequency range. For these reasons, aselection of the model order is needed. III. MODEL VALIDATION The norm of the residuals (measurements-model) over the frequencies and powers is taken and compared with its 95% uncertainty level. When circular complex distributed noise is assumed, it is proven that the 95% uncertainty level corresponds with the 3 level, where 2 ( f k, Pl ) is the variance on the measurements for a given frequency f k and a given input power P l . Hence, making this comparison (see Fig. 1) will give an idea about the quality of the obtained parametric model. This is the major advantage of this modelling procedure.14 amplitude(GBLA) [dB] -4 -22 -40 -58 400 700 1000 1300 Frequency [MHz] 1600

G Gm G Gm 3-level

(1)

The goal is to get a 2-dimensional rational parametric model in the frequency and the power, that is valid for every couple of the frequency and the input power within the considered frequency and input power range. A parametric model for this 2-dimensional rational form will be estimated by a MLE. However, this estimator is only able to reach the global minimum when good starting values for the parameters are given. For this reason, a method is proposed that combines a good minimization procedure with the good satistical properties of the MLE [1]. To do so, one starts by estimating the starting values with a Weighted Generalized Total Least Squares (WGTLS) [1]. Next, these starting values are used for a Bootstrapped Total Least Squares estimation (BTLS) [1]. This estimation procedure will give the starting values for the final estimation: a Maximum Likelihood Estimation (MLE). With this last

Fig. 1. The order of the numerator and denominator are chosen: M b = M a = 8 and N b = N a = 8 . Note that an input power slice P = -6.2dBm of the model is considered.

IV. CONCLUSION The proposed method estimates a parametric 2-dimensional rational model that takes into account the frequency as well as the input power dependency and allows to find a measure for the quality of the obtained model. REFERENCES[1] R. Pintelon, J. Schoukens (2001). System Identification. A frequency domain approach. IEEE Press, New Jersey. [2] Y. Rolain, W. Van Moer, R. Pintelon, J. Schoukens. Experimental Characterization of the Nonlinear Behavior of RF Amplifiers, IEEE Transactions on Microwave Theory and Techniques, vol. 45, no. 8, pp.3209-3218, August 2006.

PT

PVth fenelux weeting on ystems nd gontrol

fook of estrts

A Synchronization Criterion for Delay-coupled Systems based on Absolute StabilityToshiki Oguchi1) , Henk Nijmeijer2) , Noriko Tanaka3) 1) Graduate School of Science & Engineering, Tokyo Metropolitan University 1-1, Minami-Osawa, Hachioji-shi, Tokyo 192-0397 Japan Email: t.oguchi@tmu.ac.jp 2) Department of Mechanical Engineering, Eindhoven University of Technology P.O. Box 513, 5600 MB Eindhoven, The Netherlands 3) Department of Precision Engineering, Tokyo Metropolitan University 1-1, Minami-Osawa, Hachioji-shi, Tokyo 192-0397 JapanSynchronization in networks of chaotic systems has been widely investigated in applied physics, mathematical biology, social sciences and control science and interdisciplinary elds since the early work by Fujisaka and Yamada [1] and by Pecora and Carroll [2]. The study on synchronization of coupled systems [3, 4] has extensively dealt with coupled systems with delay-free coupling. In addition, from a viewpoint of control science and engineering, the synchronization problem has been also investigated via control theroy. More recently the interest is spreading to synchronization phenomena of chaotic systems with time-delay and the effect of time-delay in synchronization [5, 6]. In practical situations, time-delays are caused by signal transmission between coupled systems and affect the behavior of coupled systems. It is therefore important to study the effect of time-delay in existing synchronization schemes. The effect of time-delay in synchronization of coupled systems has been investigated both numerically and theoretically by a number of researchers, these works concentrate on synchronization of systems with a coupling term typically described by ui (t) = N j=i Ki j (xi (t ) x j (t )) and there are j=1, few results for the case in which the coupling term is described by ui (t) = N j=i Ki j (xi (t) x j (t )). In either j=1, case, the synchronization error dynamics can be described by a difference-differential equation, and the synchronization problem can be reduced to the stabilization problem for the origin of the error dynamics with suitable conditions on the coupling gain and the time-delay. However most existing conditions for synchronization are based on the LyapunovKrasovskii functional approach and are given by linear matrix inequalities (LMIs), and these criteria tend to have conservative results due to the -almost inherently- conservativeness of the Lyapunov-Krasovskii approach. In this paper, we propose a less conservative criterion for synchronization of coupled systems with time-delay. Throughout, we assume that the error dynamics can be rewritten by a feedback connection of a linear delay system with multiple inputs and outputs and nonlinear elementsPU

which are decentralized and satisfy the sector condition. Then, we derive a synchronization condition for coupled systems with delay by applying the multivariable circle criterion [8] which is an extension of the result by Popov and Halanay [7]. Unlike the conventional synchronization criteria, the derived criterion is based on a frequency-domain condition and avoid the Lyapunov-Krasovskii approach. As a result, the condition does not contain the conservativeness caused by the Lyapunov-Krasovskii approach. The effectiveness of the proposed criterion is shown by examples of coupled Chua systems with delay coupling. The condition obtained by the criterion is less conservative than the conventional LMI condition, and the boundary condition for synchronization is relatively close to the result by numerical simulations.References [1] H. Fujisaka and T. Yamada Stability theory of synchronized motion in coupled-oscillator systems, Prog. Theor. Phys. 69-1, 3247,1983. [2] L. M. Pecora and T. L. Carroll Synchronization in chaotic systems, Phys. Rev. Lett., 64-8, 821-825, 1990. [3] A. Pikovsky, M. rosenblum, and J. Kurths Synchronization -A universal concept in nonlinear sciences-, Cambridge University Press, 2001. [4] C. W. Wu Synchronization in coupled chaotic circuits and systems, World Scientic, 2002. [5] T. Oguchi and H. Nijmeijer Prediction of chaotic behavior, IEEE Trans. on Circ. and Sys. I, 52-11, 26462472, 2005. [6] T. Oguchi, H. Nijmeijer, and T. Yamamoto Synchronization in networks of chaotic systems with time-delay coupling, Chaos, 18, 037108, 2008. [7] V. Popov and A. Halanay On the stability of nonlinear automatic control systems with lagging argument, Automation and Remote Control, 23, 783786, 1962. [8] P. A. Bliman Extension of Popov absolute stability criterion to non-autonomous systems with delays, Int. J. Control, 7315, 13491361, 2000.

Acknowledgments: This work was partially supported by the Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Scientic Research (No. 20560424).

fook of estrts

PVth fenelux weeting on ystems nd gontrol

Recent advances and open questions on Peskin model for coupled oscillatorsAlexandre Mauroy and Rodolphe Sepulchre Departement of Electrical Engineering and Computer Science, University of Li` ge, B-4000 Li` ge, Belgium e e Email: alexandre.mauroy@ulg.ac.be, r.sepulchre@ulg.ac.be1 Introduction Clustering and synchronization are ensemble phenomena commonly observed in natural and articial populations of interacting oscillators. The Peskin model [3] is a seminal and simple model to study such populations. For each oscillator, the evolution of the state variable obeys a differential equation (linear for LIF oscillators, quadratic for QIF oscillators,. . . ) and is comprise between two threshold values. When the high threshold level is reached, the variable is reset to the low one (the oscillator is said to re). The coupling between the oscillators is impulsive: if a ring occurs, the state of every connected oscillators will be increased by a constant value. The behaviors exhibited by identical and all-to-all coupled Peskin oscillators are well-known. For instance, synchronization of the network was proved under some conditions in [2] and clustering phenomena were highlighted in [4]. On the other hand, there exists no mathematical tool to understand, in an extensive way, the behaviors of heterogeneous populations of Peskin oscillators. A new approach seems to be necessary. 3 Synchronous and asynchronous states The continuous model exhibits two distinct stationary behaviors (synchronous and asynchronous states) which are exactly the counterparts of synchronization and clustering phenomena observed in the original Peskin model. The arising of these behaviors depends on the coupling strength and is related, in the Peskin model, to the role of avalanches. The stability of the (a)synchronous state depends both on the coupling nature and on the curvature of the time evolution of the single oscillator state variable, as it was already pointed out in [4]. Considering both homogeneous and heterogeneous populations, we will report on stability analysis and expectations in the two cases of LIF and QIF oscillators. The results will be related to conjectures originally formulated in the context of the Peskin model. References [1] L. F. Abbott and C. van Vreeswijk, Asynchronous states in networks of pulse-coupled oscillators, Physical Review E, 48 (1993), pp. 14831490. [2] R. E. Mirollo and S. H. Strogatz, Synchronization of pulse-coupled biological oscillators, Siam J. Appl. Math., Vol. 50, No. 6, pp. 1645-1662, December 1990 [3] C. S. Peskin, Mathematical Aspects of Heart Physiology, Courant Institute of Mathematical Sciences, New York University, New York, pp. 268-278, 1975 [4] A. Mauroy and R. Sepulchre, Clustering behaviors in networks of integrate-and-re oscillators, Chaos, 18 (2008), p. 037122. Acknowledgments This work was supported by the Belgian National Fund for Scientic Research (FNRS) through a Research Fellowship at the University of Li` ge. This paper presents research ree sults of the Belgian Network DYSCO (Dynamical Systems, Control, and Optimization), funded by the Interuniversity Attraction Poles Programme, initiated by the Belgian State, Science Policy Ofce. The scientic responsibility rests with its author(s).

2 A continuous model Due to the limitations and issues encountered with the original Peskin model, it is relevant to consider a mean-eld approach in which the population of large networks is assimilated to a continuum of oscillators and approximated by a density function. Classical tools of calculus enable the study of this continuous model, through the analysis of a partial differential equation. Furthermore, considering heterogeneous populations, by the introduction of white noise, is mathematically tractable and leads to the study of a FokkerPlanck equation. The present work deals with the development and the study of such a continuous version of the Peskin model. The pulse discrete coupling is shown to become a continuous coupling, which is proportional to the network ring rate, i.e. the ux of oscillators through the high threshold, and which actually corresponds to a limit case of a model studied in [1].PV

PVth fenelux weeting on ystems nd gontrol

fook of estrts

A rst-order phase transition in a multi-dimensional clustering modelFilip De Smet and Dirk Aeyels SYSTeMS Research Group Department of Electrical Energy, Systems and Automation Ghent University Technologiepark Zwijnaarde 914, 9052 Zwijnaarde Belgium Email: Filip.DeSmet@UGent.be, Dirk.Aeyels@UGent.be1 Model description We investigate a multi-agent model describing the formation of clusters, a phenomenon that can be observed in e.g. swarming behavior of animals, opinion formation, or synchronization in systems of coupled oscillators. Each agent belongs to a multi-dynamical state space, and is characterized by an autonomous component and attraction towards the other agents: xi (t) = bi + K j fi j ( x j (t) xi (t) )ex j (t)xi (t) , j=1 N

4 Results for P > 1 The case P > 1 is harder to analyze, and for the investigation of the emerging cluster structure we focus on two special cases: a system with 3 agents, all-to-all interaction and equal weights, and a system with an innite number of agents in a spherically symmetric conguration. In the rst case we investigate the transition between a single cluster containing all three agents, and a conguration with three clusters, each containing a single agent, without an intermediate stage with two clusters. For P > 1 this may happen for generic values of the model parameters, as opposed to the behavior for the case P = 1, where a transition generically involves at most two clusters. In the second case we calculate a lower and an upper bound for the critical value of the coupling strength corresponding to the origination of a central cluster with velocity zero and containing a non-zero fraction of the population. At this transition value, the central cluster has an initial size different from zero, characteristic of a rst-order phase transition. Acknowledgements This paper presents research results of the Belgian Network DYSCO (Dynamical Systems, Control, and Optimization), funded by the Interuniversity Attraction Poles Programme, initiated by the Belgian State, Science Policy Ofce. The scientic responsibility rests with its authors. During this research Filip De Smet was supported by a Ph.D. fellowship of the Research Foundation - Flanders (FWO). References [1] F. De Smet and D. Aeyels. Clustering in a network of non-identical and mutually interacting agents. Accepted by Proceedings of the Royal Society A, 2008. [2] F. De Smet and D. Aeyels. A multi-dimensional model for clustering exhibiting a rst-order phase transition. In preparation, 2009.

(1)

for all i {1, . . . , N}, where j > 0, K 0, N > 1, and xi (t), bi RP . The differentiable functions fi j are nondecreasing with fi j (0) = 0, fi j = f ji , and lim + fi j ( ) = Fi j , for all i and j in {1, . . . , N}, for some symmetric matrix x P F RNN . Furthermore, ex x , for all x in R \ {0}, with e0 0.

2 Clustering behavior In [2] we show that the long term behavior of the system (1) can be characterized by a set partition H of {1, . . . , N}, dening different clusters, such that agents belonging to the same cluster have a common long term average velocity (and agents from different clusters have different long term average velocities). We will refer to this behavior as clustering behavior with respect to cluster structure H. 3 Preliminary results for P = 1 In [1] we show that, for any choice of the model parameters, the system (1) with P = 1 exhibits clustering behavior with respect to some cluster structure H, and we formulate necessary conditions and sufcient conditions (only differing in inequality signs being strict or not) characterizing this cluster structure. Furthermore, if the interaction functions are increasing, then distances between agents from the same cluster approach constant values that are independent of the initial condition.PW

fook of estrts

PVth fenelux weeting on ystems nd gontrol

Semi-passivity and synchronization of diffusively coupled neuronal oscillatorsErik Steur* e.steur@tue.nl* Dept.

Ivan Tyukin I.Tyukin@le.ac.uk

Henk Nijmeijer* h.nijmeijer@tue.nl

of Mechanical Engineering, Eindhoven University of Technology, P.O. Box 513 5600 MB, Eindhoven, The Netherlands1 Introduction It is well known that neurons might synchronize their behavior to each other. Examples include synchronous oscillations in the visual cortex, motor cortex and the olfactory bulb. Moreover, epileptic seizures are characterized by an increase of synchronized activity. We discuss synchronization in networks of electrically coupled neuronal oscillators, i.e. neurons that are linearly coupled via gap junctions. In particular, we present sufcient conditions for asymptotic synchronization of an ensemble of electrically coupled neurons. Semi-passivity and synchronization Let the neurons in the network be represented by the systems x1, j x2, j = f1 (x1, j , x2, j ) + f2 (x2, j , x1, j ) uj 0 , y j = x1, j , (1)

of Mathematics, University of Leicester, University Road, Leicester, LE1 7RH, United KingdomDene the electrical coupling between the systems (1) as where the gains ji = i j 0 represent the synaptic conductances. Dening the k k coupling matrix as k 12 ... 1k i=2 1i 21 2k k i=1,i=2 2i . . . = . . . , .. . . . . . . . k1 k2 . . . k1 ki i=1 u j = j1 (y j y1 ) j2 (y j y2 ) . . . jk (y j yk ) , (2)

Department

we can write (2) as u = y, where u := col (u1 , u2 , . . . , uk ) and y := col (y1 , y2 , . . . , yk ). Note that is symmetric, singular and positive semi-denite (Gerschgorins theorem). Moreover, assuming that the network cannot be divided into two or more disconnected networks, the matrix will have a simple zero eigenvalue. If each system (1) is semi-passive, then the network of electrically coupled systems possesses ultimately bounded solutions. Theorem 1.1 (Synchronization). See [1]. Suppose that 1. each system (1) is semi-passive; 2. each subsystem x2, j = f2 (x2, j , x1, j ) is a convergent system1 . Let i be an eigenvalue of and let the eigenvalues be or dered as 0 = 1 < 2 . . . k . Then there exists > 0 such that if 2 the systems (1) synchronize asymptotically. In [2] we show that the neuronal models of Hodgkin-Huxley, Morris-Lecar, FitzHugh-Nagumo and Hindmarsh-Rose satisfy both assumptions of the theorem. Hence, for some , i.e. for a given network topology in combination with sufciently strong coupling, the neurons in the network asymptotically synchronize. References[1] A.Yu. Pogromsky and H. Nijmeijer, Cooperative Oscillatory Behavior of Mutually Coupled Dynamical Systems, IEEE Trans. Circuits Syst. I, vol. 48, pp. 152-162, 2001 [2] E. Steur, I. Tyukin and H. Nijmeijer, Semi-passivity and synchronization of neuronal oscillators, (submitted)1 see

where j {1, 2, . . . , k} denotes the number of the oscillator in the network, output y j = x1, j R denotes the membrane potential, input u j R is a current stimulus which the jth neuron receives, state x2, j Rm represents internal, possibly biophysically meaningful variables and C 1 vectorelds f1 : R Rm R and f2 : Rm R Rm . Note that many neuronal models are in this normal form or can be put in this form via some (well-dened) transformation of coordinates. Denition 1 (Passivity and semi-passivity). See [1]. A system (1) is semi-passive if there exists a nonnegative storage function V : Rm+1 R+ , V (0) = 0 such that where the function H : Rm+1 R is nonnegative outside some ball. If the function H() is positive outside some ball, then the system (1) is strictly semi-passive. The most important property for our purpose is that a semipassive system, when being interconnected by a feedback u = (y) satisfying y (y) 0, has ultimately bounded solutions [1].QH

V (x j ) y j u j H(x j ),

[1] or [2] for details

PVth fenelux weeting on ystems nd gontrol

fook of estrts

On the inuence of positive and negative feedback loops on the phase response curve of biological oscillatorsPierre Sacr and Rodolphe Sepulchre e Department of Electrical Engineering and Computer Science, GIGA-Systems Biology and Chemical Biology, University of Li` ge, Belgium. e Emails: Pierre.Sacre@ulg.ac.be, R.Sepulchre@ulg.ac.be1 Introduction Rhythmic phenomena are essential to the dynamic behavior of biological systems. They nd their roots in the many regulatory mechanisms that control life at the cellular level. Understanding those molecular and cellular mechanisms is crucial to advances in systems biology. Dynamic models of regulatory mechanisms are made of complex interconnections of feedback loops often described by bloc diagrams. In this