A neural network sliding mode method for nonlinear motors

Abstract

External disturbance and random noise put forward higher requirements for the stability of motors in daily work. Here, a neural network sliding mode method is showed in this article. First, a new sliding mode surface change rate and a disturbance compensation measure are defined. They reduce the gain value to prevent chattering. Second, to improve the neural network, the adaptive theory is used to update the weights, and a peak-valley measure is designed to limit the output range. Third, the stability of this method is proved by mathematical derivation of a typical nonlinear system. Last, a typical motor model is built on the Simulink, and the influence of external disturbance and random noise on the motor is analyzed. The result shows the effectiveness of this method.