基于神经网络和证据理论融合的水电机组振动故障诊断研究  

Research on fault diagnosis of hydropower generating unit vibration based on neural network and D-S evidence theory

作　　者：李郁侠 刘立峰[1] 陈继尧[1] 张宝芳[1] 

机构地区：[1]西安理工大学水利水电学院,陕西西安710048

出　　处：《西北农林科技大学学报（自然科学版）》2005年第10期115-119,共5页Journal of Northwest A&F University(Natural Science Edition)

基　　金：陕西省教育厅专项科研计划项目(05JK266)

摘　　要：以D em pster-Shafer证据理论为基础,提出了一种神经网络局部初步诊断与证据理论融合决策诊断相结合的水电机组振动故障诊断方法,通过故障征兆信息的有效组合,从不同侧面对水电机组振动故障进行了初步诊断,对每一个子神经网络的输出结果归一化处理后,作为此证据下各种状态的基本概率分配函数,再用证据组合理论融合各个证据信息,得出最终的诊断结果。仿真试验结果表明,诊断结论的可信度显著提高,不确定性明显减少,证明了该诊断方法是有效的。A fault diagnosis method of hydropower generating unit vibration,which is based on D-S evidence theory and combines the way of BP neural network diagnose partially and D-S evidence theory decision-making diagnosis was presented in this paper. The fault of hydropower generating unit from different symptom fields may be diagnosed through effective features combination. The results of each sub-network were normalized as basic probability distributed function for each state of evidence theory and the final diagnostic results could be obtained by fusing each evidence. The simulation showed that the reliability diagnostic results improved while its uncertainty decreased prominently. The validity of this method has been proved significantly.

关 键 词：水轮发电机组 故障诊断 BP神经网络 Dempster Shafer证据理论 

分 类 号：TM312.071[电气工程—电机]