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

作　　者：马富齐 王嘉勋 贾嵘[1] 王波[2] 马恒瑞 MA Fuqi;WANG Jiaxun;JIA Rong;WANG Bo;MA Hengrui(School of Electrical Engineering,Xi'an University of Technology,Xi’an 710054,Shaanxi Province,China;School of Electrical and Automation,Wuhan University,Wuhan 430072,Hubei Province,China)

机构地区：[1]西安理工大学电气工程学院,陕西省西安市710054 [2]武汉大学电气与自动化学院,湖北省武汉市430072

出　　处：《中国电机工程学报》2025年第7期2554-2566,I0009,共14页PROCEEDINGS OF THE CHINESE SOCIETY FOR ELECTRICAL ENGINEERING

基　　金：国家科技部高端外国专家引进计划(G2023041010L)。

摘　　要：大型施工机械是电力检修、改造等作业的重要工具,其近电作业过程的安全距离有严格要求,因此,研究施工机械安全距离智能测量方法对保障本质安全生产意义重大。但其作业过程具有作业范围广、带电环境复杂等特点,现有方法表现出明显局限性。为此,该文提出一种基于双目立体感知和安全区域分割的施工机械安全距离智能测量方法。首先,通过PSMNet立体感知模型获取作业场景的双目视差值,并基于坐标转换的三维重建方法得到作业场景的三维世界空间坐标信息;然后,通过区域元素辨识和Canny边缘检测模型精确识别施工机械位置与安全区域边界;最后,通过监测施工机械与安全区域边界间水平和垂直方向的最小欧式距离变化,实现对大型施工机械作业安全距离的量化及动态预警。实验数据表明,所提方法对于实际施工机械复杂场景的三维空间重建平均误差仅3.3%,对于该场景多元化作业元素的识别精确度达到94.5%,相比现有方法表现出更好的检测性能及实用价值。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 

分 类 号：TM85[电气工程—高电压与绝缘技术]