一种改进型BP网络在汽轮发电机组故障诊断中的应用  

Application of Improved BP Neural Network in Fault Diagnosis for Turbogenerator Set

作　　者：何成兵[1] 顾煜炯[1] 杨昆[1] 

机构地区：[1]华北电力大学

出　　处：《电站系统工程》2002年第6期55-57,共3页Power System Engineering

摘　　要：在标准BP网络基础上,提出一种加优先权值的改进BP网络模型,给出该网络模型的结构、具体的学习训练算法和反馈算法,并阐述输入层与输出层神经元部分连接的依据及其连接优先权值的确定问题。结合汽轮发电机组故障诊断实例,从单故障识别和多故障识别两个角度证明了该模型具有较强的故障识别能力,其诊断结果也更符合故障实际情况。Based on standard BP neural networks, the improved BP neural network is given with priority power value. The model structure, BP algorithm, learning algorithm and the method assuring priority power value for the neural network are set forth. Combined with the example of fault diagnosis for turbogenerator set, it is proved that the model has a strong ability of fault recognition from two sides, single fault and multi faults, and the diagnosis results are accord with the facts.

关 键 词：改进型BP网络 汽轮发电机组 故障诊断 神经网络 

分 类 号：TM311[电气工程—电机] TP183[自动化与计算机技术—控制理论与控制工程]