变压器故障诊断中信息融合技术的应用

Application of Information Fusion Technique to Transformer Fault Diagnosis

电力变压器的可靠性直接影响供电的稳定,为此结合电力变压器故障诊断的具体情况,从信息融合的角度提出了基于信号类型及不同特征向量组合的集成诊断模型。该模型利用基于故障机理内在因素的专家系统,采用匹配知识规则,通过模糊推理,得特征向量与变压器故障信息间的关联性质,完成故障诊断,且诊断完成后,利用数据挖掘技术将诊断过程中的特征量、推理过程、结果以编码形式存入专家系统规则库。该模型将油中溶解气体、电气参数等特征量结合作为故障判据,评估变压器状态,并利用评估结果丰富专家系统的规则库。在讨论了多传感器信息融合技术用于变压器在线故障诊断时的具体实现方法后通过实例进行分析说明:利用来自传感器的各种故障信息进行融合,作为变压器故障诊断的判据,提高了诊断的可靠性和准确性,这将对提高电网安全性能及电气设备的诊断自动化与智能化水平起到推动作用。

According to the facts of fault diagnosis for power transformer, an integrated diagnostic model based on signal type and composing of different eigenvectors is presented from the viewpoint of information fusion. The model uses the expert system based on the inner factors of the fault mechanism and the knowledge fitting rules to get the relation between the eigenvectors and the fault information through fuzzy illation to accomplish the diagnosis. After the diagnosis, through data mining the value of the eigenvectors, illation process and results of the diagnosis will be coded and saved in the rule library of the expert system. Within the diagnostic model, the dissolved gas-in-oil analysis, the results of conventional electrical tests of power transformers and so on are combined tightly as the fault criterions. The multi-level fuzzy integrated evaluation mode is used to evaluate the status of the transformer, and the evaluation results enrich the rule library of the expert system. The paper also discusses the realization of multi-sensor information fusion technique to the fault diagnosis. Then an example is introduced to show the procedure of the fault diagnosis. The result shows that the reliability and veracity of the diagnosis which adopts multi-sensor information fusion technique are improved. The diagnosis will also contribute to the security of power system, and improve the automatic and intelligent level of the power equipment fault diagnosis greatly.