利用递归神经网络建立异步电机转速辩识模型

MODEL FOR IDENTIFYING INDUCTION MOTOR SPEED BUILT BY RECURRENT NEURAL NETWORK

本文利用递归神经网络来建立异步电机转速辩识模型，其网络学习采用实时递归学习算法。这种网络具有较强地表达和处理瞬态信息的能力，适用于解决非线性动态辩识与控制问题。仿真结果表明这种转速辩识模型具有良好的响应能力，可以准确跟踪电机转速变化轨迹，特别是当电机负载有较大范围变化时，仍然具有稳定的动态性能。

This paper presents a model for identifying induction motor speed using the recurrent neural network, which is trained by a real time recurrent learning algorithm. The network has the capability of represening and processing for temporal information and is applicable for nonlinear indentification and control problems. Simulation result shows that the speed identification has a better response and can track accurately the varied trajectory of motor speed. Particularly it still has stable dynamic characteristic while the load varies within a big scope.