Fault Analysis And Diagnosis Of Intelligent Electric Energy Meter Based On Bayesian Network

Safety instrumented systems are widely used in high-risk industries such as petroleum,chemical,metallurgy,etc,and are important equipment to ensure the safe production of enterprises.However,in real life,there are problems of low maintenance efficiency and lack of specific maintenance.This paper takes the intelligent electric energy meter as an example to analyze it and extend to the safety instrument research.At present,the fault tree analysis method with the characteristics of intuitive and causal logic is often used in the study of electric meter faults in China.However,the Fault Tree(FT)model has problems that the model is not easy to change and the model parameters cannot be updated in time.Bayesian network(BN)can solve.BN has the characteristics that the model parameters are continuously updated to predict potential hidden dangers in time,but it is often difficult to model due to lack of data.Therefore,this paper based on BN analysis and diagnosis of electric meters,master its failure mode and failure mechanism,identify weak links and defects,timely diagnosis and prediction,improve maintenance efficiency and maintenance specificity.The specific research is as follows:Through the statistics and analysis of the failure data of the intelligent electric energy meter,the failure probability and the vulnerable key components of each module are obtained,and then the display module,the metering module and the communication module are easy to be faulty.The failure mode and mechanism of each module are analyzed respectively,and the FT is established,provide the basis for conversion to BN.In view of the problem that the intelligent electric energy meter BN is difficult to build,the original historical information is fully utilized to establish the FT and the transformation rules are established according to the logical relationship with the or gate.The information extraction and reorganization method is used to establish theBN in accordance with the actual situation,and the knowledge based on the expert experience is overcome.The shortcomings of modeling subjectivity and low efficiency are completed.Aiming at the problem that the root node information cannot be calculated by the existing data when converting to BN,the method of combining the Delphi method with the trapezoidal fuzzy number is used to provide the scoring principle for the experts in the form of trapezoidal fuzzy numbers,and then the mean area method is used.Turn expert opinions into corresponding probability values.The BN method is used to solve the problem that the FT model structure is not easy to change and the model parameters cannot be updated in time.When the meter fails,the maintenance inspection result is used as new evidence for BN update.If the connection is open,no fault “0” is generated,which can be used as evidence to BN for data update,realizing dynamic update of model parameters,thereby improving maintenance efficiency.Therefore,fault analysis and diagnosis of the meter through BN can dynamically update the model parameters and prevent faults from occurring in time.It also provides important value and theoretical basis for the reliability research of relevant safety instrument systems.