DC motor position and speed tracking (PAST) system using neural networks

Abstract

The aim of the thesis is to design and implement an artificial neural network based speed and position control system for a DC motor. The PAST system can drive the DC motor to achieve any arbitrary desired trajectory as well as the speed control. Such tracking is of importance in robotic manipulators. The goal is to achieve accurate control over trajectory especially when unknown load parameters are used. The inverse model of the motor with unknown characteristics is identified through a multi-layer perceptron neural network. The trained neural network is integrated with the concepts of model reference adaptive controller to achieve speed trajectory control. To enhance the capability of the controller to achieve accurate position control, an intuitive feedback module is designed which amplifies the position error through a feedback gain and modifies the speed inputs to the inverse model to achieve accurate position control.