旋转机械故障诊断中的主从混合神经网络模型研究

A Study of the Master Slave Neural Network in Fault Diagnosis of a Rotating Machinery

针对ＢＰ网络在旋转机械故障诊断应用中的不足，借助Ｈｏｐｆｉｅｌｄ网络的优良特性，建立了以反馈式Ｈｏｐｆｉｅｌｄ网络为主控网络、前馈式ＢＰ网络为从网络的主从混合神经网络模型。通过这个网络模型的设计、动力学行为分析、学习算法的描述和测试以及它在旋转机械故障诊断中的应用，结果表明：该网络模型具有收敛速度快、稳定性好、最小系统误差等优点，是一种实现旋转机械故障诊断的优良网络模型。

Based on good properties of the Hopfield neural network,a new master slave neural network model is presented,where the master neural network is a Hopfield neural network and the slave neural network is a standard backpropagation neural network.Through the new network model designed, its dynamic property analysed,its training algorithms described and applied to fault diagnosis of a rotating machinery,the results show a number of advantages of this model,such as a quick asymptic convergence rate,good stabability and the smallest network system error.It will be a good neural network model in fault diagnosis of rotating machinery.