基于油中溶解气体能量强度的变压器故障案例推理技术研究

Transformer Fault Case-Based Reasoning Technology Based on Energy Intensity of Dissolved Gas in Oil

针对电力变压器故障诊断既缺乏明确因果关系又需要大量经验的情况,基于范例推理方法建立了归一化能量强度模型。该模型先通过事先确定的分类原则确定出待诊变压器的范例检索空间,在该检索空间内依据预先定义的相似度指标进行源范例的检索;再将源范例CH4,C2H6,C2H4,C2H2的浓度提取出来;最后通过计算得到源范例故障生成气体的能量强度,根据能量强度获得待诊变压器状态的诊断结论。实例分析结果表明:该方法避免了利用每种气体浓度的大小判断故障的繁琐和局限性;对于运行过程中产生的多种结构化及非结构化状态信息进行数据归纳与整理并确定设备状态特征量,建设完备坚强的案例库意义重大。

A normalized energy intensity model is established based on case-based reasoning(CBR) for power transformer fault diagnosis, which needs not only a clear causal relationship but also a lot of experience. The model first determines the sample retrieval space of the transformer to be diagnosed by the pre-determined classification principle, and then retrieves the source sample according to the pre-defined similarity index. Then extract the concentration of CH4, C2H6, C2H4 and C2H2 of the source sample;calculate the energy intensity of the gas generated from the source sample fault;obtain the diagnosis conclusion of the transformer state to be diagnosed based on the energy intensity. As the results of case analysis show, the method avoids the complexity and limitation of judging faults by the concentration of each gas. It is of great significance to summarize and sort out the data of various structured and unstructured state information generated during operation and determine the characteristic quantity of equipment state.