Research On Online Insulation Fault Diagnosis And Prediction For Traction Transformers

Traction transformer is one of the key devices in traction power supply system , whose reliability directly concerns the safety of the whole railway power supply. The traditional off-line oil meteorologic chromatogram method is mainly applied to judge its running status. This preventive method is not able to detect the fault nowadays, due to its shortcomings such as complexity of processing, time costing and so on. It is an urgent problem that how to find out the fault status in transformer, how to control the system effectively and manage the information safely, using the oil meteorologic chromatogram analysis method.Combined with the advantages of the oil chromatogram analysis and the on-line monitoring method, the on-line insulation diagnosis and prediction system for traction transformer is able to monitor the dissolved gases in the oil and analyze it without stopping the devices. As a key componet, the insulation diagnosis information system is used to control the oil chromatogram on-line monitoring system, display the chromatogram result in real time and process the oil chromatogram chart data. According to the requirement of the "Guide to the analysis and the diagnosis of gases dissolved in transformer oil", a series ways of gas analysis and fault diagnose are developed based on the conventional IEC three-ratio method. Weighted GRA method is introduced and applied to in order to avoid the inherent shortcoming of the traditional three-ratio method. The artificial neural network combined by genetic algorithm is also applied to the power transformer fault diagnosis and its result analyzed, which can predict the faults of traction transformer.The software system is developed on the popular C++ Builder 6.0 RAD(rapid application development) environment, which uses Microsoft SQL 2000 to found its database. The system is running on Windows 2000 platform in order to keep the reliability and the popularity. The Role-Based access control theory is adopted to control the access privilege in order to maintain the data security. The system's running status, monitoring result and the error occurred are analyzed at the end of the thesis.