Neural Network Control For Markov Jump Stochastic Nonlinear Systems Based On ELM

Markov jump system is a special class of switching system.The switch process between the different subsystems of the entire system is random,and the switching process follows the Markov process.Markov jump system has broad application prospects in the fields of power system,electronic control,locomotive control,network control and engineering control.Because of the random interference widely exist in scientific research and engineering application,and the nonlinear system also has a wider scope of application,the Markov jump stochastic nonlinear systems get attention by scholars because of its applicability.Therefore,the study of Markov jump stochastic nonlinear system is a problem which is worthy of discussion,which has profound practical significance and theoretical value.The main contribution of this thesis is to introduce a single hidden layer feed-forward neural network trained by ELM(extreme learning machine)algorithm into Markov jump stochastic nonlinear systems.We design the controller and adaptive control law by using the single hidden layer feed-forward neural network which trained to ELM algorithm,Backstepping method and Lyapunov function,so as to the stabilization and adaptive tracking problems are solved for the Markov jump stochastic nonlinear system.The main work of this thesis includes:1?The neural network control scheme based on ELM algorithm is proposed to solve the stabilization for Markov jump stochastic nonlinear systems.The single hidden layer feed-forward neural network is used to compensate the unknown nonlinear part of the stochastic nonlinear system.The hidden layer node parameters of neural network are trained by ELM algorithm.Through the Backstepping method and Lyapunov function method design the controller and adaptive control law for stochastic nonlinear system,which under this circumstance the Markov jump stochastic nonlinear system is still asymptotically bounded in probability,so as to realize the stability of the system.2?The neural network control scheme based on ELM algorithm is proposed to solve the tracking problem of Markov jump stochastic nonlinear systems.The single hidden layer feed-forward neural network trained by ELM algorithm is employed to approximate the unknown nonlinear functions in the systems.By using Backstepping method and Lyapunov function method,the controller and adaptive control law are designed to guarantee such the Markov jump stochastic nonlinear system is asymptotically bounded in probability and track the desired reference model.