基于神经网络模型的知识获取方法与故障诊断

A Learning Algorithm Based on the Neural Network Model and Its Application in Fault Diagnosis

分析了神经网络模型的应用环境，提出了一种ＢＰ模型的快速算法，并将其应用于ＦＭＳ故障诊断系统。实践证明，这种具有柔性节点的自适应ＢＰ网络模型结构为建立诊断系统知识库提供了一条有效途径。

The BP network model for an artificial neural network(ANN)is ineffective in dealingwith real-time promems with strict requirements owing to the large amount of calculatingwork and possible local convergence,An improved BP network model is proposed as a reme-dy. Generally speaking,it is difficult to determine in advance a suitable network structurewhen a hierarchical neural network is used for a spectific problem;For this reason,a conceptof flexible hidden nodes is presented and an adaptive learning algorithm is developed to com-press the redundant hidden nodes into a suitable network size. It is proved that this improvedalgorithm is an effective approach to building the knowledge base for att FMS fault diagnosissystem.