中国电机工程学报2025，Vol.45Issue(7)：2554-2566,中插9,14.DOI:10.13334/j.0258-8013.pcsee.232097
基于双目立体感知和安全区域分割的大型施工机械近电安全距离智能预警方法研究
Research on Intelligent Early Warning Method for Large Construction Machinery Near-electric Safety Distance Based on Binocular Stereo Perception and Safety Area Segmentation
摘要
Abstract
Large construction machinery serves as a crucial tool for power system overhaul and modification operations.The safety distance during its operation near live electrical equipment is strictly regulated,making the development of intelligent measurement methods for safety distance monitoring essential for ensuring operational safety.However,the operation process is characterized by extensive working ranges and complex electrified environments,presenting significant challenges that existing methods fail to adequately address.Therefore,this paper proposes an intelligent measurement method for safety distance of construction machinery based on binocular stereo perception and safety zone segmentation.First,the stereo disparity value of the work scene is obtained using the PSMNet stereo perception model,while the 3D world space coordinate information of the work scene is acquired through a coordinate transformation-based 3D reconstruction method.Then,by using regional element identification and the Canny edge detection model,the positions of construction machinery and the contours of safety area boundaries are precisely identified.Finally,the minimum Euclidean distance changes horizontally and vertically between construction machinery and safety area boundaries are tracked,allowing us to quantify and dynamically alert for large machinery's operational safety distances.The result shows that this method achieves an average error rate of merely 3.3％for the 3D spatial reconstruction of complex construction machinery scenes,and the recognition accuracy for the diversified elements within these scenes reaches 94.5％.Compared with the existing methods,the proposed method has superior detection accuracy and enhanced practical applicability.关键词
近电作业/电力深度视觉/安全距离测量/双目立体感知/安全区域分割Key words
near electricity work/power depth vision/safety distance measurement/binocular stereo perception/safety area segmentation分类
动力与电气工程引用本文复制引用
马富齐,王嘉勋,贾嵘,王波,马恒瑞..基于双目立体感知和安全区域分割的大型施工机械近电安全距离智能预警方法研究[J].中国电机工程学报,2025,45(7):2554-2566,中插9,14.基金项目
国家科技部高端外国专家引进计划(G2023041010L).National High-end Foreign Expert Introduction Program of the Ministry of Science and Technology(G2023041010L). （G2023041010L）
