基于机器视觉的电力工程质量智能监控方法研究

Research on intelligent monitoring method of power engineering quality based on machine vision

传统的电网基建工程通常采用光学摄像头进行监控拍摄,并由人工完成分析。针对该方式易受雨雪和环境光照的影响且无法准确判断线路铺设质量的问题,提出了一种基于机器视觉的电力工程质量监控方法。该方法使用双目摄像系统代替光学摄像头来完成输电线路的深度图像采集,采集到的图像中所包含的信息量大,硬件铺设成本也较低。同时使用YOLOv3深度框架对图像进行特征识别与分析,其能够自动、准确地识别输电线路的各个部件。实验结果表明,文中所设计的监控系统拥有较高的准确率,相比于其他对比算法,线路小部件的识别准确率有明显的提升,且算法具有良好的收敛性,训练速度快、效率高。

Traditional power grid infrastructure projects usually use optical cameras for monitoring and shooting,and are manually analyzed.Aiming at the problem that this method is easy to be affected by rain,snow and ambient light,and can not accurately judge the quality of line laying,a power engineering quality monitoring method based on machine vision is proposed.This method uses binocular camera system instead of optical camera to complete the depth image acquisition of transmission line.The collected image contains a large amount of information and the hardware laying cost is low.At the same time,the YOLOv3 depth framework is used for feature recognition and analysis of the image,which can automatically and accurately identify each component of the transmission line.The experimental results show that the monitoring system designed in this paper has high accuracy.Compared with other comparison algorithms,the recognition accuracy of line widgets has been significantly improved,and the algorithm has good convergence,fast training speed and high efficiency.