基于云-边协同的电作业人员行为识别算法

打开文本图片集

中图分类号：TP311 文献标识码：A文章编号：1001-5922（2025）06-0172-04

Abstract：With the gradual improvement of the intelligent requirements of the power grid， more and more substa⁃ tions， transmission lines and other scenes are equipped with dedicated UAVs for safety inspection. By carrying dif⁃ ferent processors and sensors to complete the specified tasks，UAV inspection shows the advantages that traditional manual methods cannot match. However，at present，the endurance of the UAV and the performance of the airborne processor are still difficult to complete more complex and real-time tasks，such as the safety behavior identification and early warning of field workers. The essence of the problem is that due to the limitation of performance and cost， the on-board processor carried by the UAV is often difficult to complete high-complexity operations while ensuring real-time performance，such as personnel behavior analysis. Therefore，this paper proposes a two-stage human be⁃ havior recognition algorithm to support on-site security control，and constructs a cloud-edge computing architecture according to the characteristics of the algorithm. This architecture can make full use of the computing resources on the site and further improve the real-time performance of the algorithm on the basis of ensuring the accuracy of be⁃ havior recognition. In addition，the algorithm has been tested experimentally， and the results show the accuracy and effectiveness of the proposed method.

Key words：key points detection；two stages；behavior recognition；cloud-edge collaboration

考虑到电力作业场景的特殊性（复杂性和危险性），现场安全管控往往需要投入大量的资源。（剩余5875字）