摘要
本文利用人工神经网络对异步电机转子磁链和转速进行直接辨识,多层前传网络具有良好的函数逼近能力,利用误差反传算法对多层前传网络进行训练,使神经网络准确地反映电机转子磁链和转速,在此基础上建立了一个异步电机矢量控制系统,系统仿真结果表明这种方法能较好地辨识电机转子磁链信号和转速,系统具有良好的动态性能。
In this paper, we propose a method for direct identifying the rotor flux linkage and speed of induction motors using Artificial Neural Networks. With the good capability of approximating any function and trained using the Back Propagation algorithm, the multilayer feedforward network can express the rotor flux linkage and speed of induction motors accurately. Based on that we designed a field oriented control system. The simulation results show that this method can identify the rotor flux linkage and speed and the system has good dynamic performance.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2000年第4期478-482,共5页
Pattern Recognition and Artificial Intelligence
关键词
异步电机
矢量控制
无速度传感器
神经网络
Artificial Neural Networks, Multilayer Feedforward Network, Error Back Propagation Algorithm, Rotor Flux Linkage, Field Oriented Control
