T-S Fuzzy Modeling And Control For Nonlinear Network Control Systems

Network Control System (NCS) is a real-time feedback loop control network,connect the sensors, actuators, controllers and controlled object through thenetwork in different geographic locations. The controllers exchange informationthrough the network with sensors and actuators, and thus the controlled objectwould be remote controlled. The main problems of the network control systemare the time delay and packet loss problems existed in the process of long-distance transmission. Nonlinear control system refers to a coupling parameternonlinear system. If we want to modeling or analysis nonlinear network controlsystem, there are three basic questions need to be solve: modeling of nonlinearsystem, modeling of network time delay, and design the controller for nonlinearnetwork control system, these three problems progressive relationship.Fuzzy modeling is a very effective modeling method for nonlinear systemmodeling, which can approximate any nonlinear system by its feature ofpiecewise linear, and can design controller segmented, through fuzzy weightedget fuzzy controller output, and the pieces of T-S fuzzy model is a linear functionform, the theory of linear systems can be used to analysis and design control forthe system. Delays in data loss error by network transmission and other problemscan be approximated by the network time delay, because data loss can be viewedas an infinite delay case. Stability of real-time systems have a great time delaydepending on the size of patience, using Lyapunov stability analysis of thesystem, the maximum allowable time delay system stability can be got. Designthe controller for nonlinear network system requires a controller designed basedon the nonlinear system, and then consideration of the existence of a networktime delay, model predictive compensation control is a good solution. Delaysoften only based on the network and time delay related to the current moment.Markov chains can present data in accordance with time-delay time, and the necessary historical lag, by probability matrix to predict the most likely next timedelays. If delay compensation forecast Markov chain controller be added topredict the delay, a better control would be got.Double inverted pendulum system is a typical nonlinear system and has highreal-time requirements, with a large time lag according patience. Firstly, thedouble inverted pendulum motion would be analysis, choose a different locationto get fuzzy model, resulting in double inverted pendulum fuzzy model, and thenseparately for each sub-model design LQR controller, obtained by the fuzzyweighted fuzzy LQR controller; then Lyapunov function delay stability analysisfor the T-S fuzzy model. Making the double inverted pendulum system is stablefor the maximum allowable time delay is0.01s; Double inverted pendulum in theoriginal T-S fuzzy model by adding [0-0.009]s the controller output random delay,and compared with Markov prediction delays; Finally, no time delay model as theprediction model-the ideal model, as there is, the controller data of ideal modelcombined with controller data of real-time systems by Markov chain predictdelay time as the weighted to prediction output control, optimal the controller ofdouble inverted pendulum system existed transmission delay time.All simulations of the paper are achieved in Matlab2010, including Matlab\Simulink, Matlab\LMI and fuzzy control toolbox, The simulation results are usedto verify the modeling and control algorithm proposed in this paper.