Research On Power Transformer Condition Assessment Method Based On Mahalanobis Distanc

Since 2020,China has achieved the goal of an overall well-off society.With the steady development of the domestic economy,power industry technology has become an important component of China’s industrial development.Ensuring the efficient operation of power systems and reducing losses caused by failures has become crucial.Therefore,establishing a reasonable health management scheme and effectively evaluating and predicting the status of transformers has become one of the current research hotspots.Health status assessment is a key link in transformer condition-based maintenance.However,existing online monitoring systems for dissolved gases in transformer oil can judge the normal or abnormal status of transformers based on gas concentration within a monitoring cycle,but there are problems such as fuzzy boundaries and inaccurate judgments of the evaluation results.This paper proposes a method of transformer health assessment based on Mahalanobis distance,which provides a reference scheme for dynamic short-term assessment of power transformers.Firstly,the initial feature set is constructed using the dissolved gas data in transformer oil collected by the oil chromatography online monitoring system.Referring to the evaluation methods in DL/T 1685-2017 "Guidelines for the Evaluation of the Condition of Oil-immersed Transformers(Reactors)",the attention values when each feature may produce defects are used as the preliminary judgment criteria.Next,based on the features collected from the online monitoring data of transformer oil chromatogram,this paper introduces the Mahalanobis-Taguchi System to calculate the Mahalanobis distance,construct an orthogonal table,calculate the signal-to-noise ratio and its mean pair,and extract key features to prevent error effects on subsequent transformer status evaluation.By using the signal-to-noise ratio,features that have a positive impact on the evaluation can be selected and retained,while features that have a significant negative impact can be deleted.Finally,a health index model for a single transformer is constructed using the Mahalanobis distance calculated from the dissolved gas data in all oils of a single transformer,using the Box-Cox transformation and 3 The criteria determine the alert values and thresholds for all transformer HIs;Taking an actual situation as an example,it is verified that the proposed method can reflect the dynamic changes in the operating status of transformers and can reliably evaluate the status of transformers.It avoids the problems of subjective selection of parameters and weights in the health index,fuzzy boundaries,and easy to cause misjudgments,and can provide a decision-making basis for transformer condition-based maintenance strategies.