Research On Transformer Condition Evaluation Method Based On Differentiation Threshold

Real-time monitoring of the transformer’s operating status and ensuring its reliable operation is of great significance to the safety of the power system.To accurately evaluate the transformer status,relevant power departments attach great importance to the problem.With the continuous expansion of the scale of the smart grid system and the continuous improvement of sensing technology,the state detection data of power transformers shows the characteristics of rapid growth,laying a good foundation for the application and development of data analysis technology in the field of power transformer state evaluation.Since there are many abnormal values in the collected transformer data,which will affect the evaluation process of the transformer status,data cleaning is very important to improve the quality of the data.In terms of repeated data detection,based on the traditional MPN algorithm,an adaptive step size and sliding window method are designed to reduce the redundant detection of non-repetitive data due to the fixed window size and reduce the missed detection rate.In the aspect of abnormal data detection,an unsupervised ISODATA clustering method is proposed to establish a data anomaly detection model,combined with gray correlation analysis to obtain the correlation between state quantities,and perform targeted noise value cleaning on the data.In terms of filling in missing values,time series methods are used to make corrections to ensure data integrity.The processed data can show the characteristics of real data very well and provide a reliable data set for subsequent state evaluation.Aiming at the problem that the threshold based on characteristic parameters reflects the average level of similar devices and cannot reflect the differences between devices,this paper proposes a method for calculating the membership degree of single state variables based on dynamic differentiation thresholds.The Weibull model is established based on the historical data after the cleaning is completed,the attention,abnormal,and severe thresholds are obtained,and the dynamic threshold theory is introduced to solve the defect of poor adaptability of fixed thresholds.Finally,the degraded membership function is constructed based on the threshold to obtain the membership of the evaluation index state.In the process of empowering evaluation indicators,if only subjective analysis or objective analysis is used,it will be restricted to a certain extent.Therefore,this paper proposes a game theory weighting method based on analytic hierarchy process and entropy method to calculate the confidence probability of equipment in different states,comprehensively considering the subjective and objective weight effects,to achieve the determination of the comprehensive weight of the power transformer state and the multiple parameters of the equipment state assessment.Finally,a sample analysis based on field test data verifies the effectiveness and accuracy of the method in this paper.Compared with the traditional method,it can achieve the differentiated state assessment of the transformer in a more refined manner.