基于图像处理的变压器渗漏油检测

Detection of Transformer Oil Leakage Based on Image Processing

在无人值守变电站中,电力变压器作为常用和主要的设备,其运行状况必须能实时监测,当出现异常状况时应给出相应的报警。提出了一种基于颜色特征的变压器渗漏油检测方法,通过差分法检测出异常区域,对异常区域的HS颜色直方图进行分析,当前后图像的黄褐色部分变化较大时可认为其发生了渗漏油异常。实验表明,此方法简单易行,能较准确地检测出变压器的渗漏油异常,并能很好地消除影子对检测结果的影响。

In unattended substations, as power transformer serves as common and major equipment, its operation status should be monitored at all times, and corresponding alarms should be triggered when abnormal situations occur. A detection method for transformer oil leakage based on color feature was presented, in which abnormal area could be detected through difference method, and its H-S Color Histogram could be analyzed. When the tawny part of images before and after changed much, it could be determined that oil leaks abnormally. The experimental results show that this method is easy to carry out, the accurate detection of transformer＇s oil leakage, and it can also well eliminate the influence of shadows on testing.