The Evaluation Method Study Of Oil-paper Insulation Aging State Of Power Transformer

The power transformer is the key equipment in the power system,and its state monitoring,fault diagnosis and aging state evaluation are of real and long-term significance whether to carry out the comprehensive state maintenance or to realize the goal of the whole life cycle management of the power equipment in the smart grid planning of our country.The in sulation structure of oil-paper is the main part of transformer insulation,and it is also the most fragile part.Its aging state can basically represent the aging state of the transformer.Studying the aging state evaluation of oil-paper insulation is the key to realize the life prediction of transformer.First,based on the data of dissolved gas analysis(DGA)of the transformer preventive test,the weight analysis of the characteristics of the carbon and oxygen gas is carried out by the improved layer analysis method.On the basis of the comprehensive decision deterioration degree of each characteristic,the insulation aging state of the transformer is divided into seven stages,and the feasibility of evaluating the insulation aging state of the transformer in the seven stages of the gas characteristic amount in the oil is verified by an example.Secondly,according to the CIGRE Method II standard,the internal gas gap discharge model of the typical artificial oil paper insulation is made.At the same time,the quantitative aging stage division standard is formulated,and the rationality of the aging stage division is verified by the test.Design mineral oil-insulation paperboard accelerated thermal aging test and pulse current method to detect the original partial discharge signal in different thermal aging stages.The db8 wavelet soft threshold is used to denoise the original partial discharge signal.Finally,constructe partial discharge experience mode decomposition fractal characteristic parameters.The results show that there are obvious differences between fractal characteristic parameters in different aging stages,and there is a high degree of discrimination.Thirdly,support vector machine and particle swarm optimization support vector machine were used to identify the physical and chemical parameters and PD electrical characteristic parameters at the aging stage respectively.Comparing the recognition results of the two algorithms,whether it is based on the physicochemical characteristics of the dissolved gas in oil,or based on the electrical characteristics of partial discharge empirical mode decomposition and fractal,the recognition effect of particle swarm optimization support vector machine is obviously better than that of the former.Finally,the paper combines the physicochemical characteristics of the carbon and oxygen content in the oil and the partial discharge empirical mode decomposition combined with the fractal electrical characteristics,and uses the D-S evidence theory to fuse the two characteristics.Compared with the results of single characteristic insulation aging evaluation,multi feature information fusion has achieved better results in evaluating the accuracy and reliability of transformer insulation aging state.