面向电力变压器状态评价的油中溶解气体监测数据补全方法

Method for interpolating monitoring data of dissolved gas in oil for power transformer state assessment

基于马尔可夫理论充分考虑相邻时间点系统在所有状态间的转移特性,提出了一种基于马尔可夫模型的变压器油中溶解气体数据补全方法,将油中溶解气体数据时间序列转化为在不同状态间转移的马尔可夫链,利用正、反向的状态转移矩阵计算得到油中溶解气体数据的补全值。从数据挖掘的角度建立了油中溶解气体数据质量的综合评估体系,从多个角度对数据补全的效果进行评估,并基于D-S证据融合理论融合各个角度的评估结果,得到综合评估结果。利用所提方法对某变压器100组油中溶解气体数据中25组随机缺失值进行补全,结果表明补全后的数据与实际值相似度可以达到99.999%。进一步地,验证其中15组极值点、跃变点处缺失数据补全效果,经过综合评估,补全后的数据与实际值相似度可以达到98.956%。经过验证表明所提方法能够在不改变数据特征的前提下对变压器油中溶解气体的缺失值进行准确的补全,有利于提高变压器状态评估方法的准确性。

Fully considering the system shifting characteristics between all states at adjacent time points,a Markov model based method for interpolating monitoring data of dissolved gas in transformer oil for power transformer is proposed.The time series of data of dissolved gas in transformer oil is converted into a Markov chain that transfers between different states,and the interpolating values of the dissolved gas data in transformer oil are calculated by using the forward and reverse state transition matrix.A comprehensive evaluation system for quality of dissolved gas data in oil is established from the perspective of data mining to evaluate the effect of data interpolation from multiple aspects.Based on the D-S evidence fusion theory,the evaluation results from multiple aspects are combined to obtain a comprehensive evaluation result.25 groups of random missing data in 100 groups of data of dissolved gas in transformer oil are complemented by the proposed method.The results show that the similarity ratio between the completed data and the measured value can reach 99.999%.Furthermore,the data interpolation effect of 15 groups of missing data at extreme points and mutation points are verified.Through comprehensive evaluation,the similarity ratio between the completed data and the measured value at extreme points and mutation points can reach 98.956%.It shows that the proposed method can accurately complement the missing data of dissolved gas in transformer oil without changing the data characteristics,which is beneficial to improve the accuracy of transformer state evaluation method.