摘要
当前电力无人机巡检图像处理模型存在适用范围小、研发成本高、研发周期长等问题,文章提出一种面向无人机巡检图像分析处理的自动化深度学习系统,明确该系统设计的泛化性、可拓展性、自动化三要素,综述与三要素密切相关的技术进展,设计系统架构,构建原型系统。实验表明,在绝缘子自爆识别和鸟巢识别2项电力无人机巡检图像分析处理上,系统自动化构建的模型全类平均精度分别可达91.36%和86.13%,表明系统设计理念合理且系统架构可行。
Current models for analysing images captured by Unmanned Aerial Vehicles in transmission line inspection face limitations in their applicability,high development costs,and long development cycle.This paper proposes a new automated deep learning system,with the key principles of system design being generalisability,scalability,and automation.The literature review of related technological advances and the system architecture design are presented.Experimental results with our prototype system show that the automated model constructed by the system achieved Mean Average Precision values of 91.36%and 86.13%,respectively,in identifying insulator explosions and bird nests on inspection images,demonstrating that the system design is sound,and the architecture is feasible.
作者
李道兴
王晓辉
李黎
季知祥
LI Daoxing;WANG Xiaohui;LI Li;JI Zhixiang(China Electric Power Research Institute Co.,Ltd.,Haidian District,Beijing 100192,China)
出处
《电力信息与通信技术》
2024年第4期38-54,共17页
Electric Power Information and Communication Technology
基金
中国电力科学研究院有限公司青年基金项目“面向能源互联网优化运行的数据存储与计算分析关键技术研究”(AI84-22-002)。
