基于超像素图像分割的变电设备故障诊断研究

Study on Substation Equipment Fault Diagnosis Based on Super-pixel Segmentation

红外图像是电力大数据中一种典型的非结构化数据,对变电设备的故障诊断有着至关重要的作用。利用SLIC超像素图像分割技术对变电设备红外图像进行分割,通过HSV颜色空间转换后,动态地设定色调(H)阈值,提取出发热故障区域。实验表明,所采用方法能准确快速地提取变电设备发热故障区域,提高了红外图片的分析效率。

Infrared image is a typical unstructured data of big data in power industry, which plays an important role in fault diagnosis of substation equipment. SLIC super-pixel segmentation is used to segment the infrared image of substation equipment. After HSV color space conversion the hue（H） threshold is dynamically set and the heat fault zone is extracted. Experiment result shows that the method used in this paper can accurately and rapidly extract heat fault zone of substation equipment, and the efficiency of infrared image analysis is improved.