Research On State Evaluation Of Transformer Based On Weighted Distance Discriminant Method

In the process of power system operation,the normal operation of transformer,which is the one of most crucial equipment in power system,relates to the security and stability of the power system directly.It is necessary to diagnose the fault of transformer in time and formulate the corresponding operation and maintenance strategy.Effective evaluation of the operating status of transformer is the basis for fault diagnosis and formulation of maintenance strategies.At present,there are many deficiencies in the evaluation methods of transformer,such as the one sidedness of evaluation index selection,the subjectivity of the method is strong,practicability of the method is inadequate.Based on the research of distance discriminant method,we construct the state evaluation model of transformer.The model provides a new idea for the state evaluation of transformer and other electrical equipment.Firstly,the importance of the transformer in the whole power system and the necessity of transformer condition evaluation are proposed in this paper.This paper analyzes the significance of state evaluation of transformer by illustrating the state of the electrical equipment evaluation research status at home and abroad through a lot of Chinese literature and English literature.In order to explain the rationality and necessity of using comprehensive evaluation method to evaluate the state of transformer,the advantages and disadvantages of the commonly used methods for status evaluation of electrical equipment are compared.Secondly,this paper analyzes the basic content of transformer condition evaluation and expounds the significance of the state quantity in state evaluation of transformer.Because uniform selection of state quantities is not suitable for the complex regional operating environment of transformer,this paper puts forward a scientific method of state quantity extraction and standardize the state classification of transformer.The state of transformer must be objective,comprehensive and measurable.These studies have laid a solid foundation for the further study of condition evaluation method of transformer.Then,this paper introduces the definition of distance and analyzes advantages and disadvantages of Mahalanobis distance and Euclidean distance.In this paper,the advantages of Mahalanobis distance in dealing with multivariate variable clustering problem are summarized.In order to overcome the shortage that the state quantity is treated equally by Mahalanobis distance,weight of the state quantity is analyzed by comprehensive weight combination method.The initial weight of the state quantity is determined by fuzzy set-valued statistics and further dynamically adjusted by entropy method.The model of transformer state evaluation based on weighted Mahalanobis distance discriminant method is established and expounded.Finally,the feasibility and scientific nature of state evaluation model of transformer proposed in this paper is verified by identifying the status level of sample transformers and instance transformers.On one hand,the feasibility of the model is verified by identifying the status level of sample transformers using the model and the cloud model.On the one hand,by identifying the status level of instance transformers,the influence of the state weight on the discrimination result is verified,also,it is verified that the condition evaluation model of transformer established in this paper can be used to realize the generation of sample transformers,thus forming a new multi-sample status level fusion to identify the status level of other instance transformers.