Research And Design Of The Controlling System For Sensorless Brushless DC Motor Based On Back-EMF Neural Network Commutation Method

With the improvement of manufacturing level and development of office automation,intelligent,miniaturized and portable are becoming more and more important in people's daily appliances.As important parts of daily appliances,the motors are gaining more and more attention due to their high efficiency,stability,and miniaturization.Compared with traditional Brushless DC Motor(BLDC),Sensorless Brushless DC Motor(SLBLDCM)is being used more widely because its low noise,low cost and good stability.The most widely used commutation method since the invention of SLBLDCM is the zero-crossing detection method based on motor Back Electromotive Force(Back-EMF).This method is simple in design and low in cost.It has a relatively stable control effect when the motor is running at a constant speed.However,when the motor accelerates or decelerates,there will be a deviation between the actual delay time and the ideal delay time,resulting in high power consumption,poor working stability,and even running errors.To solve this problem,based on the algorithm of Back-EMF,this dissertation combines Back Propagation neural network to predict the delay time of motor commutation,and builds a control system of SLBLDCM.In this dissertation,the BLDCM is modeled and simulated,and the best performance network structure is obtained.By comparing the performance of different commutation strategies in the system,it is verified that the proposed method has a large improvement in power consumption,system stability,overshoot and other aspects,which is closer to the ideal commutation situation,with a mean square error of 15.906.To verify the performance of the control strategy under the actual working conditions,this dissertation built a complete control system based on GD32F103 MCU.The incremental encoder is used to realize the precise collection of the working state of the motor,and FPGA is used as the transfer module to upload the data of the motor to PC,as well as to accelerate the calculation of neural network to meet the real-time demand of the system.The experiment results show that the proposed strategy can reduce the power consumption of the system by 11.3%,effectively improve the stability of the system,and the SLBLDCM drive system also achieves a wide motor speed range of 1500-8000 rpm.