基于神经网络重构模型的直接转矩控制系统  

Direct Torque Control System Based on Neural Network Restructuring Model

作　　者：蔡斌军[1] 刘国荣[1] 

机构地区：[1]湖南工程学院

出　　处：《微特电机》2009年第10期46-49,共4页Small & Special Electrical Machines

基　　金：湖南省科技计划项目(2008FJ3029;2008CK3082);湖南省教育厅科研项目(09C261)

摘　　要：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时存在较大的脉动问题,提出了用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。实验结果表明,用该方案构成的系统具有良好的动态性能,并能有效地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 technique following the converter technique based on the vector control. To solve the problems of high current and torque ripple of asynchronous motors based on direct torque control when they are running at low speed, this paper uses neural-network in direct torque control system to restructure stator flux observer and status selector model. And the individual training neural-network was utilized to cope with the complex calculations. The experimental result shows that the system with neural-network has a good dynamic per- formance and efficiently improves the low-speed performance of direct torque control system.

关 键 词：直接转矩控制 神经网络 重构模型 低速性能 

分 类 号：TM343[电气工程—电机]