Study Of Multi-motor Power Balance Control Based On Fuzzy Neural Network

The belt conveyor is an efficient equipment of transporting materials in bulk, it hasbeen applied to mining industry and many other industrial areas because of its manyadvantages. As the enlargement of belt conveyor, its driving manner usually bemulti-motor driving. The important problem needed to be resolved in multi-motor drivingsystem is the power balance of diving motors. Using fuzzy neural network to solve thepower balance of multi-motor has advantages of fast response and high controllingprecision.The thesis analyzed the phenomenon of power unbalance appearing in multi-drivingsystem, and reviewed the principle of adjusting power balance of multi-motor with theworking process of Controlled Start Transmission (CST), and designed the ideal startingspeed curve aimed at the actual starting trait.In this thesis, the traditional multi-motor power balance control strategies werecontrasted and analyzed, and the master-slave mode multi-motor power balance controlsystem was simulated. The simulation results shows that when traditional control methodsof power balance are used, the hysteresis phenomenon of control system is evident, and thecontrol accuracy can still be improved. Because of the defect of traditional methods, thisthesis put forward one kind of multi-motor power balance compensation control strategybased on fuzzy neural network. Fuzzy neural network has both advantages of fuzzy logicsystem and superiority of neural network. The thesis introduced the composition of fuzzylogic system and structure classification and learning style of neural network in detail, andfocused on the BP learning algorithm. Because the learning rate is difficult to set intraditional learning algorithm, a method to automatically adjust the learning rate is adoptedto make the traditional learning algorithm better. This thesis also introduced the method ofdesign power balance controller of multi-motor systematically based on fuzzy neuralnetwork, and used the method of trial and error aim for the problem that the number ofnodes in the hidden layer is difficult to determine. By designing fuzzy neural network, thepressure set value of PI controller of CST can adaptive online and hysteresis of controlsystem can be eliminated and control precision can be improved effectively.In this thesis, the MATLAB software is used to simulate the control method proposedin this paper, and the simulation results show that this method has fast response, strongrobustness and high precision of power balance, and this method can fully meetrequirements of output power balance of multi-motor driving belt conveyor.