基于混沌神经网络的电机故障诊断

Fault Diagnosis of Motors Based on Chaotic Neural Network

采用全局耦合混沌神经网络模型,每个神经元的动力学行为由反对称立方映像表示。采用Hebb算法设计网络的连接权值矩阵,将记忆模式的回忆过程转化为耦合系统中参数演变的过程,从而实现了混沌神经网络的联想记忆。根据提出的能量击穿规则,扩大了样本的吸引域。在此基础上,应用该混沌神经网络对异步电机转子断条故障进行诊断。结果表明,该种方法有助于故障模式的记忆和重现。

?A chaotic neural network is made up with global coupled model, and a symmetric map produces each unit's chaotic motion. The interconnected matrix is designed by means of Hebb algorithm; dynamic associative memory of the chaotic neural network can be realized if the parameters values properly in coupled model are controled. It also can enlarge the attraction domain of samples by using the energy broken rule. Based on the proposed network and rule, the fault of three phase induction motors with broken bars is diagnosed. Diagnose results suggest that the chaotic neural network is beneficial to dynamic memory retrieval and faults identification.