Research On Fault Diagnosis Method Of Power Transformer Based On GEP Algorithm

Based on the existing academic achievements and practical accumulation of transformer condition-based maintenance,the relationship between transformer condition assessment and transformer fault diagnosis is further studied,and an effective strategy of condition assessment and maintenance is proposed.It has important academic value and practical engineering significance for condition-based maintenance of transformer equipment,reducing cost and improving operation stability.Therefore,we use classifier to clarify the relationship between gas and fault,and through optimization algorithm to get a more accurate determination of the relationship between power transformer fault and dissolved gas.Gene expression programming(GEP),as an optimization algorithm,has better search range and efficiency,so it is used to optimize the ownership relationship of classifiers.There are three aspects in this paper:Firstly,the principle of gas generation in transformer oil and the principle of transformer internal faults are introduced in detail,and the relationship between dissolved gas in transformer oil and transformer faults is analyzed.The principle,advantages and disadvantages of gene expression programming algorithm and its difference from genetic algorithm are described in detail.Secondly,a transformer fault classification model based on binary classifier is established,and the relevant transformer fault data are obtained by consulting the literature.The fault data are processed and a transformer fault judgment model is built.Thirdly,on the basis of transformer fault judgment,genetic expression programming algorithm is used to optimize,and the corresponding relationship between gas analysis in oil and transformer fault is obtained.This paper presents a new transformer fault diagnosis method.In this paper,the traditional methods related to transformer failure are introduced firstly.For example,IEC triple ratio method,Rogers method or artificial intelligence fault detection method such as fuzzy logic,expert diagnosis and artificial neural network are introduced.Then introduced the relevant principles of dissolved gas analysis and predecessors on the dissolved gas analysis and the failure of power transformers research are introduced.Part of the artificial intelligence algorithm is introduced,mainly introducing the artificial neural network and support vector machine.Theprinciples of the gene expression programming algorithm and the calculation method are introduced in detail.Finally,a transformer fault diagnosis model based on gene expression programming was established.A large number of transformer fault data was collected in the literature to build a binary classifier fault classification model and optimized by using gene expression programming.Finally,an example obtained in the literature and compared with artificial neural network method and fuzzy algorithm to prove the usability of the method in this paper.