基于SOFM神经网络的变压器油中溶解气体的故障分析

Analysis of Dissolved Gas in Transformers Based on Self-Organising Feature Map

油浸的电力变压器无论在正常老化还是故障运行时都会产生低分子烃类,CO,CO2等气体,这些气体将会溶解于油中。这样人们根据油中溶解的气体种类和浓度就可以判断变压器是否有故障、以及故障的种类等信息。但是由于各种故障对应的产气情况十分复杂,至今人们还没有建立它们之间的精确关系。因此,根据变压器油中各气体的浓度以识别变压器故障的模式识别方法应运而生。文章提出了一种基于SOFM神经网络的变压器油中溶解气体的故障分析方法,并实例分析证明该方法的有效性。

Oil-filled power transformers will generate some gases, such as hydrocarbon, carbon dioxide and carbon monoxide. They will be dissolved in oil. People can get the faults information of transformers by analyzing the dissolved gases in oil. But complex relationship exists between the faults and the gass. This paper presents a method based on self-organizing feature map to analyze the dissolved gas. The results indicate that the methods is effective.