BP网络模型重构在直接转矩控制中的应用

Direct Torque Control System Based on Neural Network Restructuring Model

直接转矩控制是继矢量控制变频技术后发展起来的一种新型具有高性能的交流变频调速技术。针对基于直接转矩控制的异步电机低速运行时存在较大的电流及转矩脉动问题,提出了用BP神经网络重构直接转矩控制系统的定子磁链观测器模型和开关状态选择模型,并用单个神经网络训练的方法来处理直接转矩控制的复杂运算。仿真结果表明,用此方案构成的系统具有良好的动态性能,并能有效的改善直接转矩控制系统的低速性能。

The direct torque control is a novel high-powered AC frequency control techniques after the converter technique based on the vector control. Aiming at the problems of great current and torque ripple of the asynchronous motor based on direct torque control when it is running at low-speed. This paper uses neural-network to restructuring stator flux observer and status selector model in direct torque control system. And uses the individual training neural-network to cope with the complex calculations. The simulating result shows that the system using the neural-network have good dynamic performance and efficiently improve the low-speed performance of direct torque control system.