Research On Transformer Fault Diagnosis Based On Tdlas

The voltage level of China's power system is constantly rising.A number of1000 kV high-voltage AC projects have been built.The corresponding transformers are upgraded and the capacity is increased.The failure will reduce the operational stability of the power system,so it is urgent to improve the transformer.The correctness of the fault identification.In this paper,the concentration measurement and fault discrimination of dissolved gases in transformer insulating oil are studied.Based on the tunable semiconductor laser absorption spectroscopy system,the fuzzy C-clustering neural network is used to construct the fault identification system,aiming at improving the transformer fault identification.The correctness.The Lambert-Beer law and wavelength modulation technology used in the traditional fault identification method of power transformer and tunable semiconductor laser absorption spectroscopy system are studied to realize the concentration measurement of various dissolved gases in transformers,and the transformer concentration diagnosis is carried out by using gas concentration data;The solution of dissolved gas detection system in transformer oil is designed,including the overall structure,circuit structure,transmitting and receiving unit and long-path absorption cell,and the selection of measuring system components,including laser,etc,using baseline-free calibration by wavelength modulation spectroscopy The baseline drifts and the characteristic gases C₂H₄,CH₄,C₂H₂ and C₂H₆ are measured.H₂ is proved by the hydrogen sensor to prove the accurate measurement of the characteristic gas.The application of neural network in fault discrimination is studied.A discriminant system based on fuzzy C-clustering neural network is built.The model is trained to obtain the model,and the accuracy of fault identification is improved.The dissolved gas concentration in the transformer insulating oil measured by the tunable semiconductor laser absorption spectroscopy system is input to the fuzzy C-clustering neural network fault discrimination system,and the fault of the measured transformer can be determined.