农网变电站安防视频检测终端研究与实现  

Research and Implementation of Alarm Terminal for Security Prevention in Rural Substations Based on Video Technology

作　　者：李强 赵峰 庄莉 张晓东 宋立华[2] 

机构地区：[1]国网信息通信产业集团有限公司,北京 [2]福建亿榕信息技术有限公司,福建 福州

出　　处：《计算机科学与应用》2023年第12期2314-2323,共10页Computer Science and Application

摘　　要：农网变电站大多位于偏远地区且安全管理薄弱,易遭受外来人员非法入侵或无意闯入,从而影响供电安全和人员生命安全。本文提出了以ESP32为核心的公共安全视频检测终端方案,基于帧差法和OpenCV实现入侵检测与合法人员的筛选。实验表明,4米范围内的入侵检测和30˚内的人脸识别精准度都达到了100%。相比传统视频监控系统,本文所提出的方案成本不足百元,且具备合法维保人员自动筛选能力,降低了系统的误报率,为成本受限状况下的农网变电站安全运行提供保障。Most rural power grid substations are located in remote areas and have weak security manage-ment, making them susceptible to illegal or unintentional intrusion by outsiders, thereby affecting power supply safety and personnel life safety. This paper proposes a ESP32-based public security video detection terminal scheme, which implements intrusion detection and legitimate personnel screening based on frame difference method and OpenCV. The experiment shows that the accuracy of intrusion detection within a range of 4 meters and facial recognition within 30 degrees has reached 100%. Compared to traditional video surveillance systems, the proposed solution has a cost of less than ¥100 and has the ability to automatically screen legal maintenance personnel, re-ducing the false alarm rate of the system and providing security for the safe operation of rural pow-er grid substations under cost constraints.

关 键 词：安防检测 人脸识别 帧差法 OPENCV 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]