变压器套管憎水性及破损工况智能监测系统的研制

Development of intelligent monitoring system for hydrophobicity and damage condition of transformer bushing

针对变压器套管在线监测的现状,为进一步提高变电站的安全性,研制了一种基于计算机视觉的变压器套管智能监测及评价系统。利用远程网络摄像机获取套管的实时图像,用于识别套管的憎水等级和运行工况。给出了该系统的整体结构和工作流程,采用同态增晰的局部直方图均衡化方式进行图像预处理,采用改进的Canny算子检测图像边缘和数学形态学运算修正进行图像分割,采用基于Hausdorff距离的并行算法进行图像匹配,通过改进型BP神经网络构建分类器,实现套管憎水等级和运行工况的识别,最后给出了运行结果与分析。通过实验和现场测试验证了该系统的有效性和工程实用性。

In view of the present situation of the transformer bushing on-line monitoring,a kind of intelligent monitoring and evaluation system of transformer bushing based on computer vision is developed in order to improve the security of the substation. Using the remote network camera to obtain the real-time image of the bushing,which is used to identify the hydrophobic grades and operating conditions of the bushing. The overall structure and working flow of the system are given. The image preprocessing by the local histogram equalization approach using the homomorphic clear realization is presented. Image segmentation is based on improved Canny operator to detect image edge and mathematical morphology operation. The parallel algorithm based on Hausdorff distance is used to match the images,and the classifier is constructed by the improved BP neural network to realize the recognition of the water level and the running condition of the bushing. The operation result and analysis are given at last. Experiments and field tests show that the system is effective and practical.