Single Neuron PID Control For Ultrasonic Motor Based On RBFNN Identification

As a new kind of direct driving micro-motor, ultrasonic motor(USM) has perfect performance and characteristics such as low speed, high torque, quick response, self-lock while power off, and etc. These characteristics make USM a good prospect in precision control and short-term-continuous movement. But, the inherent complexity nonlinear and time-varying of USM brings difficulties to control the motor. Aiming at the problems above, the paper raises one single neuron adaptive PID control method to USM, which is based on the RBF Neural Network(RBFNN) on-line identification. The ability of single neuron self-learning and self-adapting as well as the process model got from online identification can achieve PID control parameter's adaptive tuning. The main contents of the paper are as follows:Firstly, this paper introduces the USM's history, advantages and application field, as well as the speed regulation method and driving mode.Secondly, a filter is designed. The paper points out the Kalman filter algorithm which is based on M speed-measuring, and verifies the algorithm's effective by simulation experiment. The improved Kalman filter is designed after analyzing the experiment datum, so as to simplify algorithm implement on DSP.Thirdly, the paper studies identification principle and learning algorithm of the RBFNN. The Matlab experiment is used to investigate identification effects between BPNN and RBFNN. The result shows RBFNN is faster and less error than BPNN. Then, a single neuron PID controller based on RBFNN identification is designed to improve USM's self-adaption and response.Finally, according to Modbus communication protocol and Q format programming, the control program of system is designed and verified on USM80 USM platform. The result shows that the single neuron PID control based on RBFNN identification is better than the traditional PID in control, and smaller in speed error.