كنترل توزيع شده سوپروايزري در سيستم هاي ابعاد وسيع تحت شبكه

چكيده انگليسي :

Distributed Supervisory Control of Networked Large Scale Systems Mohammad Afshari m afshari@ec iut ac ir Date of Submission August 22 2012 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Persian Supervisors Dr Jafar Ghaisari and Dr Farid SheikholeslamAbstract Uncertainties in the physical model of large scale systems often induce instability andunpredicted costs Thus one of the major challenges in the control of large scale systems besidesstability is the cost paid by the system This cost is often defined in a supervisory quadratic form forthe whole large scale system To simplify the design procedures and reduce the computational burden the large scale system should be decomposed into several lower order subsystems Though with suchuncertainties and distributed computations a method which gives a proper upper bound for the costfunction of the system is necessary This approach is called guaranteed cost control which has theadvantage of providing an upper bound on a given performance index Thus the system performancedegradation incurred by the uncertainties is guaranteed to be less than this bound On the other hand the time delay of information transmission induced by communication networks between subsystemsis a source of instability and poor performance of the system Hence the stabilization andperformance analysis of uncertain large scale systems with time delays has received considerableattention by many researchers In this thesis we consider a class of large scale discrete time systemswith parameter uncertainties The uncertainty is time varying and is assumed to be norm bounded Asupervisory cost function is defined for the large scale system and divided into subsystem functions Using the Lyapunov Krasovski functional technique combined with a linear matrix inequality LMI technique we developed a robust guaranteed cost control for this system via memory less statefeedback which makes the closed loop system stable for all admissible uncertainties and guaranteesan adequate level of performance A delay dependent stability criterion for the existence of theguaranteed cost controller is then derived in terms of LMIs and their solutions provide aparameterized representation of the control A bound also is derived for the defined cost function Theminimum of this bound is then found by using LMIs These LMIs can be easily solved by variousefficient convex optimization algorithms Also in delay dependent stability criterion we have foundthe maximum allowable delay This means that the closed loop system is stable for all delays in thestates lower than this bound Finally a novel model of large scale NCSs fully describing the dynamiccharacteristic of network including delay loss and packet disordering is proposed by introducing theconcept of packet displacement The resulting close loop systems are jump linear systems due to thenewest signals executed subject to Markovian chains which is not found in the existing literature Again scheduling controller is designed for uncertain and certain system based on Markovian theorycombined with LMI techniques such that the closed loop cost function value is not more than aspeci ed upper bound Note that the controller obtained varies according to the delay packet loss andpacket disordering such that the better performance is obtained by the systems In addition numericalexamples are given to illustrate the effectiveness of the proposed method Keywords Supervisory cost function Guaranteed Cost Control Large Scale Systems Linear MatrixInequality Lyapunov Krasovski functional Gain Scheduling Controller

افشاري، محمد

كنترل توزيع شده سوپروايزري در سيستم هاي ابعاد وسيع تحت شبكه

كنترل با هزينه تضمين شده , تابع لياپانوفك - كراسوفسكي , كنترل جدول بهره , نامساوي ماتريس خطي