基于D-S证据理论信息融合的电力设备故障诊断研究  

Research on Fault Diagnosis of Power Equipment by Information Fusion in D-S Evidence Theory

作　　者：李昊 Li Hao(Three Gorges New Energy Quyang Power Generation Co.,Ltd.,Baoding Hebei 073100,China)

机构地区：[1]三峡新能源曲阳发电有限公司,河北保定073100

出　　处：《现代工业经济和信息化》2024年第12期280-282,共3页Modern Industrial Economy and Informationization

摘　　要：传统的故障诊断方法在处理电力设备故障的复杂性和不确定性方面存在局限性。基于此,对BP网络、PNN概率神经网络等四种常见的神经网络模型进行了对比研究,以变压器绝缘故障诊断为例,对比了其在故障诊断准确率方面的优劣,发现PNN概率神经网络准确性最高,但仍未超过90%,且不同模型在处理故障问题时的表现也不尽相同,存在一定局限性。提出采用D-S证据理论进行电力设备故障诊断,以水利水电系统中的电力设备为研究对象,构建了信息融合的功能模型,其通过对不同神经网络模型判断结果的信息融合,提高故障诊断的精确性和可信度。最后通过算例展示,以三种不同的诊断方法来检测某一电力设备两种可能的故障A和B,通过D-S证据理论的信息融合,故障诊断可信度从原本的最高70%提升到了90%。Traditional fault diagnosis methods have limitations in dealing with the complexity and uncertainty of power equipment faults.Based on this,four common neural network models,such as BP network and PNN probabilistic neural network,are compared.Taking transformer insulation fault diagnosis as an example,the advantages and disadvantages in fault diagnosis accuracy are compared,and it is found that the PNN probabilistic neural network has the highest accuracy,but it is still more than 90%,and the performances of the different models are different in dealing with faults,which has certain limitations.We propose to adopt D-S evidence theory for power equipment fault diagnosis,and take the power equipment in water conservancy and hydropower system as the research object,and construct the functional model of information fusion,which improves the accuracy and credibility of fault diagnosis through the information fusion of the judgment results of different neural network models.Finally,through the example,three different diagnostic methods are used to detect two possible faults A and B of a certain power equipment,and through the information fusion of D-S evidence theory,the credibility of fault diagnosis is increased from the original maximum of 70% to 90%.

关 键 词：神经网络模型 故障诊断 D-S证据理论 信息融合 

分 类 号：TM507[电气工程—电器]