Research On Transformer Insulation Aging Diagnosis And Fault Prediction Based On Large Data Analysis Method

Power transformer is the most important hub for power transmission.With the core mission of energy conversion and transmission,its health and operating status are directly related to the safe operation of the grid.If the transformer fails,it will cause widespread power failure,causing huge economic losses and causing inconvenience to people's lives.In the state grid there are many old transformer have ran more than 20 years,because the running environment,the difference of the loading and other reasons,the actual residual life of the old transformer and aging condition of insulation are differ in thousands ways.A blind replacement would result in huge economic losses,and unwarranted continued operation would create unnecessary security risks.In order to avoid these situations it is necessary to use modern analysis and diagnosis technology in the operation of transformer aging diagnosis and fault forecast analysis research,so as to realize safe and reliable operation of the transformerIn this paper,based on the study of transformer aging mechanism,from the perspective of big data,and using the DGA online monitoring data of a substation in Fuzhou.Firstly,the data cleaning method of double-loop structure is proposed on the basis of iterative test method,and the DGA data is processed.This method can distinguish the abnormal data caused by transformer failure,and can better identify and correct the noise data.Secondly,the chaos theory is used to analyze the chaotic parameters of the DGA data of the transformer aging stage through MATLAB programming.The maximum Lyapunov exponent,Loft entropy and correlation dimension are analyzed to get the insulation aging law.And then based on the maximum Lyapunov index prediction method for DGA time series fault prediction analysis.The results show that the data cleaning method can better identify the fault data and noise data,and can correct the noise data,thus effectively avoiding the loss of information.The analysis of the solidification characteristics of solid insulation shows that the reconstructed phase of CO and C02 The spatial distribution of the spatiotemporal pattern increases with the aging process,the complexity becomes stronger and the chaotic characteristic is enhanced.The maximum Lyapunov exponent and the entropy increase with the aging process,and the insulation aging is more sensitive to the use of this special regularity,a new method of transformer insulation aging diagnosis is proposed.An example analysis proves that this method is effective and feasible.It is found that the prediction method based on the maximum Lyapunov exponent can better reveal the internal evolution of the chaotic sequence,and the prediction accuracy is higher in the maximum predictable time range,and then the DGA intelligent fault diagnosis is used to analyze the prediction data to achieve the maximum prediction of transformer faults in predictable range.