基于加权模糊度时间序列分析的大型变压器故障预报

Prediction of Power Transformer Faults Based on Time Series of Weighted Fuzzy Degree Analysis

分析了变压器油中溶解气体特性及其与故障的对应情况,确定了故障现象模糊集合和故障种类标准模型库(模糊集),根据现象子集对每个故障模型进行划分,形成多个模糊向量集合族;收集了大量故障变压器的历史数据,并用数理统计方法估算故障隶属度函数的参数;采用时间序列灰色预测法估计某时间内各故障隶属度及其发展趋势,对阈值原则进行改进,并根据改进后的阈值原则推断可能发生故障的种类及其发展趋势。最后,用实例验证了该方法的有效性。

This paper analyzes the characteristics of insulation faults caused by gases dissolved in oil. The fuzzy sets of faults phenomena as well as the standard model bases of faults types are then set up. According to the subsets of faults phenomena, several fuzzy vectors are formed. The fault data of many faulty transformers are collected, and the parameters of the faults-relative membership degrees (FRMD) function are estimated by statistic method. The FRMD are predicted using time series analysis technique based on gray theory. An improved threshold value principle is proposed and used to predict faults types and the trend of their development. Test results show that the prediction model is valid.