Neural Network Based Control of a Two-Mass Drive System

Abstract

In this paper, two-mass drive system is modelled and speed control of the two-mass system is presented. The speed control of the system offers the challenge due to handle torsional vibrations. In the control structure, Particle Swarm Optimization (PSO) based conventional Proportional-Integral-Derivative (PID) controller and single-layer, feed-forward Neural Network controller with back-propagation learning algorithm are proposed. NN control is investigated to show the effectiveness of the control performance compared with the designed PID control. In order to have a fair comparison, PSO method is used to determine the optimum PID parameters and NN controller is designed with online learning algorithm. In the NN learning, back-propagation, which is the most preferred method, is adapted. Simulation studies are performed in different two parts to examine the performance of the proposed controller. In the first part, the controllers are tested for different step references and comparative results of the optimized PID and NN controllers are illustrated. In the second part, the effect of load torque is explored with proposed NN control method. According to the obtained simulation results, it can be seen that the designed NN controller provides better performances without and with load speeds.