Study Of Multi-motor Power Balance Based On Neural Network

The belt conveyor is the key equipment for colliery transportation because of its good effect in control, Its reliable and smooth motion has important significance for safe normal and efficient production in coal mine. With the long distance and large capacity's transportation. It is the inevitable trend that multi-motor drive one belt conveyor. So power balance for multi-motor driving belt system becomes key problem. Because belt conveyor is multivariable, nonlinear and time varying system, and it have coupling phenomenon between its controlled parameters, till now it isn't recovered for some accurate mechanisms and mathematical model of its parameters. So traditional control method based on accurate model is difficult to obtain the controlling requirement. It is very meaningful for using neural network to keep power balance. When traditional PID methods are used, all parameters coupled each other and control is lagged. So traditional control methods based on accurate model is difficult to obtain the controlling requirement.Because of the defect of traditional control methods, this paper has developed with improved BP neural network for control forecasting of multi-motor power balance, and with positional PID to control the position of ladle bar. Finally the control reach power balance requirements. The design uses the self-learning ability and the adaptability of BP neural network and three-level neural network is adopted that model is 4-12-3. By the control of network obtained ladle bar moves for power balance. Then the moves are the inputs of PID and tracking the ladle bars' postion. In order to achieve power balance of multi-motor, using regulated PID control servo motor. The input of PID based on self-tuning is accomplished and control lagged is erased. The power balance of multi-monitor based on this method can make effect of control becoming accurate and fast.In the control system simulation, MATLAB neural network toolbox and MATLAB/SDVIULINK module are used. The results of simulation show that the control method has the advantages of fast response, small overshoot, small steady state error, and strong robustness. The new control algorithms achieve the aim to power balance.This paper develops BP neural network and digital PID controller based on DSP. On this foundation software and hardware for this control system is designed.