On-line Fault Diagnosis And Life Prediction Of Dry-type Transformers

With the rapid development of the social economy,the power consumption of all walks of life has increased rapidly.Based on considerations such as installation and maintenance,power safety,etc.,dry-type transformers have been added to new power consumers in subways,hospitals,airports,office buildings,etc.The proportion of use is getting higher and higher.However,the current research on the fault diagnosis and life prediction of dry-type transformers is still in its infancy.This paper mainly focuses on the research on online fault diagnosis and life prediction of dry-type transformers.Dry-type transformer failures are generally caused by internal windings and iron cores,which are not easy to find directly from the outside.In addition,dry-type transformers generally have a long service life,and it is easy to ignore the loss of transformer life in general inspections.In this paper,the fault diagnosis of the transformer is carried out by the method of vibration,and the life prediction of the transformer is carried out by the method of daily operation data of the transformer.Transformer vibration is mainly caused by winding,iron core and cooling device.The fixed vibration frequency of the transformer body depends largely on factors such as the quality,material,assembly process,and degree of compaction of the transformer body.The method of theoretically solving the transformer's natural vibration frequency can only give a qualitative explanation.Therefore,the method of frequency identification of vibration signals is generally used for fault diagnosis.In this paper,the empirical wavelet transform method is used to identify the vibration signal of the transformer,the knocking experiment is carried out on the transformer under normal conditions,the natural frequency of the transformer is identified,and then the vibration frequency of the transformer is determined according to the transformer vibration under different loads and different parts It is extracted and the fault diagnosis of the transformer is realized by analyzing the abnormal vibration frequency of the transformer.The loss of transformer life is mainly caused by the thermal effect of the transformer.In this paper,the global optimization of the cross entropy theory and the comprehensive characteristics of the combined forecast information are used to establish a dry transformer life prediction model based on the cross entropy theory.Each single prediction method can obtain a sample-based probability density function.Then establish the minimum cross entropy objective function with different probability distributions,take the weight of each prediction method as a parameter,and iteratively solve,and finally obtain the optimal solution to determine its weight.Through analysis of the historical operating data of the transformer,the current transformer is obtained Gives the estimated value of its remaining life and guidance for the next use process.