粒子群算法优化神经网络的异步电机转速估计

Asynchronous Motor Speed Estimation Methods Based on Neural Network Optimized by PSO

在异步电机的矢量控制系统中,电机的转速检测是必不可少的,并且转速检测的精度直接影响磁场定向的准确性。讨论了各种无传感器速度辨识方法的特点,利用BP神经网络对异步电机转子转速进行辨识,通过粒子群算法优化使BP神经网络获得更好的网络初始权值和阀值,在此基础上利用Matlab/Simulink建立一个异步电机矢量控制系统,仿真结果表明这种方法能较好地辨识异步电机转子转速,系统具有良好的动态性能,对系统参数变化具有较强的鲁棒性。

In vector-controlled system of and the detection precision affects the sys schemes for sensorless estimation of motor asynchronou tern g speed reatly were d s motor, detection for speed is essential, In this paper, the features of common iscussed and a method to estimate motor speed by neural network was propested in which particle swarm optimization（ optimize initial weights and thresholds of the neural network. Furthermore, system of asynchronous motor was designed by Matlab/Simulink. The simul that the method estimated motor speed, exhibiting good dynamic performance ness to the variation of motor parameters. PSO） was used to a vector-controlled ation results show and strong robust