基于PCA-ILARS-DNN案例推理模型的电站设备故障诊断方法

Fault Diagnosis Method of Power Plant Equipment Based on PCA-ILARS-DNN Case-based Reasoning Model

针对电站设备故障诊断过程中故障信息提取能力不足导致诊断率低等问题,提出了从历史数据库中建立基于案例推理模型的电站设备故障诊断方法。通过主成分分析(PCA)-改进最小角回归(ILARS)案例表示方法将电站设备故障信息提取为统一形式的数据案例,以此构建电站设备典型故障案例库。在已有案例库的基础上,使用深度神经网络(DNN)算法进行案例检索,实现故障诊断,并通过诊断效果来对比PCA-ILARS算法和其他案例表示方法对故障诊断率的影响。结果表明:PCA-ILARS-DNN算法优于其他常规案例表示方法,对电站设备具有较高的故障诊断率。

A method establishing power plant equipment fault diagnosis model based on case-based reasoning from historical database system was proposed for insufficient ability to extract fault information and low diagnostic rate.Principal component analysis(PCA)-improved least angle regression(ILARS)case representation method was used to extract the fault information of power station equipment into a unified form of data cases,so as to construct the typical fault case database of power station equipment.On the basis of the existing case database,deep neural network(DNN)algorithm was used to carry out case retrieval and fault diagnosis.Through the diagnosis effect,the influence of PCA-ILARS algorithm was compared with other case representation methods on fault diagnosis rate.The result of engineering application shows that the proposed PCA-ILARS-DNN algorithm is superior to other conventional case representation methods,and has a higher identification rate for power station equipment fault diagnosis equipment.