Design and Application of Stable Predictive Controller Using Recurrent Wavelet Neural Networks

摘要: This paper presents a design methodology for stable predictive control of nonlinear discrete-time systems via recurrent wavelet neural networks (RWNNs). type controller has its simplicity in parallelism to conventional generalized and efficiency deal with complex dynamics. A mathematical model using RWNN is constructed, learning algorithm adopting recursive least squares employed identify the unknown parameters consequent part RWNN. The proposed law derived based on minimization modified performance criterion. Two theorems are presented conditions stability analysis steady-state closed-loop systems. Numerical simulations reveal that gives satisfactory tracking disturbance rejection performances. Experimental results position positioning mechanism show efficacy method setpoint changes.