基于模糊-灰色聚类的油纸绝缘状态综合诊断

Comprehensive diagnosis ofoil-paper insulation status based on fuzzy gray clustering

变压器油纸绝缘系统在不同测试方法下会得出不同老化特征量,复杂的绝缘状态全面评估存在着一些模糊与不确定问题。因此,建立基于模糊-灰色聚类和组合赋权法的变压器油纸绝缘状态评估模型。首先,建立了多个测试方法下的老化指标评估综合层次图,以层次图为基础,由收集到的变压器绝缘老化指标数据,借助于模糊C-均值聚类(FCM)算法,确定了4级油纸绝缘状态评估老化指标标准状态向量表,同时确定了白化函数阀值;接着,通过主客观意义的组合赋权法给老化特征量赋权,合理地计算出灰色聚类系数;然后,以一级灰色聚类系数作为二级模糊综合评价的隶属度矩阵,搭建出油纸绝缘状态全面评估模型;最后,通过实例验证了新评估方法的有效性与准确性,具有良好的应用前景。

The aging feature quantity of transformer oil-paper insulation system is different under different test methods. Some ambiguity and uncertainties still exist in the comprehensive assessment of complex insulation condition. Thus, the transformer oil-paper insulationcondition assessment model is established based on fuzzy-gray clustering and comprehensive weight method. Firstly,the aging index hierarchical graph based on multiple test methods is obtained. Usingfuzzt C – means clustering（ FCM,Fuzzy c-means）,the four levels of the aging state standard index vector table of oil-paper insulation condition assessment is identified and the threshold of whitening function is determined with the collected transformer insulation aging index data. Secondly,the gray clustering coefficient is calculated by assigning weight to the aging quantity by weighting method combined with subjective and objective sense. Thirdly,a comprehensive assessment model of oil-paper insulation state is built taking the grade-one gray clustering coefficient as the membership matrix of grade-two fuzzy comprehensive assessment. Finally,the validity and accuracy of new assessment method are verified by examples,which shows a good prospect.