Research On The Fault Diagnosis Method Of Substation Based On Improved Bayesian Network

The fault diagnosis technology of substation is based on intelligent diagnosis method.At present,intelligent algorithms are used mostly in the analysis of fault causes.After the failure,a large number of alarm information will be sent to the dispatching center in a short time to generate redundant information,and the computation workload of the large amount will exceed the processing capacity of the operator.The fault diagnosis technology of substation can not only help dispatchers to find fault components quickly but also the premise of quick recovery of power supply.Actually using Bayesian network model to guide the substation diagnostic work is a probability problem,first set again in the prior probability of each component failure or failure symptom as evidence of input to the model of probability reasoning,find a posteriori probability failure mode and failure cause of the highest.If troubleshooting in this direction finds fault root and solves it,the function of the system is restored and the diagnosis result is accurate.If any troubleshooting is not found in accordance with the direction of fault source you need to find from other failure modes and failure causes of posterior probability is higher,for troubleshooting again until you find fault source to solve problems.Will be available in the cause of the problem,failure mode and failure affect the causal relationship between a series of accurate and complete records formed a breakdown maintenance data,also can provide staff with a positive judgment.Aimed at the limitation of the amount of nodes in the substation fault model,using principal component analysis(PCA)is used to simplify the fault feature,reduce the input dimension of the fault features,so as to reduce the size and complexity of diagnosis system,after reduction Bayesian network fault model.Using only partial models related to the symptom of failure can be inferred,which will not affect the result of model inference.The Bayesian model was established for the optimized data to improve the diagnostic efficiency.Therefore,the combination of Bayesian network and principal component analysis can solve their own shortcomings and give full play to their advantages.In substation fault diagnosis by using principal component analysis(PCA)has the advantage that only need to use a few conditions,USES concise greatly improving the efficiency of diagnosis rules: to avoid the main component is that if a fault information is too big,in the case of data screening of the disadvantage of a lot of work.Bayesian fault diagnosis can't identify fault component fault type of problem,put forward Bayesian classifier based on S transform substation fault identification method,element type for the staff to provide further diagnosis basis.S transform has good time-frequency analysis capability,which can accurately extract the key characteristic information of the mutation signal.Using S transform to analyze the time frequency of voltage sag signal and extract all kinds of transient characteristics;The characteristics were trained and identified by Bayesian classifier.Through the simulation example,this method can effectively identify the voltage sag source and can be applied to the substation fault diagnosis system.Choose Bayesian classification is used to solve the problem of substation fault element type recognition,substation fault element type recognition is an important part of the whole power system fault analysis,Bayesian fault diagnosis method proposed in this paper is of further reasoning,for safe,stable and reliable operation is of great significance.