基于紫外成像技术的电气设备放电故障定位

Based on UV Imaging Technology Locating Discharge of Power Equipment

夜间型紫外成像仪获得的放电图像往往因其拍摄条件及其设备的限制,不能很好的实现电晕放电位置的定位,为此在对紫外图像进行降噪预处理的基础上,分别依据紫外图像灰度值的不同、放电部分的灰度值均高于环境部分灰度值、放电部分与环境部分灰度值相差明显且人眼对黑色敏感习惯以及放电部分边界梯度变化大的原理,采用等位图法、阈值法、反转法和边缘法研究了基于紫外成像技术的电气设备放电故障定位。实验结果表明4种方法均可实现放电定位。通过对结果的对比分析,确定了各种方法的适用情况,此研究有利于基于紫外成像技术的电气设备放电故障诊断的实现。

Because the UV image is shot at night and limited by equipment, the location of corona discharge can not be found easily on the image. The locating methods of corona discharge are researched in this paper, which are based on UV imaging processing techniques. Four methods are applied for locating the area of corona discharge, namely, the contour method, the threshold method, the image reverse method, and the edge method. The contour method has a fast computing speed, can realize the comparison between discharge area and un-discharge area, but be easily influenced by image noise. When using this method, the UV image of corona discharge had to be reduced noise at first. The threshold method can decide the size of discharge area, and this size has a direct relationship with the discharge degree of corona. It is important and difficult to select the threshold value in the threshold method with removing the images of electrical equipment themselves. The image reverse method can obviously present the area of larger corona discharge on electrical equipment, but when the corona discharge is smaller, or the gray value of corona discharge approaches electrical equipment, image reverse method will be ineffective. The edge method can extract the boundary between corona discharge and other background, and locate the corona discharge, even when corona discharge is not intense, or the area of corona discharge is small, this method is also effective, but its calculation is more complex than that of other three methods. Through analyzing the different use conditions of these methods, a guide of locating the area of corona discharge in practice use is offered to study electrical equipment fault diagnosis.