| The asynchronous motor at present is more and more used in the field of ac speed regulation, the direct torque control (DTC) technology has many advantages, such as the system is simple and strong robustness, good dynamic response performance and static performance is good, also gradually in the asynchronous motor speed regulation control applications. However, the traditional direct torque control technology also exist some problems, such as torque ripple is larger, the flux observation is not accurate, switching frequency is not fixed, etc., these problems influence the system to further improve the overall performance, limit the application of expanding the range of the technology. Therefore, study how to improve the performance of the traditional system is of great importance, the main task of the subject is combined with the traditional direct torque control technology and neural network technology, achieve the goal of optimization and improve the control performance.First of all, the thesis introduces the development of the direct torque control technology and the neural network, and the research status at home and abroad, then according to the mathematical model of asynchronous motor and inverter, this thesis expounds the theory related to space voltage vector and the core of the direct torque control, analyzed the observation model of asynchronous motor and the implementation scheme of direct torque control system. In order to improve and optimize the traditional control system, the thesis introduces the artificial neural network theory, designs the BP neural network PID control algorithm, using the neural network instead of the switch table to realize the space vector modulation, and neural network was used to optimize the observation model of flux linkage and torque, in the aspect of speed closed-loop control using neural network PID controller instead of traditional PI controller.Direct torque control system simulation model was set up in MATLAB/Simulink platform, and the way of realization of each module is introduced in detail. The three different working process of asynchronous motor are simulated, the simulation results show that the direct torque control system based on neural network can effectively reduce the torque ripple, ripple component of the stator flux linkage locus is also obvious drop, at the same time it has good effect in low speed running, but cannot effectively reduce the overshoot in the speed control, needs to be further optimized. The experimental results show that the design scheme can improve the control performance of traditional direct torque control system, and still has good performance when speed is low. |
