改进灰色聚类熵法在电力变压器故障诊断中的应用

Application of the Improved Grey Clustering Entropy Method to the Diagnosis of Transformer Faults

为了提高传统油中溶解气体分析(DGA)的诊断能力,解决传统灰色聚类分析中指标权重确定存在的问题,引用熵权原理并加以改进,通过构造梯形白化权函数对变压器故障类型进行分析和判断。该算法有两点改进:①将诊断模型数据的单一标准值优化为范围区间,考虑了相邻灰类界限的相互影响;②通过熵函数确定权重,避免了主观因素的影响,使评价结果更加客观,准确。通过实例介绍了该方法的具体应用,并与三比值法相比较,验证了该方法具有一定的有效性和实用性。

In order to enhance the ability of conventional dissolved gas analysis（DGA） and overcome the disadvantages of the index weight when a conventional grey clustering analysis is made.Entropy weight is introduced into the method and is improved.Transformer faults are analyzed and judged by constructing trapezium whiting function.There are two improvements on the algorithm.Firstly,a single standard value of diagnosis model＇s data is replaced by interval rage,considering interactional relationship of adjacent grey kinds.Secondly,weight is determined by entropy function,which avoids the effect of subjective factors and makes the evaluation results more objective and accurate.The application is introduced by examples.