基于综合关联度分析的电力变压器故障诊断

Power Transformer Fault Diagnosis Based on Comprehensive Correlation Analysis

在电力变压器故障诊断中,针对油中溶解气体分析,传统的三比值法难以包括和反映电力变压器内部故障的所有形态,在实际工作中存在许多变压器故障因查不到故障编码而无法判断的问题。结合油中溶解气体分析技术和灰色关联的相关理论,提出了一种面积关联度和斜率关联度相结合的综合关联度分析方法,给出了变压器故障诊断的算法步骤,并对权重系数的选取进行了探讨。该方法既能表征序列曲线的离散程度,又能反映序列曲线的变化走向相似度,能全面描述序列间联系的紧密程度。实验表明,将该方法用于变压器故障诊断,不仅克服了三比值法存在的问题,而且故障诊断准确率也高于面积关联分析方法和斜率关联分析方法。在收集到的数据中随机选取350组进行计算,诊断准确率达到93.7%。

Aiming at dissolved gases analysis, the traditional three-ratio method is difficult to include and reflect all types of power transformer faults. Actually some faults are unable to be judged because of the absence of fault coding. The power transformer fault system can be regarded as a typical grey system. Based on the dissolved gases analysis and grey correlation theory, a new comprehensive analysis method of grey correlation grade is proposed, which combines grey area correlation degree and grey slope correlation degree. The step of arithmetic for power transformer fault diagnosis is given and the choice of weighing coefficient is discussed. The method reflects the discrete degree between sequence curves the similar degree of changing trend of sequence curves. The experimental results demonstrate that the comprehensive correlation analysis method overcomes the shortcoming of three-ratio method, and it has higher accuracy compared with grey area correlation analysis and grey slope correlation analysis. 350 groups of random data are selected and calculated. The diagnosis accuracy is 93.7%.