基于模糊数学和概率论的变压器故障诊断

为克服现有专家系统人工神经网络和模糊数学等诸多不足,基于前人对变压器故障诊断理论的研究,综合考虑变压器电气试验、油中溶解气体检测分析及观测信息,并引入模糊数学、概率推理和节约覆盖集理论,建立了一个新的变压器故障综合诊断模型。模糊隶属函数利用各专家的语义强度和经验,建立变压器故障性质与故障征兆之间的因果强度关系,再将概率统计和模糊数学相结合,利用节约覆盖集的理论重新筛选故障征兆集,重构相对似然函数。形成的模型能最大限度融入专家的知识,充分考虑了变压器的历史检修记录,使诊断结论更准确、可信。案例分析表明,该模型即使在只有部分征兆的情况下亦能准确诊断,对综合诊断变压器故障性质具有明显的优越性。

Fuzzy Theory and Probability Reasoning Applied for Identifying Power Transformer Fault

Due to the complexity of electric power transformer fault,an improved fault diagnosis model based on the research theories of electric power transformer fault diagnoses by predecessors was introduced.The new fault diagnosis model is based on the fuzzy theory,probability reasoning and parsimonious covering theory.Not only the dissolved gas in oil analysis and electric tests data were used,but also other observed information was taken into account in the fault diagnosis model.Using the linguistic expression and experience of experts,the relationship between transformer fault characteristics and fault symptoms was built up.Then fault symptoms were filtered by the parsimonious covering set theory.Based on the probability reasoning and parsimonious covering theory,the relative probability function was rebuilt.Not only expert knowledge and experience are imported for the process of transformer fault diagnoses,but also the records of tests and repair history was taken into account in the model.Application of the fault diagnoses system shows that the fault diagnoses model can identify the fault characteristic correctly even with some symptoms absent.There was great superiority for power transformer fault synthetic diagnoses.