Fault Diagnosis And Failure Rate Estimation Method For Transformer Internal Latent Fault Based On Energy Weighted DGA

Power transformer is one of the important power transmission equipments in power system. It's in the hub in the power system, so the safety and reliability of its operation is directly related to the security of the power system. To make accurate fault diagnosis of transformer can help ensure the safe and stable operation of the power grid. Transformer latent fault develops for a long time, the effective means of detection is through dissolved gas analysis, DGA. It can detect whether the transformer is in fault condition,and differentiate fault categories and the severity of the fault to judge. In addition, to make accurate estimates of the transformer latent fault failure rate,can help gurantee the safety of transformer and help arrange maintenance an operation activities. This paper focuses on the faultdiagnosis and failure rate estimation of transformer and the main works are as follows:Firstly,the current widely used intelligent methods in transformer fault diagnosis were introduced. As Extreme Learning Machines(ELM) has many advantages, it is mainly introduced and a fault classification model ia established based on ELM. DGA data is used to testify the validity of the model in fault diagnosis, and the results shows that the research in this paper provides a new way for fault diagnosis of transformer.Secondly, the concept of energy weighted DGA is introduced.transformer fault severity of judgments. Introduces the concept of thermodynamic enthalpy, and thus the transformer oil decomposition chemistry analyzes enthalpy of formation of different fault gases are weighted according to the criteria put forward the concept of energy weighted DGA,numerical examples show EWDGA could offer a more objective judgment than the traditional gas increase rate data when determining the degree of fault has. Although the extent of discrimination EWDGA fault shows someadvantages, it is a relatively new concept and more tests are needed.And finally, based on EWDGA and Markov process, taking into account the failure rate is proposed energy-weighted gas yield data transformer latent faults estimation method, and a numerical example, the results show that compared with the traditional method, to better distinction in the same state,the different levels of development of the transformer fault latent faults failure rate.