7709

7181

كارشناسي ارشد

كنترل

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

1391

شش،93ص.: نمودار

ص.ع.به فارسي و انگليسي

مريم ذكري

بهرام كريمي

22/2/92

جواد عسگري، مرضيه كمالي

مهندسي برق و كامپيوتر

ID7181

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

94 Adaptive Control for Nonlinearly Parameterized Periodically Time Varying Systems using Neural Networks Niloofar Ghasemi n ghasemi@ec iut ac ir Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 IranDegree M Sc Language FarsiSupervisor Maryam ZekriAbstractMany control strategies has been made for compensating the uncertainties of nonlinearsystems where uncertain parameters are assumed to be constant or have slow variations Some uncertainties can be caused by time varying parameters or disturbances thatentered the system By considering the fact that conventional adaptive strategies haverestrictions in dealing with time varying uncertainties investigation of an adaptivecontrol for time varying systems is a challenging problem to the control community Thisstudy presents an adaptive neural control scheme for a class of unknown nonlinearsystems with periodic time varying parametric uncertainties The proposed approach usesa novel neural network structure to approximate nonlinear functions with time varyingperiodic parameters that are nonlinearly parameterized The only prior knowledge is theperiodicity of the parameters A periodic adaptation law updates the network weightsover one entire period which uses the states of the system as inputs The stability ofcontrol system is guaranteed and asymptotic tracking convergence of the system isproved First an affine nonlinear system is considered where RBF neural network is usedas an approximator of unknown functions The extension of the proposed approach to thesystems with unknown nonlinear input gain as a function of states and time is alsodiscussed A robust term is used for compensating the neural network approximationerror The proposed adaptation laws achieve desired performance For the next step anonlinear non affine system with time varying uncertainties is studied The existence ofa controller is proved that the stability of the closed loop system has been ensuredconsidering some theorems Then a RBF neural network is used to design this idealcontroller Also a robust term is applied for compensating RBF neural network errorapproximation Proposed scheme ensured stability of the closed loop and tracking errorconverged to zero For the last step we use WNN with adjustable parameters as aapproximator of uncertainties In addition to updating weights using periodic laws scaleand shift parameters are updated by integral laws Desired performances achieve usingWNN as an approximator with less neurons than RBF neural network Some simulationresults are provided to illustrate the efficiency of proposed control scheme in this study inevery section Keywords Time varying uncertainties Adaptive Neural Control Nonlinear affinesystems Radial Basic Function RBF Periodic adaptation Nonlinear non affinesystems Wavelet Neural Network WNN

مريم ذكري

بهرام كريمي

جواد عسگري، مرضيه كمالي