Research On Power Transformer Condition Assessment Method Based On Hybrid Data Processing Method

The power transformer is the core component of the power system.The condition inspection of the power transformer is an important part of maintaining the safety and stability of the power system,and the basis of the condition inspection work is the condition assessment.The state evaluation of the transformer is actually a process of evaluating the health status of the transformer by processing the sample data with an algorithm.However,the characteristics of a single algorithm are not compatible with the transformer condition assessment work.This paper aims at the problems of subjective and objective methods are one-sided,mutual interference between state variables,too much data to processing,and the evaluation process needs to satisfy the problem of fuzziness and randomness,synthesizing the advantages and disadvantages of various algorithms,and at the same time,under the different backgrounds of a small and sufficient amount of transformer sample data,two hybrid algorithms are proposed for transformer state assessment.First of all,this article comprehensively considers the research experience in the State Grid regulations and related documents,establishes a hierarchical state evaluation system,selects the influencing factors related to the transformer state level as the intermediate layer,and selects the most relevant state for different intermediate layer influencing factors.Then,explore the state evaluation in the case of a small amount of sample data.Aiming at the problem of mutual interference between the state variables when the Fuzzy Analytical Hierarchy Process(FAHP)calculates the weight of the transformer state variables,the subjective method of Decision-making Trial and Evaluation Laboratory(DEMATEL)is introduced to modify the weight.In view of the one-sided problem of individual subjective and objective methods,the CRITIC method(Criteria Importance Though Intercriteria Correlation)is selected to calculate the objective weight value,and then the optimal weight is calculated based on the combined weighting method.Then,on the basis of the optimal weight,combined with the state score of each state quantity,the comprehensive state evaluation result is obtained.An example analysis verifies the reliability of the hybrid method,and adds white noise to the experimental data in MATLAB to simulate interference,which proves that the method has better stability.Finally,explore the state assessment in the case of sufficient data.Aiming at the problem of too complex sample data,the complementary capabilities of fuzzy sets and rough sets in solving uncertain problems are integrated,Fuzzy Cluster Analysis(FCA)is used to cluster transformer sample data and combined with Rough Set Theory(RST)calculates the weight value of the state quantity;for the problem of maintaining the fuzziness and randomness of the model during the evaluation process,the multidimensional state cloud model theory is used to calculate the comprehensive certainty of the transformer for each state evaluation level.Part of the transformer sample data is selected for comprehensive grade evaluation examples,and the normal random number generation method is combined with the original sample data for sample expansion.Based on sufficient data,multiple methods are used to compare,verifying the effectiveness of the method.