Limits of stability and stabilization of time delay systems a small gain approach 1st edition jing z

Page 1


Limits of Stability and Stabilization of Time Delay Systems A Small Gain Approach 1st Edition Jing Zhu

Visit to download the full and correct content document: https://textbookfull.com/product/limits-of-stability-and-stabilization-of-time-delay-syste ms-a-small-gain-approach-1st-edition-jing-zhu/

More products digital (pdf, epub, mobi) instant download maybe you interests ...

Analysis and Synthesis of Switched Time Delay Systems

The Average Dwell Time Approach Dan Zhang

https://textbookfull.com/product/analysis-and-synthesis-ofswitched-time-delay-systems-the-average-dwell-time-approach-danzhang/

Stability and Control of Nonlinear Time varying Systems

Shuli Guo

https://textbookfull.com/product/stability-and-control-ofnonlinear-time-varying-systems-shuli-guo/

Thermal and Time Stability of Amorphous Alloys 1st

Edition Cheretaeva

https://textbookfull.com/product/thermal-and-time-stability-ofamorphous-alloys-1st-edition-cheretaeva/

Finite-time stability: an input-output approach First Edition Amato

https://textbookfull.com/product/finite-time-stability-an-inputoutput-approach-first-edition-amato/

Computer Controlled Systems with Delay A Transfer Function Approach Efim N. Rosenwasser

https://textbookfull.com/product/computer-controlled-systemswith-delay-a-transfer-function-approach-efim-n-rosenwasser/

Dynamic Systems with Time Delays Stability and Control

Ju H. Park

https://textbookfull.com/product/dynamic-systems-with-timedelays-stability-and-control-ju-h-park/

Virtual Equivalent System Approach for Stability Analysis of Model based Control Systems Weicun Zhang

https://textbookfull.com/product/virtual-equivalent-systemapproach-for-stability-analysis-of-model-based-control-systemsweicun-zhang/

Space Time and the Limits of Human Understanding 1st Edition Shyam Wuppuluri

https://textbookfull.com/product/space-time-and-the-limits-ofhuman-understanding-1st-edition-shyam-wuppuluri/

Stabilization and H∞ Control of Switched Dynamic Systems Jun Fu

https://textbookfull.com/product/stabilizationand-h%e2%88%9e-control-of-switched-dynamic-systems-jun-fu/

Jing Zhu

Tian Qi

Dan Ma

Jie Chen

Limits of Stability and Stabilization of Time-Delay Systems

A Small-Gain Approach

AdvancesinDelaysandDynamics

Volume8

Serieseditor

Silviu-IulianNiculescu,Paris-Sud,CNRS CentraleSupèlec Universitè, Gif-sur-Yvette,France

e-mail:Silviu.NICULESCU@l2s.centralesupelec.fr

Delaysystemsarelargelyencounteredinmodelingpropagationandtransportation phenomena,populationdynamicsandrepresentinginteractionsbetweeninterconnecteddynamicsthroughmaterial,energyandcommunication flows.Thoughtasan openlibraryondelaysanddynamics,thisseriesisdevotedtopublishbasicand advancedtextbooks,explorativeresearchmonographsaswellasproceedings volumesfocusingondelaysfrommodelingtoanalysis,optimization,controlwitha particularemphasisonapplicationsspanningbiology,ecology,economyand engineering.Topicscoveringinteractionsbetweendelaysandmodeling(from engineeringtobiologyandeconomicsciences),controlstrategies(includingalso controlstructureandrobustnessissues),optimizationandcomputation(including alsonumericalapproachesandrelatedalgorithms)bycreatinglinksandbridges between fieldsandareasinadelaysettingareparticularlyencouraged.

Moreinformationaboutthisseriesathttp://www.springer.com/series/11914

JingZhu • TianQi •

JieChen

LimitsofStability andStabilization ofTime-DelaySystems

ASmall-GainApproach

JingZhu CollegeofAutomationEngineering

NanjingUniversityofAeronauticsand Astronautics

Nanjing China

TianQi CenterforControlandOptimization SouthChinaUniversityofTechnology

Guangzhou China

DanMa CollegeofInformationScienceand Engineering NortheasternUniversity Shenyang China

JieChen DepartmentofElectronicEngineering CityUniversityofHongKong HongKong HongKong

ISSN2197-117XISSN2197-1161(electronic)

AdvancesinDelaysandDynamics

ISBN978-3-319-73650-1ISBN978-3-319-73651-8(eBook) https://doi.org/10.1007/978-3-319-73651-8

LibraryofCongressControlNumber:2017963525

© SpringerInternationalPublishingAG,partofSpringerNature2018

Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpart ofthematerialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations, recitation,broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmission orinformationstorageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilar methodologynowknownorhereafterdeveloped.

Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthis publicationdoesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfrom therelevantprotectivelawsandregulationsandthereforefreeforgeneraluse.

Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthis bookarebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernorthe authorsortheeditorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinor foranyerrorsoromissionsthatmayhavebeenmade.Thepublisherremainsneutralwithregardto jurisdictionalclaimsinpublishedmapsandinstitutionalaffiliations.

Printedonacid-freepaper

ThisSpringerimprintispublishedbytheregisteredcompanySpringerInternationalPublishingAG partofSpringerNature

Theregisteredcompanyaddressis:Gewerbestrasse11,6330Cham,Switzerland

Tomyparentsandhusband

Tomyparentsandchildren

Tomyparents

Tomyparents

JingZhu
TianQi
DanMa
JieChen

Preface

Timedelaysareaprevailingsceneinnaturalandengineeredsystems.Modern interconnectednetworksareespeciallypronetolongandvariabledelays;systems andnetworksinthiscategoryaremany,rangingfromcommunicationnetworks, sensornetworks,multi-agentsystems,cyber-physicalsystems,biologicalsystems, tonetworkedcontrolsystems,tonameafew.Exceptinrareinstances,timedelays arelikelytoresultindegradedperformance,poorrobustness,andeveninstability, whichconsequentlyposesigni ficantchallengestotheanalysisanddesignofcontrol systemsunderdelayedfeedback.

Whilearecurringsubjectofstudy,overthelasttwodecadesorsotherehave beenparticularlynotableadvancesinthestabilityanalysisoftime-delaysystems, thankstothedevelopmentofanalysismethodsdrawinguponrobustcontroltheory andthedevelopmentofcomputationalmethodsinsolving linearmatrixinequality (LMI)problems.Anextraordinaryvolumeoftheliteratureisinexistenceonstabilityproblems,andvarioustime-andfrequency-domainstabilitycriteriahavebeen developed.Ofthesedevelopments,whileanoverwhelmingmajorityoftheavailableresultsareobtainedbasedupontime-domainLyapunov –Krasovskiimethods andrequirethesolutionofLMIs,frequency-domainconditionsinthespiritof small-gaintheoremhavealsobeensoughtafter.Generally,time-domainstability conditionsareapplicabletobothconstantandtime-varyingdelays,butareknown tosufferfromavaryingdegreeofconservatism.Incontrast,frequency-domaintests arelargelyrestrictedtoconstantdelaysthoughtheyoftenprovidetightconditions andappearmoresusceptibletofeedbacksynthesis.

Despitetheconsiderableadvancesinstabilityanalysis,controldesignproblems fortime-delaysystemsprovefarmorechallenging.Feedbackstabilizationof time-delaysystemsposesadifficultproblemandhasbeensomewhatanunderdevelopedresearcharea.Fundamentalrobustnessissueshavebeenseldominvestigatedaswell.Furthermore,recentadvancesinbroad fieldsofscienceand engineeringbroughtforthnewissuesandproblemstotheareaoftime-delaysystems;timedelaysresultedfromtheinterconnectedsystemsandnetworkspresent newchallengesunexploredinthepastandareincreasinglyseentohavefarmore graveeffects,whichtheexistingtheoriesdonotseemtobewellequippedwith.

Amongotherchallenges,theseissueshaveledtoourworkthatformsthecore ofthepresentbook.

Wepresentinthismonographastudyonfundamentallimitsandrobustnessof stabilityandstabilizationoftime-delaysystems,withanemphasisontime-varying delay,robuststabilization,andnewlyemergedareassuchasnetworkedcontroland multi-agentsystems.Wedevelopsystematicallyanoperator-theoreticapproachthat departsfromboththetraditionalalgebraicandthecurrentlypervasiveLMIsolution methods.Thisapproachisbuiltontheclassicalsmall-gaintheoremandisofa distinctive flavorofrobustcontrol,whichenablesustodrawuponrichtoolsand techniquesfromrobustcontroltheory.Thebookisorganizedasfollows:

InChap. 1,weprovideanumberofmotivatingexamplesofbothclassicaland contemporaryinterest,togetherwiththeconciseliteraturesurveymostrelevantto thebookcontents.Chapter 2 collectssomekeymathematicalfactsandresults requiredinthesubsequentdevelopments,includingthesmall-gaintheoremand rudimentsofrobustoptimalcontrol.InChap. 3,wedevelopstabilityconditionsfor linearsystemssubjecttotime-varyingdelays.Leveragingonthesmall-gaintheorem,wecastthestabilityproblemasoneofrobuststabilityanalysisandderive accordingly L2 -and L1 -typestabilityconditionsreminiscentofrobuststability boundstypicallyfoundinrobustcontroltheory.Thedevelopmentshowsthata varietyofstabilityconditions,bothexistingandnew,canbeunifiedintheformof scaledsmall-gainconditions,which,otherthantheirconceptualappeal,canbe checkedusingstandardrobustcontroltoolboxes.

Chapter 4 studiesstabilizationproblemsforlinearsystemssubjecttounknown variabledelays.Weinvestigatethefundamentallimitofstabilizationbylinear time-invariant(LTI)controllers.Thisproblem,commonlyreferredtoasthe delay margin problem,concernsthelimitationforaLTIcontrollertorobustlystabilizea time-delaysystem,addressingthequestion:Whatisthelargestrangeofdelaysuch thatthereexistsafeedbackcontrollercapableofstabilizingalltheplantsfordelays withinthatrange?Chapter 4 focusesonsingle-inputsingle-output(SISO)systems withaconstantunknowndelay.Drawinguponanalyticinterpolationtheoryand rationalapproximationtechniques,wedevelopfundamentallimitsonthedelay margin.Theresultsaresubsequentlyextendedtosystemswithtime-varyingdelays inChap. 5,whichdisplayasignifi cantlyincreasedlevelofintricacyandcomplexity.Chapter 6 focusesonthedelaymarginachievablebyPIDcontrollers, wherebycommonpractice,weexaminelow-ordersystems;PIDcontrollersare favoredfortheireaseofimplementationandarewidelyusedincontrolling industrialprocesses.InChap. 7 wegeneralizethedelaymargintothenotionof delayradius,whichconcernstherangeofnonuniform,multivariatedelaysfor multi-inputmulti-output(MIMO)systems.Boundsandestimatesareobtainedina cohesive,unifiedmanner,whichinmostcasesamounttosolvinganeigenvalue problem.

Chapters 8 and 9 thenprogresstocontemporarytopicsonnetworkedcontroland multi-agentsystems.Speci fically,Chap. 8 studiesnetworkedfeedbackstabilization problemsoverlossycommunicationlinks,wheretimedelaymayresultfromthe systemitselforfromthecommunicationchannel.Networkedcontrolisbroadly

referredtoassuchamechanisminwhichcontroltasksareexecutedbyexchanging informationamongsystemcomponentsviasomeformofcommunicationlinks. Sinceinformationtransmissioncannotbeidealandisingeneralnoisy,communicationlossesarise.Wemodelthelossycommunicationchannelsbystochastic multiplicativeuncertainties,whichfurnishageneraldescriptionforsuchcommunicationlossessuchasdatadropoutandfading.Theproblemunderconsiderationis todeterminethefundamentalthresholdofcommunicationnoisesanduncertainties, sothatthedelaysystemcanbestabilizedrobustly.Basedonthemean-square small-gaintheorem,wederivenecessaryandsufficientconditionsforasystemtobe stabilizableunderamean-squarecriterion,forbothSISOandMIMOdelay systems.

Chapter 9 studiesconsensusrobustnessproblemsforcontinuous-time multi-agentsystems.InaMASconsensustask,asetofagentscoordinateto reachaglobalcommonstatebasedontheexchangeoflocalinformationthrougha communicationnetwork.WeassumethattheagentsintheMASreceivedcertain delayedsignals,whereasthedelaymayarisesheerbecauseoftheagentdynamics orduetocommunicationdelaybetweentheagents.Thecentralissueunderour studyistheeffectofthedelayedinformationexchangeonconsensus:Canconsensusstillbeachievedunderdelayedfeedbackofanagent’sneighborinformation?Tomaintainconsensus,howwilldelayconstrainnetworktopology?Fora giventopology,whatisthelargestpossiblerangeofdelayallowedinorderto insurerobustnessofconsensus?Drawinguponconceptsandtechniquesfrom robustcontrol,notablythoseconcerninggainmarginandgain–phasemargin optimizations,wederiverobustnessconditionsforgenerallinearagentstoachieve consensusunderdelayeffects.Theresultsshowthatdelayedcommunication betweenagentswillgenerallyhinderconsensusandimposerestrictionsonthe networktopology.

Acknowledgments:WewishtothankRonG.Chen,XiangChen,LeiGuo, DanielHo,RickMiddleton,andDanielMillerforusefuldiscussions.Theirvaluablesuggestionshelpedshapethisbook.Wealsowanttoexpressourgratitudeto MinyueFu,LiQiu,andWeizhouSu.Partofthematerialspresentedinthisbook hasbenefitedfromourcollaborativeworkwiththem.WeareindebtedtoSilviu Niculescu,forhisenthusiasminourbookproject.FinancialsupportfromHong KongResearchGrantsCouncil(undergrantnumberCityU111012,CityU 11201514,andCityU11200415)andNaturalScienceFoundationofChina(under grantnumber61603179,61603141,61603079,and61773098)isalsogratefully acknowledged.

Nanjing,ChinaJingZhu Guangzhou,ChinaTianQi Shenyang,ChinaDanMa HongKong,ChinaJieChen

3.3 L2 -StabilityConditions

3.3.1Range-IndependentConditions

3.3.2Range-DependentConditions

3.3.3ExtensiontoNoncentricDelayIntervals

3.3.4ExtensiontoTransformedNominalSystems

3.4 L1

5StabilizationofLinearSystemswithTime-VaryingDelays

6.2First-OrderUnstableSystems

6.3Second-OrderUnstablePlants

6.3.1RealPoles

6.3.2ComplexConjugatePoles

6.3.3OscillatoryPoles

6.4Second-OrderExamples

6.5EffectofNonminimumPhaseZeros

6.6UnstablePlantswithTime-VaryingDelay

7.1Introduction

7.2RegionofStability

7.3UniformDelays

7.4RegionofInstability

7.5Summary

SymbolsandAbbreviations

SetsandSpaces

C Thespaceofcomplexnumbers

C Theopenleft-halfofthecomplexplane

C þ Theopenright-halfofthecomplexplane

C Theclosedleft-halfofthecomplexplane

C þ Theclosedright-halfofthecomplexplane

Cn Thespaceof n-dimensionalcomplexvectors

Cm n Thespaceof m by n complexvectors

R Thespaceofrealnumbers

Rn Thespaceof n-dimensionalrealvectors

Rm n Thespaceof n by n realmatrices

Rn þ The n-dimensionalspaceofpositiverealnumbers

Lp Thespaceofthemeasurablefunctions f : R ! Rn with f kkLp \1

H2

Thesubspaceof L2 analyticin ReðsÞ [ 0

H? 2 Thesubspaceof L2 whichconsistsoffunctionsanalyticin ReðsÞ\0

H1 Thesubspaceof L1 analyticin ReðsÞ [ 0

RH1 Therealrationalsubspaceof H1

supp(f Þ Thesupportofareal-valuedfunction f : X ! R

VectorsandMatrices

xi The ithelementofvector x aij Theunitofmatrix A attheintersectionof ithrowand jthcolumn

I Theidentitymatrix

Df diag f1 ; ...; fn ðÞ ofan n-tupleofscalars,vectors,andmatrices ff1 ; ...; fn g

z Theconjugateofacomplexnumber z

xH Theconjugatetransposeofacomplexvector x

xT Thetransposeofacomplexvector x

xviSymbolsandAbbreviations

x kkp TheHölder p-normofavector x with x 2 Cn

f kkLp The Lp normofafunction f with f 2Lp

AH Theconjugateofacomplexmatrix A

Tr ðAÞ Thetraceofmatrix A

F l ðA; BÞ Thelowerlinearfractionaltransformationofmatrix A and B

rðAÞ Thelargestsingularvalueofmatrix A

qðAÞ Thespectralradiusofmatrix A

kmax ðAÞ Thelargestrealeigenvalueofmatrix A

kðM Þ ThelargesteigenvalueoftheHermitianmatrix M

\ðu; vÞ Theprincipalanglebetweentheunitaryvectors u; v 2 Cn

cos\ðu; vÞ Thevalueof uH v jj

A B TheKroneckerproductofmatrix A and B.If

Abbreviations

AQMActivequeuemanagement

AREAlgebraicRiccatiequation

FDEFunctionaldifferentialequation

GEVPGeneralizedeigenvalueproblem

IQCIntegralquadraticconstraint

LFTLinearfractionaltransformation

LKFLyapunov–Krasovskiifunctional

LMILinearmatrixinequality

LTILineartime-invariant

MAREModi fiedalgebraicRiccatiequation

MASMulti-agentsystem

MIMOMulti-inputmulti-output

NCSNetworkedcontrolsystem

SISOSingle-inputsingle-output

SNRSignal-to-noiseratio

TCPTransmissioncontrolprotocol

Chapter1 Introduction

Abstract Physicalresponsesandeffectsrarelytakeplaceinstantlyuponexerting externalsignals.Itmoreorlesstakestimetotransportexternalinputs,suchas material,information,orenergyfromoneplacetoanother.Thisleadsto timedelay. Timedelayprevailsininterconnectedsystemsandnetworks,whichmayarisefrom varioussources,includingphysicaltransportdelay,computationdelay,andsignal transmissiondelay.Inthischapter,wegiveanumberofmotivatingexamplesranging fromclassicalmechanicalsystemstocontemporarybiologicalstudies.Alongwith thisintroduction,wealsoprovideaconcisesurveyintotheliteraturemostrelevant tothetopicsofthisbook.

1.1ExamplesofTime-DelaySystems

Examplesoftime-delaysystemsareplentiful.Manyclassicalmechanicalandelectricaldelaysystemscanbefoundin,e.g.,[1, 2].Conspicuouscontemporaryexamples includethoseofcommunicationnetworksandbiologicalprocesses.

1.1.1AGyroscopicSystem

Agyroscopicsystemisamechanicaldeviceformeasuringandmaintainingorientationbasedontheconservationofangularmomentum.Gyroscopicsystemsare widelyusedininertialnavigationsystems,for,e.g.,precisioncontrolofintercontinentalballisticmissiles(ICBMs),attitudecontrolofhelicopters,andconstructionof gyrocompassesforshipsandaircrafts.

ShowninFig. 1.1 isatypicalgyroscopicsystem,namelyagyroscope.Mechanically,agyroscopehasaspinningwheelmountedonanaxleandtheaxleisfree toassumeanydirection.Whenanoutsideforceisappliedtoahorizontalspinning wheel,thebarycentreofthegyroscopeisshifted,makingthespinningwheelcircle

©SpringerInternationalPublishingAG,partofSpringerNature2018 J.Zhuetal., LimitsofStabilityandStabilizationofTime-DelaySystems, AdvancesinDelaysandDynamics8,https://doi.org/10.1007/978-3-319-73651-8_1

Fig.1.1 Gyroscope

arounditsverticalaxiscounterclockwise.Thereactivemotiontakesplaceapproximately90◦ laterintheplaneofrotation.Translatedintime,thisresultsinatime-delay systemwhosedynamicscanbedescribedbytheapproximatelinearmodel[3, 4]:

where M , G , and K reflectthemass,gyroscopicforce,andstiffness,respectively. Thedelaytime τ referstothetimeforthereactionexertedbythegyroscopicforce.

Let x (t ) = x 1 (t ) x 2 (t ) T ,wherethestatevariablesareselectedas x 1 (t ) = y (t ), x 2 (t ) =˙ y (t ).Thenthesystemcanbeexpressedintermsofthestate-spaceform

1.1.2AClassicalEconomicModel

Asacornerstoneofeconomicstheory,theSolow–Swanmodel,independentlydevelopedbySolow[5]andSwan[6]in1956,seekstoexplaineconomicgrowthbytaking intoaccountthreefactors:physicalcapitalaccumulation,laborgrowth,andproductivityincreasecausedbytechnologicalimprovements.Inthismodel,anincreasein

physicalcapitalaccumulationandlaborgrowthwillcontributetoincreaseineconomicgrowthrate.Oncetheeconomygrowsatasteadyratewithnomorecapital andlaborinvestment,theeconomicgrowthratecanonlybeincreasedthroughtechnologicaldevelopments.

Let k (t ) bethecapitalintensityattime t ,describingthecapitalstockperunitof effectivelabor,and δ, n ,and g betheratesatwhichthecapitaldepreciates,thelabor andthetechnologygrow,respectively.Then,theSolow–Swanmodelisdescribedby thedifferentialequation[5, 6]:

where sk isthefractionsofincomeinvestedinthephysicalcapital.Theparameter α ,0 <α< 1,representstheelasticityofoutputwithrespecttocapital.However, inrealisticeconomicgrowth,itisevidentthattheaccumulationdoesnothappen solelytophysicalcapital.InordertovalidatetheoriginalSolow–Swanmodel(1.3) empirically,Mankiwetal.[7]providedanimprovedmodelwhichaccountsfor capitalandlaboraccumulationsimultaneously,describedbythetwo-dimensional differentialequation

where h (t ) isthelaborintensityand sh isthefractionsofincomeinvestedinthelabor capital.Similarly,theparameter β istheelasticityofoutputwithrespecttolabor.It isrequiredthat α + β< 1.

Disregardingtimedelaybetweentheinvestmentandtheaccumulationofcapital, theproductionandaccumulationofcapitalareassumedtotakeplaceimmediatelyin themodel(1.3)and(1.4),bothofwhichimposelimitsonexplainingfluctuationsin income.Amodifiedmodelwithtimedelay(see,e.g.,[8]andthereferencestherein) isthenproposed:

where τ1 and τ2 arethetimespentintheprocessofthecapitalandlaboraccumulation, respectively.Thissystemhasauniqueequilibriumstate (k e , h e ),where

PerformingaTaylorexpansionaroundtheequilibriumstate (k e , h e ),weobtainan approximatelinearizedsystem

Then,thesystem(1.6)becomes

1.1.3CommunicationNetworks

Acommunicationnetworkconsistsofacollectionof(networkusers/sources)elementsinterconnectedtotransferinformationfromonenodetoanotherthroughsome communicationlinks.Duetotheinherentconflictbetweenanetwork’sbandwidth andnetworktraffic,communicationdelaysarisenaturally,whichgenerallyleadto performancedegradationsandintheextreme,packetlossesandserviceinterruptions [9, 10].Oneofsuchexamplesistheactivequeuemanagement(AQM)intherouters ofcommunicationnetworkswithtransmissioncontrolprotocol(TCP)flows,seenin acasestudypresentedin[11].Thenetworkconsistsofarouterwithtransmission capacity C and N homogeneousTCPsources.Theimplementationofthenetwork isillustratedinFig. 1.2.

Let W (t ) bethecongestionwindowsizeinTCPwithadditive-increaseand multiplicative-decreasebehavior,asshowninFig. 1.3.Letalso q (t ) bethequeue lengthattherouter.Then,anonlinearmodelfordescribingthedynamicsofthe averagecongestionwindowsizeandthequeuelengthoftheroutercanbeobtained as

Fig.1.2 TheimplementationofAQMwithTCPflows

Fig.1.3 TCPwindowsizedynamics

˙ W (t ) = 1 R (t ) W (t ) 2 W (t R (t )) R (t R (t )) p (t R (t )), ˙ q (t ) =

where R (t ) istheround-tripdelaytime, p (t ) istheprobabilityofpacketmarkdue totheAQMschemes.Denoteby ( R0 , W0 , p0 ) theequilibriumpoint,whichisthe solutiontotheequations: W

with T p beingthepropagationdelay.Thelinearizationof(1.8)atthisequilibrium givesriseto

˙ W (t ) =− N R 2 0 C ( W (t ) + W (t R0 )) 1 R 2 0 ( q (t ) q (t R0 )) R0 C 2 2 N 2 p (t R0 ),

˙ q (t ) = N R0 W (t ) 1 R0 q (t ),

where W (t ) = W (t ) W0 , q (t ) = q (t ) q0 ,and p (t ) = p (t ) p0 .Denote x (t ) = W (t ) q (t ) T , u (t ) = p (t ).Thentheequationsin(1.10)canbe rewritteninthestate-spaceform

1.1.4BiologicalSystems

Biologicalandpopulationdynamicsareareaswheredelaysareusefulindescribing manyprocesses,suchasareactionchain(distributedcharacter[12]),atransportprocess(breathingprocessinthephysiologicalcircuitcontrollingthecarbon-dioxide levelintheblood[13]),latencyandshortintercellularphases(inepidemics,as forexample,cell-to-cellspreadmodelsinaparticularcompartment,thebloodstream[14]),spreadofinfectionswithafamily,andepidemicswithintermediate classes,orrecurrentdiseases,assuggestedinvariousrelapse-recoverymodels[15]. Aquintessentialexampleisthebiochemicaloscillatormodelfordescribingcellregulatorymechanisms(see,e.g.,[16]andthereferencestherein),whichcontainsdelays duringthetranslationandtranscriptionprocessesofprotein(endproduct)mRNA self-repression.Delayisalsoaconspicuousphenomenoninchronicmyelogenous leukemia(CML)(see,e.g.,[17]andthereferencestherein);thereareinfactalonglist ofdynamicaldelaymodelsusedtodescribeCML(see,e.g.,[17]andthereferences therein).

Considerforexampletheconcentrationdynamicsofdonorbloodcells D ,which canbeillustratedbythefollowingsimplemodel

Donorbloodcells D movethroughthediagraminFig. 1.4 astheresultofinteraction withanti-donorcells T D .Stemcellsprovidenewdonorbloodcellsatarate S D .Except fordyingatthenaturaldeathrate d D ,thenewcells D perishwithaprobability p . k presentsthekineticmixingrate.Theperishstateindicatesthattheinteractionwith anti-donorcells T D hastriggeredacytotoxicresponse,andthatthecellswillbedead asaresultafteratime ρ .

Adistinctivefeatureaboutthedelaysinbiologicalsystemsisthattheyareof differentscales,andforthisreason,theyarebestmodeledasmulti-inputmulti-output (MIMO)processes.Forexample,amodelincludingfourdelayswasestablishedin [17]ondifferentscalestodescribetheimmunedynamicbetweenTcellandcancer cellsinleukemiapatientsafterbonemarrowtransplants.Itisofbiologicalinterestto seehowlongdelays(1to8days)interactwithshortones(1to5min),andtoidentify whichismoresensitivethantheothers(Fig. 1.5).

Fig.1.4 Thediagramofthe generaldonorbloodcell[17]

1.1.5TheDujiangyanIrrigationProject:AShowcase

ofEarlyMankind’sIngenuity

MinRiver(Minjiang, ),aturbulentupstreamtributaryofthegrandYangtzeRiver, isafast-flowingriverwithplentyofwater.Downstreamisthefertile,richlycultivated anddenselypopulatedChengdubasininthesouthwesternChina.Initsolddays,Min Riverreckedhavocinsummerwithitsviolentfloodswhileprovidedinsufficientwater supplyinwinter.Toalleviatethischronicalplague,sometwothousandyearsago, LiBing,acivilengineerandthenthegovernorofSichuan,begantheconstruction ofDujiangyanirrigationproject.

Hailedasoneofthegreatestengineeringtriumphsinthemankind’searlycivilization,andoneofthemost niubi1 mega-engineeringprojectsinthehistory,the Dujiangyanirrigationsystemisavastwaterregulationprojectcompletedin251BCto servethedualpurposesofwaterconservationandfloodcontrol.Thiscolossalproject, seeninFig. 1.6 andillustratedschematicallyinFig. 1.7,extendsseveralkilometers longalongMinRiver.Itconsistsofthreecoherentlyconstructedparts:Yuzui(Fish Mouth),Feishayan(FlyingSandWeir),andBaopingkou(MouthofTreasureBottle). Yuzui,theprincipalportionoftheproject,dividesMinRiverintotheinnerandthe outertributaries.Forthistohappen,alowislewasconstructedinthecenterofthe river,whichfunctionsasaleveebuiltwithaparticularcurvature,andacrossit,Mount Yulei(JadeFortress)wascutopentoprovideawaterchannel,bytheforceoffireand water.2 Byconstruction,theriverbedoftheinnerriverissignificantlylowerthanthat oftheouterriver,whichinadryseasondiverts60%oftheincomingwaterintothe innerriver,thusprovidingadequatewaterfortheChengduPlainsinwinter.When

1 AcontemporaryChineseslangforsuperlativedescription,meaningextraordinary.

2 Inlackoftoolsbeforetheinventionofgunpowder,fireandwaterwereusedtoheatandcoolthe rockymountainside,sothattherockswerecrackedandremoved.

Fig.1.5 LiBing,(approximately302–235BC,TheWarringStatesPeriod),thechiefarchitectand supervisorofDujiangyanProject;statuedatedsomefourhundredyearsafterhisdeath,totheHan dynasty(202BC–220AD)andunearthedin1974atDujiangyan.

Fig.1.6 ABird’s-EyeviewoftheDujiangyanIrrigationSystem

Fig.1.7 TheschematicdiagramoftheDujiangyanIrrigationSystem

MinRiverfloodsinsummer,sincetheouterriverismuchwiderthantheinnerone, thepercentageofthewaterdistributionreverts,i.e.,60%ofthewaterflowsintothe outerriver,thuspreventingthedownstreamChengduheartlandfrombeingflooded. Becauseofthecurvatureofthewatercourseandsincetheinnerriverissignificantly deeper,itleadstoslowercurrentsthanthoseintheouterriver.Whenobservedat Baopingkou,thiscreatesa defacto delayandonthedownside,theaccumulationof siltandsediment.

Feishayan,alowleveethatindryseasondisconnectstheinnerriverfromtheouter river,servesasanextcascadeindrainingfloodwater,aleakagechannelwhichalong withtheaidofBaopingkoudrainsofffurthertheflood.Infloodingtime,muchofthe waterfromtheinnerriverisblockedbyBaopingkou,whereinthewaterlevelrises. WhenthewaterintheinnerriverreachestoalevelhigherthanFeishayan,theexcess wateroverflowsbacktotheouterriver.Actinglikeanactuator,theswirlingflowat Baopingkoualsomovestheaccumulatedsiltandsediment“flyingover”Feishayan leveetotheouterriver.

1.2LinearTime-DelaySystems

Inoursubsequentstabilitystudies,weshallmainlybeconcernedwithlineartimedelaysystemsdescribedbythestate-spaceequation

x (t ) = A 0 x (t ) + A 1 x (t τ (t )) , (1.13) where A 0 , A 1 ∈ Rn ×n aregivenconstantstatematrices,and τ (t ) isadelayfunction. Toformallyintroducethebackdropofthismodel,considertheclass C ([−r, 0], Rn ),

consistingofcontinuousfunctionsmapping [−r, 0] to Rn .Forany a > 0andany continuousfunctionoftime ψ ∈ C ([t0 r, t0 + a ], Rn ),and t0 ≤ t ≤ t0 + a ,let ψt ∈ C ([−r, 0], Rn ) beasegmentofthefunction ψ definedas ψt (θ) = ψ(t + θ), r ≤ θ ≤ 0.Thegeneralformofa retardedfunctionaldifferentialequation (FDE) (orfunctionaldifferentialequationof retardedtype)is ˙ x (t ) = f (t , x t ), (1.14)

where x (t ) ∈ Rn and f : R × C ([−r, 0], Rn ) → Rn .Equation(1.14)indicatesthat thederivativeofthestatevariables x attime t dependson t and x (ξ) for t r ≤ ξ ≤ t Assuch,todeterminethefutureevolutionofthestate,itisnecessarytospecifythe initialstatevariables x (t ) inatimeintervaloflength r ,say,from t0 r to t0 ,i.e.,

where φ ∈ C ([−r, 0], Rn ) isgiven.Inotherwords, x (t0 + θ) = φ(θ), r ≤ θ ≤ 0.

AretardedFDEmayalsoinvolvehigher-orderderivatives,whichisknownasa higher-orderFDE.Aswithdifferentialequationswithoutdelay,wemayintroduce additionalvariablestotransformahigher-orderFDEtoastandardfirst-orderfunctionaldifferentialequationoftheform(1.14).NotealsothatinaretardedFDE,the highest-orderderivativedoesnotcontainanydelayedvariables.Whensuchaterm doesoccur,thenweencounterafunctionaldifferentialequationof neutral type,or aneutral-delaysystem.Neutral-delaysystemswillnotbeconsideredinthisbook, butmanyofoursubsequentresultscanbegeneralizedtosuchsystems.Time-delay systemscangenerallybedescribedusingFDEs,whichconstituteperhapsthemost prevailingformfoundintheliterature(see,e.g.,[1, 2, 18–20]).Indeed,asearlyas intheeighteenthcentury,FDEswerealreadyusedtosolvegeometryproblemsby EulerandPoisson.Sincethelastcentury,FDEshavebeenfoundtobeusefulfor modelingawidevarietyofproblemsinthefieldsofphysics,economics,andlife sciencesandhavebeenextensivelystudiedin,e.g.,[18]and[19].

Ofparticularinterestinthisbookisasubclassoflinearretardeddelaysystems describedby

where A k ∈ Rn ×n , k = 0, 1,..., q aregivenstatematrices,and τk ≥ 0, k = 1,..., q areconstantortime-varyingdelays.Inthismodel,multipledelays arise.If τk = k τ for k = 1,..., q ,thatis,thedelaysallbecomeintegermultiples ofacertainpositive τ ,wesaythatthedelays τk are commensurate.Otherwise,the delaysaresaidtobe incommensurate.Forsystemswithcommensuratedelays,the state-spaceequation(1.16)canbealternativelyrepresentedintheformof(1.13)by rewriting(1.16)as ˙

,with

Thus,weshallfocusonthesystem(1.13).Similarly,instudyingstabilizationproblems,weshallconsiderthenonautonomoussystem

Here A , B , C aregivenconstantmatrices, x (t ) isthesystemstate, y (t ) isthesystem output,and τ(t ) isasystemdelayatinput.Analogously,wemayalsoconsider systemswithadelaytakingplaceatthesystemoutput.

1.2.1StabilityofLinearTime-DelaySystems

Stabilityoflinearsystemswithdelayshasbeenlongandwellstudied(see[1, 2, 21, 22]andthereferencestherein).Aidedbyrichanalysistoolsinlinearcontrol theory,thelasttwodecadeshavewitnessedextraordinaryadvancesinstabilitystudies oftime-delaysystems,owinglargelytothedevelopmentofoptimizationmethods, [23, 24]androbustcontroltheory[25–27].Sheerbyitsvolume,onecannothelp butconcedethatevenamoderatetrackingofthepastworkbecomesamission impossible.Thus,weshallrestrictourattentiontoresultsthataremostrelevantto thesubsequentdevelopment.

Thestabilitystudyonlineartime-invariantsystemswithconstantdelayshasbeen comprehensiveandfruitful.Thankstodecadesofendeavor,variousstabilityconditions,underbothdelay-dependentanddelay-independentcriteria,havebecome available,bymeansofbothtime-domainandfrequency-domainapproaches[1, 28]; herebydelay-independentstabilityofasystem,wemeanthatthesystemisstable forallvaluesofdelay,andotherwisethesystem’sstabilityisdelaydependent.The time-domainresultsaremainlybuiltonextensionsofLyapunovstabilitytheory, i.e.,theLyapunov–Krasovskii[29]andLyapunov–Razumikhintheorems[30].This approachtypicallyresultsinsufficientstabilityconditionscharacterizedbythesolutionsto linermatrixinequalities (LMIs)[1, 24].Overall,thetime-domainLMI-based methodsoffertheversatilitythattheyarewidelyapplicable,totime-varyingandnonlineardelaysystems,albeitattheexpenseofincreasedconservatism.

Classicalfrequency-domainstabilitytestsarebasedonanalysisofsystemcharacteristicfunctions[31]andoriginatedfromtheearlystudiesofPontryaginon quasipolynomials,whichhaveledtothelaterdevelopmentofthewell-known

two-variablecriterion.Notablemethodsofthetwo-variablecriterionincludethe 2-Dstabilitytestandpseudo-delaymethod(see,e.g.,[1, 32, 33]).Morerecently, inspiredbytechniquesfoundinrobustcontroltheory[26, 34],small-gainstability conditions[35]begantoberecognizedandsubsequentlydevelopedinthe1990sand onwards,leadingtoavarietyoffrequency-sweepingandmatrixpencil-basedcriteria [36–39]thatareefficientlycomputable.Tothiseffect,akeystepistoreformulate delaysbyfictitiousmodelinguncertaintiesinanonconservativeway,whichconsequentlyallowsustodrawuponrichtoolsandtechniquesfromrobustcontroltheory. Thisideawasfirstadvocatedin[37],whichwassoonfurtherdevelopedtolendboth delay-independentanddelay-dependentstabilityconditions.WhilerestrictedtoLTI systemswithconstantdelays,thefrequency-domainstabilityresultshavetheappeal thattheyprovidenecessaryandsufficientconditions.Inparticular,fortheirclose linkagetorobustcontroldesign,thesmall-gainstabilityconditionsappearespecially amenabletofeedbacksynthesisforstabilizationandcontrolperformance.

Generally,whileforspecialcasesequivalencemaybeestablishedbetween Lyapunov-typeandsmall-gainconditions[40],thesmall-gainfrequency-domain testsarefavoredfortheircomputationalefficiencyandnumericalprecision.Bynow itisknownthatwithconstantcommensuratedelays,thestabilityofaLTIsystem, whetherdelaydependentordelayindependent,canbedeterminedefficientlyby solvingamatrixeigenvalueproblem,whileforsystemswithmultipleincommensuratedelays,thestabilityproblemhasbeenfoundtobe NP-hard [41],andthus posesafundamentalcomputationaldifficulty.Inotherwords,forLTIsystemswith constantdelays,thestabilityproblemisbyandlargeresolvedorotherwiseisdeemed computationallyintractable.

Whenthedelayistimevarying,however,stabilityanalysisbecomesmoreintricate.Unlikeinthecaseofconstantdelays,whichenjoysawidevarietyofnonconservativeandreadilycomputablestabilityconditions,onlysufficientstability conditionsareavailableforsystemswithtime-varyingdelays,withvaryingdegrees ofconservatism.Indeed,itisunlikelythatnonconservativenecessaryandsufficient conditionscanbeobtained.Thedifficultyistwofold.First,withatime-varyingdelay, frequency-domainanalysisceasestobeapplicable.Furthermore,itisoftenthecase thatfast-varyingdelayscanresultinunexpected quenchingphenomenon,contrary tointuitionsgainedbyanalyzingsystemswithconstantdelays.Thus,formuchof thetime,thestabilityresultsforsystemswithtime-varyingdelaysaretime-domain conditions,whichintheiressentialflavorisobtainedbasedontheconstructionof Lyapunov–Krasovskiifunctionalsandasthesolutionstolinearmatrixinequality (LMI)problems.Inattemptstoreducetheconservatism,sophisticatedmanipulationsofLyapunov–Krasovskiifunctionalswereemployed(see,e.g.,[42–47]).Yet theLyapunov–Krasovskiifunctionalsandtheimprovedstabilityconditions,which attimesmayseemincremental,donotcomewithoutaprice;theytendtoresultin large-scaleLMIsandhenceforthincreasedcomputationalcomplexity.

Ithasbeenrecognizedforsometime,nonetheless,thatlikeitscounterpartwith constantdelays,thestabilityproblemwithtime-varyingdelayscanstillbecast andconsequentlytackledasoneofrobuststabilityundertime-varyinguncertainties [1],byreformulatingthetime-varyingdelayasalineartime-varyinguncertainty.

Likewise,thisrecognitionhasledtosmall-gainstabilityconditionsinanumber offorms,forexample,intermsof integralquadraticconstraints (IQC)[48, 49], optimallyscaled small-gain conditions[50, 51],orbasedonan quadraticseparation argument [52, 53].Theresultsarecontingentupontheestimationofthe L2 -induced normofthelineartime-varyingoperator.Stabilityconditionsinthisspiritwillbe systematicallydevelopedinChap. 3.Whileincomputationtheymayfarebetterin somecasesandworseinotherstotheirLMIcounterparts,thesesmall-gainfrequencydomainconditionsappeartobeconceptuallysimpleandcomputationallyefficient; variouswell-establishedrobustcontroltoolboxescanbeappliedinastraightforward manner.

1.2.2StabilizationofLinearTime-DelaySystems

Notwithstandingtheremarkableprogressonstabilitystudies,thestabilizationof time-delaysystemsprovesfundamentallymoredifficultaproblem.Theexisting workhasbeenlargelyfocusedonsynthesisproblemsforsystemswitha fixed delay. Feedbacksynthesisforsuchsystemscanbeconductedbasedonclassicaltechniques suchasfinitespectrumassignment[54]andtheSmithpredictorapproach[55], onLQRand H∞ techniques(see,e.g.,[56, 57]andthereferencestherein),via predictorfeedback[58, 59],orusingLMI-basedsolutions[2, 60].Ontheother hand,fundamentalrobustnessofstabilizationinthepresenceof uncertain,variable delayshasbeenseldominvestigated.Norisitclearhowthemethodsalludedtoabove maybeextendedtoaddresstherobuststabilizationproblem.Inthisvein,particularly noteworthyistheproblemof delaymargin [61],whichbynatureaddressesasystem’s robuststabilizationagainstuncertaindelaysandseekstoanswerthequestion: What isthelargestrangeofdelaysuchthatthereexistsasinglefeedbackcontrollerthat canstabilizealltheplantssubjecttodelayswithintherange? Anage-oldproblem byitself[62, 63],thisproblembearsaclosesimilaritytothegainmarginandphase marginoptimizationproblems,whichcanbestatedasfollows:

• GainMarginProblem: Considerthefamilyofplants P ={kP0 (s ) : k ∈[1, k 1 )}.

Supposethat P0 (s ) canbestabilizedbysomecontroller K (s ). Whatisthelargest k 1 suchthatthereexistsacertaincontrollerK (s ) whichstabilizesallplantsin P ?

• PhaseMarginProblem: Considerthefamilyofplants

P ={e j θ P0 (s ) : θ ∈[−θ1 ,θ1 ]}

Supposethat P0 (s ) canbestabilizedbysomecontroller K (s ) Whatisthelargest θ1 suchthatthereexistsacertaincontrollerK (s ) whichstabilizeseveryplant in P ?

Thegainmarginandphasemarginoptimizationproblemsaretwoclassicalrobust stabilizationproblemssolvableanalyticallybysolvingafinite-dimensional H∞ optimalcontrolproblem[25].Boththeseproblemswillserveourpurposesintackling thedelaymarginproblem(cf.Chap. 4)andmulti-agentconsensusproblems(cf. Chap. 9).Nonetheless,unlikethegainandphasemarginproblem,thedelaymargin problemprovesfundamentallymorechallenging,duetoobstaclesinsolvinginfinitedimensionaloptimizationproblems.Indeed,theproblemisingeneralmoredifficult thantheaforementionedstabilityproblem,orthestabilizationproblemwithafixed delay,andforthatmatter,hasbeenopenexceptinisolatedcases.

Indeed,itappearssafetocontendthatmuchofthepastworkonthedelaymargin problemhasonlymetwithsporadicsuccesses.In[2](p.154),thedelaymarginwas determinedforthefirst-ordersystemachievablebystaticfeedback,whilein[64], thedelaymarginwasfoundforthefirst-ordersystemwhenPIDcontrollersareused. Morerecently,upperboundsonthedelaymarginwereobtainedin[61, 65]for generalSISOsystemssubjecttoanuncertainconstantdelay.Theseboundsserveto providealimitbeyondwhichnosingleLTIoutputfeedbackcontrollermayexistto robustlystabilizethedelayplantfamilywithinthemargin.Theresultsshowthatthis fundamentallimitisdeterminedbytheunstablepolesandnonminimumphasezeros intheplant.Initsessence,however,theseresultsarebyandlargelimitedtosystems withnomorethanoneunstablepoleandnonminimumphasezero,forwhichthe boundswerefoundtobeexact;otherwise,undermoregeneralcircumstances,the boundsmaybecrudeandpessimistic.Moreover,theanalysiswascarriedoutlargely casebycase,andforthisreason,itstechniquedoesnotappearreadilygeneralizable. Theauthorsrecentlydevelopedlowerboundsonthedelaymarginofgeneral SISOdelaysystemsandfurtherestimatesonthedelayregionofgeneralmulti-input multi-output(MIMO)delaysystems.Unliketheupperbounds,whichaddressthe questionwhenadelaysystemis not stabilizable,thelowerboundsanswertothe questionwhenit is robustlystabilizable,i.e.,theentirefamilyoftheplantscanbe robustlystabilizedbyasingleLTIcontrollerforallpossibleuncertaindelayswithin thatbound.Thus,theseresultsprovidea guaranteed rangeofdelayensuringstabilizability.Builtonsmall-gainstabilityconditions,weadoptedanoperator-theoretic approachthatemploysrationalapproximationofdelayelements,whichenablesus tocasttheproblemasoneoffinite-dimensional H∞ optimization;thelattermaythen betackledandsolvedusingsuchanalyticinterpolationtechniquesasNevanlinna–Pickinterpolation[66].Thisapproachwillbefullydevelopedinthisbook.Chapter 4 addressesthestabilizationofSISOsystemswithaconstantunknowndelay.Explicit boundsonthedelaymarginarederivedforSISOsystemswithanarbitrarynumber ofplantunstablepolesandnonminimumphasezeros,whichrequirescomputingthe largestrealeigenvalueofaconstantmatrix.Thisresultisthenextendedtosystems withtime-varyingdelaysinChap. 5,givingrisetoboundsbothdependentonand independentofdelayvariationrate.Chapter 6 providesboundsonthedelaymarginoflow-ordersystemsachievablebyPIDcontrollers.PIDcontrollersrepresent themostprevailingclassofcontrollerswithfixedstructureandcomplexity;hence, theseresultsareofsignificantinterestfortheireasyofimplementationandfortheir implicationonindustrialcontroldesign.Finally,thedelaymarginisgeneralizedto

1.2LinearTime-DelaySystems15 thenotionofdelayradiusinChap. 7,toaddressrobuststabilizationproblemsfor MIMOdelaysystems.

Theoperator-theoreticapproachalludedtoabove,whichisunifiedunderthe bannerofsmall-gainconditionsandanalyticfunctioninterpolation,isnotonlycomputationallyattractive,butalsoisconceptuallyappealing;whenspecializedtomore specificcases,e.g.,toplantswithoneunstablepoleandonenonminimumphasezero, theyyieldanalyticalexpressionsexhibitingexplicitdependenceoftheboundsonthe poleandzero,showinghowfundamentallyunstablepolesandnonminimumphase zerosmaylimittherangeofdelaysoverwhichaplantmayberobustlystabilized. Furthermore,sincetheapproachamountstosolvingastandard H∞ controlsynthesis problem,itinfactyieldsarobustlystabilizingcontrollerthatachievesthebounds andguaranteesthestabilizationforallpossibledelayvalueswithinthebounds.It shouldbeemphasizednonethelessthattheseresultsandconclusionsareonlylimitedtofinite-dimensionalLTIcontrollers.Infinite-dimensionalLTIcontrollersvia delayedfeedbackcanbeemployedtostabilizeunstableplants[67, 68],andmore generalcontrollerswithvaryingdegreesofimplementationcomplexity,suchaslinear periodiccontrollers[69],nonlinearperiodiccontrollers[70],andnonlinearadaptive controllers[71, 72]canbeconstructedtolendaninfinitedelaymargin,allowinga LTIdelayplanttobestabilizedforarbitrarilylonguncertaindelays.

1.3DelaysinNetworkedControlSystems

Networkedcontrol,whichcangenericallybedescribedbytheschematicdiagram depictedinFig. 1.8,isbroadlyreferredtosuchasettingthatcontroltasksareexecutedoverwiredorwirelesscommunicationlinks.Inbroadterms,networkedcontrol systems(NCS)consistofactuators,sensors,andcontrollerswhicharespatiallydistributed,andwhoseoperationsareexecutedbyexchanginginformationandcontrol signalsamongthesystemcomponentsviasomeformofcommunicationnetworks consistingofencoders,decoders,andcommunicationchannels.Bythisnature,networkedcontrolsystemsdifferfromtheconventionalfeedbackcontrolsystemsin anumberofimportantways,andtheyusherinnovelchallengesunexploredinthe past.Correspondingly,thestudyofnetworkedcontrolproblems(seeforexample

Fig.1.8 Networkedcontrol system
Plant Sensor

Another random document with no related content on Scribd:

grana.

¿Hay mas beldad que ver la pradería estrellada con flores de las plantas, que van mostrando el fruto y la alegría?

Donde, con profundíssimas gargantas, las tiernas avecillas estudiosas están de señalar cuales y cuántas.

Allí veréis pastoras más hermosas (no con maestra mano ataviadas), que las damas en Cortes populosas.

Allí veréis las fuentes no tocadas distilando, no agua al viso humano, mas el cristal de piedras variadas.

Allí veréis el prado abierto y llano, donde los pastorcillos su centella descubren al Amor, furioso, insano.

Este, de su pastora se querella; aquél de sí, por que miró la suya; el otro, más grossero, se loa della.

No hay quien por defeto se

lo arguya, ni quien de rico ponga sobrecejo, ni quien á los menores dexe y huya.

En el prado se oye el rabelejo, la zampoña resuena en la floresta, en la majada juegan chueca ó rejo.

Pues qué ¿venido el día de la fiesta, hay gusto igual que ver á los pastores haciendo á las pastoras su requesta?

Uno presenta el ramo de las flores, y cuando llega, el rostro demudado, otro dice suavíssimos amores. Uno llora, y se muestra desamado; otro ríe, y se muestra bien querido; otro calla, y se muestra descuidado.

El uno baila, el otro está tendido; el uno lucha, el otro corre y salta, el otro motejado va corrido. En esta dulce vida, ¿qué nos falta? y más á mí que trato los pastores,

y cazo el bosque hondo y la sierra alta, Con arco, perchas, redes y ventores, ni basta al ave el vuelo presuroso, ni se me van los ciervos corredores.

Este sabuesso era un perezoso, y ya es mejor que todos: halo hecho que, como mal usado, era medroso.

Tiene buen espinazo y muy buen pecho y mejor boca: ¡oh pan bien empleado! toma, Melampo, y éntrete en provecho.

Quiérome ya sentar, que estoy cansado; ¡oh seco tronco, que otro tiempo fuiste fresno umbroso, de Ninfas visitado!

Aquí verás el galardón que hubiste, pues te faltó la tierra, el agua, el cielo, después que este lugar ennobleciste.

Assí passan los hombres en el suelo; después que han dado al mundo hermosura, viene la muerte con escuro

velo.

Ya me acuerdo de ver una figura que estaba en tu cogollo dibujada, de la que un tiempo me causó tristura.

Estaba un día sola aquí sentada; ¡cuán descuidado iba yo de ella, cuando la vi, no menos descuidada!

Puse los ojos y la vida en ella, y queriendo decirla mis dolores, huyó de mí, como yo ahora della.

Por cierto grande mal son los amores, pues al que en ellos es más venturoso, no le faltan sospechas y temores.

Igual es vivir hombre en su reposo.

¿Quién es aquel pastor tan fatigado? Debe de ser Florelo ó Vulneroso.

La barba y el cabello rebuxado, la frente baxa, la color torcida. ¡Qué claras señas trae de enamorado!

¿Es por ventura Fanio?

¡Qué perdida tengo la vista! Fanio me parece.

¡Oh Fanio, buena sea tu venida!

FANIO

Amado D, el cielo que te ofrece tanta paz y sossiego, no se canse, que solo es bien aquel que permanece.

DELIO

Aquesse mismo, F mío, amanse el cuidado cruel que te atormenta, de suerte que tu corazón descanse.

He desseado que me diesses cuenta, pues que la debes dar de tus pesares á quien contigo, como tú, lo sienta.

Y quiero, F, por lo que tratares perder la fe y el crédito contigo, cuando en poder ajeno lo hallares.

Sabe que al que me ofrezco por amigo,

la hacienda pospuesta y aun la vida, hasta el altar me hallará consigo.

FANIO

D, tu voluntad no merecida no es menester mostrarla con palabras, pues en obras está tan conocida.

Pero después que tus orejas abras, más lastimosas á escuchar mi duelo en un lenguaje de pastor de cabras,

Ni á ti podrá servirte de recelo, pues ya tienes sobradas prevenciones, ni á mí de altivo en tanto desconsuelo.

Y no son de manera mis passiones que se puedan contar tan de camino, que aunque sobra razón, faltan razones.

DELIO

Conmigo te han sobrado de contino, entendiendo que la hay para

encubrirme lo que por más que calles adivino.

Y aunque me ves en porfiar tan firme, sabe que poco más que yo barrunto de tu importancia puedes descubrirme.

Y pues me ves en todo tan á punto para mostrarme amigo verdadero, no me dilates lo que te pregunto.

Cuéntame tus passiones, compañero, cata que un fuego fácil encubierto suele romper por el templado acero.

FANIO

Oh, caro amigo mío, y cuán más cierto será hacer mis llagas muy mayores, queriéndote contar mi desconcierto.

Porque siendo mis daños por amores, tú pretendes saber, contra derecho, más que la que ha causado mis dolores.

Salga el nombre de L de

mí pecho y toque á tus orejas con mi daño, ya que no puede ser por mí provecho.

No me quexo de engaño ó desengaño, de ingratitud, de celos ni de olvido, quéxome de otro mal nuevo y extraño.

Quéxome del Amor, que me ha herido; abrióme el corazón, cerró la boca, ató la lengua, desató el sentido.

Y cuanto más la rabia al alma toca, la paciencia y firmeza van creciendo y la virtud de espíritu se apoca.

De tal manera, que me veo muriendo, sin osarlo decir á quien podría sola dar el remedio que pretendo.

DELIO

Amigo F, aquessa tu porfía tiene de desvarío una gran parte, aunque perdones mi descortesía.

Díme, ¿por qué razón debes guardarte de descubrir tu llaga á quien la hace?

¿ó cómo sin saberla ha de curarte?

FANIO

Porque de L más me satisface que me mate su amor que su ira y saña, y en esta duda el buen callar me aplace.

DELIO

No tengo á L yo por tan extraña, ni entiendo que hay mujer que el ser querida le pudiesse causar ira tamaña. Cierto desdeño ó cierta despedida, cuál que torcer de rostro ó cuál que enfado, y cada cosa de éstas muy fingida.

Aquesto yo lo creo, F amado; empero el ser amada, no hay ninguna que no lo tenga por dichoso hado.

Y si, como me cuentas, te importuna

aquesse mal y tienes aparejo, no calles más pesar de tu fortuna.

Tú no te acuerdas del proverbio viejo: que no oye Dios al que se hace mudo, ni da ventura al que no ha consejo.

FANIO

Pues dame tú la industria, que soy rudo, grossero y corto, y en un mismo grado mi razonar y mi remedio dudo.

Bien que llevando L su ganado por mi dehesa, junto con el mío, me preguntó si soy enamorado.

Y el otro día estando junto al río llorando solo, en medio de la siesta, L llevaba al monte su cabrío.

Y díxome: Pastor, ¿qué cosa es ésta? y yo turbado, sin osar miralla, volvíle en un suspiro la respuesta.

Mas ya estoy resumido de buscalla, y decirle por cifra lo que

siento, al menos matárame el enojalla.

De cualquier suerte acaba mi tormento, con muerte, si la enojo, ó con la vida, si mi amor y mi fe le dan contento.

Veremos esta empresa concluída, venceré mi temor con mi deseo, la vitoria, ó ganada ó bien perdida. ¿Oyes cantar? D. Si oyo. F. A lo que creo, L es aquélla. D. Eslo, F. Al valle viene.

¡Ay, que te busco y tiemblo si te veo!

Ascóndete de mí, que no conviene, si tengo de hablarle, que te vea.

DELIO

Ascóndeme, pastor; Amor ordene que tu mal sienta y tus cuidados crea.

LIRIA

El pecho generoso, que tiene por incierto

serle possible, al más enamorado ser pagado, y quejoso vivir estando muerto, y verse en medio de la llama helado; cuán bienaventurado le llamará el extraño, y en cuánta desventura juzgará al que procura hacerse con sus manos este daño, y por su devaneo á la razón esclava del Deseo. Memoria clara y pura, voluntad concertada, consiente al alma el corazón exento; no viene su dulzura con acíbar mezclada, ni en medio del placer ama el tormento sano el entendimiento, que deja el Amor luego más que la nieve frío, pero el franco albedrío y el acuerdo enemigo, á sangre y fuego; y en tan dañosa guerra, sin fe, sin ley, sin luz de cielo ó tierra.

Promessas mentirosas, mercedes mal libradas son tu tesoro, Amor, aunque no quieras; las veras, peligrosas; las burlas, muy pesadas;

huyan de mí tus burlas y tus veras, que sanes ó que hieras, que des gloria ó tormento, seas cruel ó humano, eres al fin tirano, y el mal es mal y el bien sin fundamento; no sepa á mi morada yugo tan duro, carga tan pesada.

Corran vientos suaves, suene la fuente pura, píntese el campo de diversas flores, canten las diestras aves, nazca nueva verdura, que estos son mis dulcíssimos amores; mis cuidados mayores el ganadillo manso, sin varios pensamientos ó vanos cumplimientos que me turben las horas del descanso, ni me place ni duele que ajeno corazón se abrase ó hiele.

FANIO

Por essa culpa, F, ¿qué merece L? L. Lo que padece; pues, penando, quiere morir callando. F. Gran engaño

recibes en mi daño. ¿Tú no sientes que las flechas ardientes amorosas vienen siempre forzosas? Si de grado tomara yo el cuidado, bien hicieras si me reprendieras y culparas.

LIRIA

Déxame, que á las claras te condenas: pudo Amor darte penas y matarte, y no debes quexarte, pues que pudo; de ti, que has sido mudo y vergonzoso, debes estar quexoso. ¿De qué suerte remediará tu suerte y pena grave quien no la ve ni sabe? F. ¡Ay, L mía! que yo bien lo diría, pero temo que el fuego en que me quemo se acreciente.

LIRIA

Pues, ¿tan poquito siente de piadosa quien tu pena furiosa ensoberbece?

FANIO

Mas antes me parece, y aun lo creo, que tan divino arreo no es posible en condición terrible estar fundado; pero considerado aunque esto sea, no es justo que yo vea mi bajeza, y aquella gentileza soberana, y que sufra de gana mis dolores sin pretender favores. L. Grande parte ha de ser humillarte, á lo que creo, para que tu deseo se mitigue, porque Amor más persigue al más hinchado, que está muy confiado que merece, que al otro que padece, y de contino se cuenta por indino; pero cierto, tú no guardas concierto en lo que haces:

¿no se sabe que paces las dehessas, con mil ovejas gruessas abundosas y mil cabras golosas y cien vacas?

¿No se sabe que aplacas los estíos

y refrenas los fríos con tu apero, y tienes un vaquero y diez zagales?

Todos estos parrales muy podados, que tienes olvidados, ¿no son tuyos?

Pues estos huertos, ¿cuyos te parecen?

Todo el fruto te ofrecen; pues si digo del cielo, ¿cuán amigo se te muestra, tecuánto la maestra alma Natura y dió de hermosura, fuerza y maña?

¿Hay ave ó alimaña que no matas?

¿Hay pastor que no abatas en el prado?

¿Hate alguno dejado en la carrera?

Pues en la lucha fiero ó en el canto, ¿hay quién con otro tanto se te iguale?

Pues esso todo vale en los amores, porque de los dolores no se sabe

si es su accidente grave ó si es liviano.

Todo lo tienes llano. F. ¿Qué aprovecha tener la casa hecha y

abastada, si en la ánima cuitada no hay reposo?

LIRIA

Vivir tú doloroso, ¿qué te vale, si aquella de quien sale no lo entiende?

Tu cortedad defiende tu remedio.

FANIO

¿Parécete buen medio que lo diga?

LIRIA

Antes es ya fatiga amonestarte.

FANIO

Pues, ¿tienes de enojarte si lo digo?

LIRIA

F, ¿hablas conmigo ó desvarías?

¿Pensabas que tenías y mirabas presente á quien amabas? F. Sí pensaba y en nada me engañaba. L. No te entiendo, aunque bien comprehendo que el amante

tiene siempre delante á la que ama, y allí le habla y llama en sus passiones.

FANIO

No glosses mis razones. L. Pues, ¿qué quieres?

FANIO

Hacer lo que quisieres, aunque quiero preguntarte primero: ¿si mis males y congojas mortales me vinieran por ti y de ti nacieran, y el cuidado te fuera declarado, ¿te enojaras?

LIRIA

Si no lo preguntaras, te prometo que fueras más discreto. Tú bien sientes los rostros diferentes de natura en una compostura de facciones; pues, en las condiciones, es al tanto, aunque no debe tanto ser piadosa,

á mi ver, la hermosa que la fea, que en serlo hermosea su fiereza.

FANIO

¡Ay, cuánta es tu belleza! L. Assí que digo, que no debes conmigo assegurarte, pues sé certificarte que en tal caso, aquello que yo passo por contento puede ser descontento á tu pastora, y no imagino agora por qué vía con la voluntad mía quiés regirte.

FANIO

Porque puedo decirte que, en belleza, en gracia y gentileza, eres trassunto, sin discrepar un punto, á quien me pena.

LIRIA

¿Es por dicha S tu parienta? Si es ella, no se sienta entre la gente,

que eres tan su pariente como mío; pueda más tu albedrío que tu estrella.

FANIO

¡Ay, L, que no es ella! ¿Y aún te excusas y de decir rehusas el sujeto que en semejante aprieto mostrarías?

LIRIA

Horas me tomarías si lo digo, que como fiel amigo te tratasse; y horas que me enojasse, que aun no siento mi propio movimiento. F Dessa suerte más me vale la muerte y encubrillo, que al tiempo de decillo verla airada.

LIRIA

Bien puede ser quitada tu congoxa, si aquella que te enoja me mostrasses y en mis manos fiasses tu remedio.

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.