The Application And Research Of Fuzzy Multi-criteria Decision Method Based On Bayesian Network In Fault Diagnosis

With the application of industrial enterprise informatization and the development of science and technology,there are more and more industrial equipment and the structure is more complex.In the industrial production process,it is difficult to avoid the occurrence of faults.If a fault occurs suddenly,it will not only cause economic losses,but also cause casualties,so fault diagnosis technology is particularly important.However,when the fault information is incomplete or uncertain,the difficulty of fault diagnosis will be greatly increased,and even the best time for repair will be missed.At present,in view of this situation,data-driven fault diagnosis technology has developed rapidly.Among them,the application of Bayesian Network in fault diagnosis has been a research hotspot in recent years and the application of fuzzy multi-criteria decision-making methods in fault diagnosis is a new research topic.Therefore,in this dissertation,the transformer fault is studied from Bayesian Network and fuzzy multi-criterion decision-making method.The details are as follows:(1)Research on fault diagnosis methods,including the development status and future research direction of fault diagnosis technology.This dissertation mainly introduces the theoretical knowledge of Bayesian Network and its application in fault diagnosis.In addition,the relevant knowledge of fuzzy theory is introduced and the fault diagnosis method of fuzzy multi-criteria decision-making method is studied deeply.(2)Due to the low accuracy and missing code of the three ratio judgment method in the traditional transformer fault diagnosis method,a fault diagnosis method based on TOPSIS-grey relational degree is proposed by using fuzzy multi-criteria decision making method.In order to obtain relatively standard fault data types,TOPSIS method was used to process each fault type sample.The fault type of the sample to be tested is determined by the grey relational grade method.Experimental results show that the method in this dissertation can further improve the accuracy of fault diagnosis.(3)In order to solve the problem of conditional independence assumption in Naive Bayesian Network,in this dissertation,the degree of correlation between different conditional attributes and class attributes will be used to improve the performance of classification and an improved attribution-weighted Naive Bayesian Network classification algorithm is proposed.In order to ensure the reliability of weights,the combination of analytic hierarchy process and entropy weights is used to weight the Naive Bayesian Network.Then,the improved algorithm is applied to transformer fault diagnosis to verify its feasibility.(4)The redundancy,uncertainty,and fuzziness of the transformer fault samples.In order to make the training samples better trained and learned by the Bayesian network and improve the accuracy of fault diagnosis,this dissertation combining Bayesian network and fuzzy multi-criteria decision-making method,then a fault diagnosis method of TOPSIS and gray correlation degree of Bayesian network is proposed.The fuzzy theory is used to process the fault samples,and then the Bayesian network is used for fault diagnosis,which can quickly and accurately determine the fault type of the transformer,so that it can operate safely and stably.