Design And Implementation Of Transformer Fault Diagnosis Subsystem Based On The Classification Method

Power transformer is an important competent of energy conversion and voltageconversion and plays an important role in power system.Its steady state is directlyrelated to the security of the entire power system.Thus,predicting in advance thepotential fault type of the transformer is very important. Currently, a large powertransformers are oil-immersed transformers in power system.Transformer willproduce a variety of gases dissolved in the insulating oil in the running course, andthus the transformer potential fault type can be to predict according to the oildissolved gas analysis (DGA). With the rapid development in recent years, the use ofDGA techniques for fault diagnosis is based on various categories methods, such asBayesian classification methods and Case-Based Reasoning classification methods.For these reasons and user application needs, the paper designs and implementstransformer fault diagnosis subsystems based on classification methods.Firstly, this paper introduces the principle of gas generating and theories ofthree classification methods of three ratio methods, Naive Bayesian classificationmethods and Case-Based Reasoning classification methods used by the proposedsubsystem, and describes the use conditions and the use processes of threeclassification methods. Meanwhile, the paper tells the design ideas and patterns andthe necessary development technology used by the transformer fault diagnosissubsystem. These theoretical knowledge lay a solid foundation for the design andimplementation of the transformer fault diagnosis subsystem.Secondly, the paper designs and implements the subsystem based on theclassification methods. The design part includes the data layer design, functiondesign and the detail design of the subsystem.The data layer design mainly includesthe subsystem internal data input, data preprocessing and data output to meet theformat requirements of various classification methods. System design includesarchitecture design and function modules design of the subsystem and lay thefoundation for the subsequent design. In the detail design of the subsystem, mainlyintroduce internal processes of the three ratio diagnostic module, Naive Bayesiandiagnostic module and case-based reasoning diagnostic module, and then introducethe required class diagram of the subsystem.By designing the subsystem, identify theoverall architecture, business logic and the division of functional modules of thesubsystem. The implementation of the subsystem uses C++and MFC framework in accordance with the system class diagram in detail design.Thirdly, the paper tests the subsystem. Because this subsystem is mainly usedfor transformer fault diagnosis, and thus the major testing of subsystem is functionaltesting. The functional testing uses the original gas content collected from insulatingoil for fault diagnosis and tests the subsystem diagnostic process and theeffectiveness and practicability of the subsystem. The test results show that thedesigned and implemented transformer fault subsystem can effectively diagnosepotential fault type of the transformer and provide a reference for maintenancepersonnel. The subsystem meets the needs of the user.