基于递归神经网络的转速估计器仿真研究

Study of Speed Estimators Based on RNN

针对硬件传感器安装、调试、维护复杂等缺点,采用了一种基于递归神经网络(recurrent neural network,RNN)的转速估计器,取代传统传感器完成转速检测任务.递归神经网络采用带遗忘因子的最小二乘(RLS)估计算法,该方法利用RNN强的非线性动态特性,可以在线训练权重,从而可以快速跟踪参数变化、负载变动等情况.最后,通过MATLAB/Simulink仿真验证了此方法的有效性.

In light with the weaknesses of installation,debugging,maintenance and so on of hardware sensors,a new speed estimator based on recurrent neural network is adopted to replace the conventional hardware sensors.The algorithm for recurrent least square（RLS） with forgetting factors is employed in recurrent neural network（RNN）.The weights of RNN can be estimated on-line.It can track the information of the parameter variation and load variation.The effectiveness of the proposed method is verified through the MATLAB and Simulink model.