论文部分内容阅读
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。
Aiming at the shortcomings of BP network in fault diagnosis of rotating machinery, a master-slave hybrid neural network model with feedback Hopfield network as master network and feedforward BP network as slave network is established by virtue of the excellent characteristics of Hopfield network. Through the design of this network model, the analysis of dynamic behavior, the description and test of learning algorithm and its application in the fault diagnosis of rotating machinery, the results show that this network model has the advantages of fast convergence, good stability and minimum system error, It is an excellent network model for rotating machinery fault diagnosis.