基于贝叶斯网络的水轮机故障诊断研究

Research on Fault Diagnosis of Hydraulic Turbine Based on Bayesian Network

为解决水轮机故障诊断过度依赖专家经验且效率低的现状,利用水轮机故障历史数据和相关专家经验,建立故障树模型,寻求风险隐患和故障诊断之间的映射,通过故障树模型和贝叶斯网络模型的转换,利用贝叶斯网络的反向诊断技术深入研究根节点的概率重要度和敏感性,推理出导致故障发生的水轮机部件及故障原因,实现对水轮机的故障诊断。

In order to solve the current situation that hydraulic turbine fault diagnosis relies excessively on expert experience and has low efficiency, the historical data of hydraulic turbine fault and relevant expert experience were used to establish the fault tree model for seeking the mapping between risk hidden danger and fault diagnosis. Through the conversion of fault tree model and Bayesian network model, the probability importance and sensitivity of root node were deeply studied by using the reverse diagnosis technology of Bayesian network. Hydraulic turbine components and fault causes were inferred to realize the fault diagnosis.