Multiple Models Adaptive Control Based on Time Series for a Class of Nonlinear Systems

Abstract

For a class of nonlinear discrete time systems,a multiple models adaptive controller(MMAC) based on time series is proposed.It uses clustering method to establish some linear local models,utilizes time series and directional derivative to establish a weighted model to approximate the real system when its working point jumps abruptly,and adds one global adaptive model with a re-initialized adaptive model to get the multiple models.Then a switching mechanism is designed to select the optimal controller to realize the control.Finally,in the simulation result,it can be seen that the proposed controller not only improves the transient response and speed up the control effect,but also reduces the number of the multiple model sets greatly,especially for a similar control response.