Direct Torque Control Based On Wavelet Neural Network Research

Direct torque control(DTC) is a control scheme with high performance in modern AC speed regulating system, It has become a hot trend of motor control system. Because of its simple structure, easy to control, improving the rapidity of the system and good dynamic and static performance. Although it has a achieved breakthrough progress for direct torque control of relevant theoretical research and practical application. But the system still exists a lot of disadvantages. But there are still large torque ripple, Magnetic chain round slow convergence speed and not smooth. system control defects such as defective.For motor speed regulation of large torque ripple, slow convergence speed and flux circle is not smooth. system control defects such as defective. We have come up with some solutions to optimize system performance. The article start from the dynamic mathematical model of asynchronous motor, introducing the basic principle of direct torque control. After understanding the traditional torque and flux linkage control. Aiming at the shortcomings of the traditional SVM-DTC is put forward based on the change of stator flux linkage phase Angle of the direct torque control strategy. And We verify improvement scheme by simulation.The stator flux is more close to round.Convergence speed is faster, The system control is more accurate, control effect is better.This article in view of the existing in the direct torque control system of non-linearity,robustness, etc. Using wavelet neural network approximation function ability, wavelet algorithm to optimize the BP neural network structure is adopted to form the speed regulator to achieve the given torque of motor control. Avoid unplanned and BP neural network into a local minimum of faults, effectively improve the operation performance of the system, obtain a more precise torque control system. By comparing with BP neural network PID control, we can get that wavelet neural network better ability of online learning, the error of accuracy is higher. For motor direct torque control system, wavelet neural network structure is more simple, less harmonic torque pulsation, system has faster convergence speed, and the speed overshoot volume is almost zero. The system has better steady-state performance, and the control effect better.Finally, the simulation model is set up in Matlab, the results show that the improved motor direct torque control system with wavelet neural network, no matter in improving the motor performance at low speed, Or in improving anti-interference ability of the system and parameter time-varying adaptive aspects are obvious improvement effect is obtained.