Application of an improved RBF neural network in sliding mode control system

Equivalent sliding mode control based on RBF neural network uses the traditional gradient descent algorithm to achieve the control function. Because of local minima, training is slow and so on. The algorithm has slow convergence, poor adaptability problems. This paper presents a RBF network based on variable learning rate of W which can be used to equivalent sliding mode control system. Experimental results of the simulation show that the new algorithm has fast convergence and tracking precision. It can effectively avoid the interference caused by unknown divergence, and have a good control of reliability. © 2010 IEEE.