Condition Assessment Of Electric Power Equipment Based On Data Mining Technology

Condition-based maintenance strategy for power equipment is gradually replacing the traditional time-based maintenance strategy.Condition assessment,one of the core tasks of power equipment condition-based maintenance,is to determine the condition of power equipment.At present,with the development of condition assessment and the application of power information system,the condition data of power equipment has been collected and stored on a large scale,which provides a good foundation for the application of data mining technology.The condition assessment method based on data mining technology is studied for primary equipment represented by transformer and overhead transmission line and secondary equipment represented by relay protection equipment.A transformer condition assessment method based on SMOTE(Synthetic Minority Over-sampling Technique)and decision tree algorithm is presented.According to the viewpoint of.buckets effect",the transformer is divided into five parts,and the criteria are selected.SMOTE is used to alleviate the category imbalance of the raw sample set.The current condition assessment of transformers is mainly carried out by the way of deducting points.The eondition assessment process of transformer is regarded as a classification process,and the model's output is defined as the condition grade.C4.5 algorithm is selected to construct condition assessment model.Because the model is a white box model,it is used to acquire knowledge,the key criteria and key assessment rules of transformers are extracted from the model.The case study is performed on using the actual data of a city-level power supply company,the results show that the proposed method can build a transformer condition assessment model with good classification performance,and can complete the process of knowledge acquisition.An assessment method for overhead transmission line based on random forest algorithm is presented.The special situation affecting the condition of the overhead transmission line is considered,and the criteria is selected.The criteria of the overhead transmission line are much more than that of the transformer,and random forest is more powerful than a single decision tree in high-dimensional feature processing ability,random forest algorithm is selected to construct the condition assessment model of the overhead transmission line.The case study is performed on using the actual data of a city-level power supply company,the results show that the proposed method can construct a condition assessment model of overhead transmission lines with good classification performance.A condition assessment method and a risk assessment method for relay protection equipment based on data mining technology are proposed,criteria of body and criteria of secondary circuit are both selected.Because the current condition assessment of relay protection equipment usually adopts the comprehensive scoring method,the condition grade is directly corresponded to the condition score.So as the condition process of relay protection equipment is regarded as a regression proeess.BP neural network algorithm is used to construct the condition score prediction model of relay protection equipment,and the output of the model is defined as the condition score.On this basis,a method for assessing the relay protection,s risk based on condition assessment score is proposed.The defect rate-condition score function is introduced to obtain the method of calculating the defect rate from the eondition score,which establishes the relationship between the condition score and the risk assessment.In the case study,the condition prediction model of relay protection equipment based on BP neural network is constructed with actual data,which has high prediction accuracy.Then the risk assessment method is used to analyze the risk of relay protection equipment,which proves the practicability of the risk assessment method proposed in this paper.