Research On Condition Assessment And Maintenance Decision Of Distribution Transformer Based On Multi-source Data Fusion

Based on the background of big data in power grid,considering the increasing reliability requirements of distribution network equipment,and based on the analysis of multi-source data in the distribution network,the distribution transformer is taken as an example to construct its state assessment model and decision-making model of condition-based maintenance strategy.This paper mainly completes the following tasks.Firstly,after preprocessing the multi-source data involved in the distribution network equipment,a method of outlier diagnosis combined with the wind-driven algorithm and K-means algorithm is proposed,and the optimal number of clusters is determined by the contour coefficient method.Secondly,based on the processed data,the random forest algorithm based on Tomek links undersampling is used to classify and regress the distribution transformer faults,and the fault state and the estimated outage time under the influence of meteorological conditions are obtained.Thirdly,the state assessment model of distribution transformer is established from the two perspectives of qualitative and quantitative indicators.The cloud model is used to determine the initial state level of equipment.On this basis,the influencing factors of equipment state in external environment,self-health and operation conditions are considered.The calculation method of each quantitative index is put forward,and the comprehensive deterioration assessment result of equipment is obtained,which is the precondition of condition-based maintenance strategy.Finally,the decision-making model of condition-based maintenance strategy is established from the target level,scheme level and decisionmaking level.Considering the two situations of insufficient and excessive maintenance of equipment,the attributes of each index are divided into two categories: reliability and economy.On the one hand,the evidence theory method is used to make the maintenance decision under single attribute or without special preference information.On the other hand,the index attribute optimization model based on satisfaction and consistency is used to make the maintenance decision under group attribute.Based on the massive data sources of distribution network,combined with data mining algorithm and intelligent optimization algorithm,this paper obtains the distribution condition assessment and maintenance strategy model which integrates the external environment,self-health and operation conditions.The assessment method and maintenance decision model proposed in this paper are validated by using distribution equipment and meteorological data in a certain area of China.The results show that the decision-making model can be used as a reference for the scheme selection of distribution transformer maintenance personnel,and have certain engineering significance.