摘要
滚装电梯因高效运载而被广泛应用,但其异常载客行为存在安全隐患。传统检测依赖重量传感器、视频监控、人工巡检,存在实时性差、误检率高、成本大等问题。基于网络通信与服务器架构的AI检测,因受传输延迟与带宽限制的影响,存在检测延迟、数据丢失等风险。提出YOLOv3-R异常载客检测系统,采用高分辨率工业相机采集数据,并结合智能预处理与深度学习识别超载、闯入、违规运输等行为。结果表明,YOLOv3-R在精度、误检率、推理速度方面优于传统方法,可实现高效、低延迟的实时监测,为电梯安全管理提供智能检测方案。
Roll on/roll off elevators are widely applied due to their efficient transportation capabilities,yet abnormal passenger behaviors in these elevators pose safety hazards.Traditional detection methods rely on weight sensors,video surveillance,and manual inspections,which suffer from issues such as poor real-time performance,high false detection rates,and significant costs.AI-based detection utilizing network communication and server architectures is affected by transmission delays and bandwidth limitations,leading to risks of detection delays and data loss.The YOLOv3-R abnormal passenger detection system is proposed,which employs high-resolution industrial cameras for data collection and integrates intelligent preprocessing with deep learning to identify behaviors such as overloading,intrusion,and illegal transportation.The results demonstrate that YOLOv3-R outperforms traditional methods in terms of accuracy,false detection rate,and inference speed,enabling efficient,low-latency real-time monitoring and providing an intelligent detection solution for elevator safety management.
作者
张德萌
刘闯
ZHANG Demeng;LIU Chuang(Dingtao District People's Hospital,Heze,Shandong 274199,China)
出处
《自动化应用》
2025年第16期31-33,共3页
Automation Application
关键词
滚装电梯
异常载客检测
YOLOv3-R
深度学习
目标检测
roll on/roll off elevators
abnormal passenger detection
YOLOv3-R
deep learning
object detection
作者简介
张德萌,男,1987年生,助理工程师,研究方向为电梯控制与智能化。
