基于YOLOv3和HSV颜色空间的绝缘子低/零值故障诊断研究

Research of Low/Zero Defect Diagnosis of Porcelain Insulator Based on YOLOv3 and HSV Color Space Image Processing Technology

针对绝缘子在运行过程中长期受恶劣环境的影响而导致其绝缘性能恶化的问题,文中提出了基于YOLOv3和HSV颜色空间的绝缘子低/零值故障诊断技术,以判断出其低/零值故障。首先,采用YOLOv3算法构建瓷绝缘子劣化预测模型,从红外图像中提取瓷绝缘子的最小矩形面积。然后,利用K⁃means算法从矩形区域中分割出绝缘子,以避免背景干扰。在此基础上,利用旋转矩形按长轴方向绘制HSV颜色空间特征归一化值,包括H分量的最大值和最小值、S分量的平均值和V分量的最大值,对绝缘子图像进行校正。最后,结合HSV分量与温升的相关性,推导出瓷绝缘子低/零故障的检测判据,完成对绝缘子运行状态的评价,验证了该方法能够准确区分正常绝缘子和低/零绝缘子。

In view of dielectric performance deterioration of insulator due to harsh environment during long term op⁃eration,the low/zero fault diagnosis technique for insulator based on YOLOv3 and HSV colour space is proposed in this paper to judge its low/zero faults.Firstly,the YOLOv3 algorithm is used to construct a porcelain insulator deteri⁃oration prediction model and extract the minimum rectangular area of the porcelain insulator from the infrared image.Then,the insulator is segmented from the rectangular area using the K⁃means algorithm to avoid background interfer⁃ence.Based on this,the insulator images are corrected by plotting the normalized values of HSV colour space fea⁃tures,including the maximum and minimum values of the H component,the mean value of the S component and the maximum value of the V component,in the direction of the long axis using a rotating rectangle.Finally,the correla⁃tion between the HSV component and temperature rise is combined to derive a criterion for detecting low/zero faults in porcelain insulators,completing the evaluation of insulator operating conditions and verifying that the method can accurately distinguish normal insulators from low/zero insulators.