Algorithm Research And System Design Of Fault Diagnosis For Main Transformer Of Electric Locomotive

As the main power supply equipment of railway electric locomotive,the safety and reliability of the main transformer plays an important role in ensuring the normal operation of the electric locomotive.The research and establishment of the fault diagnosis system for the main transformer of electric locomotive can help workers to diagnose the fault of the main transformer more quickly and accurately,and to work out the scheme of fault treatment in time.What's more,the possibility of main transformer fault occurs in the operation will be reduced.Therefore,it is important for ensuring the safety of railway transportation to study and establish a fault diagnosis system for the main transformer of electric locomotive.With the development of railway transportation,The traditional main transformer fault diagnosis method has been unable to meet the needs of railway maintenance production modernization.Therefore,according to the fault characteristics of the main transformer,the fault diagnosis system was established with the theory of support vector machine based on binary tree and case retrieval.Through this system,the efficiency and quality of locomotive main transformer fault diagnosis and treatment are improved.The main research work of this paper is as follows:(1)This paper introduced the mechanism of the dissolved gas in the transformer oil,and analyzed the relationship between the fault type and the dissolved gas.What's more,the fault diagnosis method of the main transformer based on the analysis of the dissolved gas(DGA)was introduced.In addition,the necessity of developing intelligent diagnosis method for main transformer was demonstrated.(2)In view of the characteristics of the electric locomotive main transformer,which is nonlinear and few samples,the diagnosis model based on the binary tree support vector machine was introduced in the fault diagnosis of the main transformer.In order to reduce the cumulative error of the binary tree support vector machine,this paper used the kernel fuzzy clustering to generate the structure of the binary tree.In the training of support vector machine,K fold cross validation and genetic algorithm were used to search the optimal parameter combination to improve the classification accuracy of support vector machine.The experimental results show that the proposed method is more accurate than the one-against-one support vector machine,one-against-rest support vector machine,BPNN and improved three ratio method.(3)The database of main transformer fault case was designed.Moreover,the methods of case representation and organization were introduced.After analyzing several commonly used case retrieval algorithms,the weighted KNN algorithm was used to retrieve the fault cases of the main transformer,and the process of case retrieval was designed by using multilevel index model.(4)Based on the Microsoft.NET Framework 4.0 development platform and Microsoft SQL Server 2008 database,the software of electric locomotive main transformer fault diagnosis system based on B/S mode was designed by using ASP.net.