Rearch On Fault Diagnosis Method Of Oil-immersed Transformer Based On The Neural Networks

The role of the transformer in the power system is transforming voltage,distributing and transmiting electrical energy.The failure of the transformer will directly affect the safety of power production and transmission,so the operating status of the transformer is monitored in real time.In this paper,the fault diagnosis method of oil-immersed power transformer is studied.To design the transformer fault online monitoring and diagnosis system,the method based on the analysis of the dissolved gas composition in the oil is used.Firstly,the fault types and causes of oil-immersed transformers,in this article,is analyzed.Then the relationship between the characteristic gas components dissolved in the oil of the oil-immersed transformers and the types of faults is analyzed.The BP neural network algorithm is used to diagnose and analyze the transformer fault.Aiming at the long training time and poor convergence in the neural network training,the adaptive rate method and the additional momentum method are used to optimize the diagnosis process.The proposed method has been simulated and verified on the MATLAB platform.The simulation results show that the improved BP neural network has greatly improved training time and convergence.Finally,an online monitoring device for transformer faults and an upper computer management system are designed.The fault on-line detection device uses STM32F103C8T6 as the main control chip,which has the functions of gas composition detection,data storage,data display and remote communication.The upper computer management system has functions such as user management,transformer data acquisition,fault diagnosis and data management.The system test results show that the designed system can realize the online monitoring and diagnosis of oil-immersed transformer faults.