T-S Fuzzy Model-based Synthesis Approach Of Model Predictive Control And Its Applications In Networked Control Systems

Due to the complexities of actual process control systems,the issues of nonlinearity,uncertainty and the feasibility and stability in constrained predictive control attract much attention in the research field of predictive control.In this dissertation,the Takagi-Sugeno(T-S)fuzzy model and methods of control theory including of robust invariance set,linear matrix inequality(LMI)technique are introduced to develop the robust constrained predictive control algorithms,especially in the basic issues of robusteness,feasibility and online optimization.Based on these results,the design methods of predictive controller for networked control systems is explored.The main work of the dissertation includes the following aspects:1.A robust constrained model predictive control synthesis approach for discrete-time T-S fuzzy system with structured uncertainty is investigated.The fuzzy predictive controller is designed on the basis of non-parallel distributed compensation control law,relaxed stability conditions of the closed-loop fuzzy system are developed by employing an extended nonquadratic Lyapunov function and introducing additional slack and collection matrices.By utilizing the LMI technique,the online min-max optimization problem is parameterized as minimizing the performance objective function in the infinite time horizon at each sampling time.In addition,robust invariance set is constructed to deal with the physical constraints and study the feasibility and robust stability.Simulations on a highly nonlinear continuous stirred tank reactor(CSTR)are eventually carried out to demonstrate the effectiveness of developed theoretical approach.2.In view of a class of linear time-varying systems described by polytopic uncertainty model,which involves all adverse effects of packet loss,network-induced delay and data quantization in the sensor to controller link.Based on the state feedback control law,the network predictive controller is designed by solving the online convex optimization problem in the infinite time horizion to weaken the influence of network data transmission and realize the stable control of closed-loop systems.The simulation results under the different network environment conditions verify the effectiveness of the network predictive control algorithm.3.A constrained model predictive control method for nonlinear networked control systems in the presence of parameter uncertainties and random packet loss over the communication link is investigated.To deal with the parameter uncaetainties and packet loss over the communication link,both of interval type-2(IT2)Takagi–Sugeno fuzzy model and a time homogenous Markov process are utilized to describe the nonlinear plant and model the packet transmission over the controller to actuator link.An IT2 fuzzy predictive controller is designed by minimizing an upper bound on the expected quadratic performance objective function over the infinite horizon at each sampling time such that the resulting closed-loop IT2 fuzzy system is asymptotically mean-square stable.Stochastic invariance set is constructed to deal with the input constraint and study the feasibility in predictive control and asymptotically stability in mean-square sense of closed-loop systems.The simulation experiment results under the different network environment conditions demonstrate the effectiveness and feasibility of the proposed theory method.