基于神经网络的曳引轮槽磨损测量系统设计  

Design of a Traction Wheel Groove Wear Measurement System Based on Neural Network

作　　者：董灵军 张雍 李琛 邓焜文 周泽丞 Dong Lingjun;Zhang Yong;Li Chen;Deng Kunwen;Zhou Zecheng(Taizhou Special Equipment Inspection and Testing Research Institute,Taizhou 318000;College of Quality and Safety Engineering,China Jiliang University,Hangzhou 310018)

机构地区：[1]台州市特种设备检验检测研究院,台州318000 [2]中国计量大学质量与安全工程学院,杭州310018

出　　处：《中国特种设备安全》2023年第8期54-57,62,共5页China Special Equipment Safety

基　　金：浙江省市场监督管理局科研计划项目“电梯曳引系统定量检测方法的研究和综合检测仪的研发”(ZC2021B101)。

摘　　要：本文针对电梯曳引轮轮槽磨损难以实时定量检测的问题,设计了轮槽深度定量实时测量系统。利用高频高清图像捕获设备对曳引轮槽形貌进行采集,利用神经网络图像算法技术对轮槽表面形貌进行三维重构,同时提取轮槽深度等关键数据并与基准数据进行比较,进而计算得到轮槽磨损量。对5槽曳引轮进行测量实验,测量总体准确率大于97%。实验结果表明,本系统能够实现电梯曳引轮准确、快速的现场检测,为电梯公共安全提供了技术保障。For solving the problem that the wear of elevator traction wheel groove is difficult to detect quantitatively in real time,a quantitative real-time measurement system of wheel groove depth is designed in this paper.The high-frequency HD image capture equipment is used to collect the traction wheel groove morphology,and the neural network image algorithm technology is used to reconstruct the wheel groove surface morphology.At the same time,the key data such as wheel groove depth are extracted and compared with the benchmark data,so that the wheel groove wear is calculated.The experiment on the 5-groove traction wheel is carried out,and the overall accuracy of the measurement is more than 97%.The experimental results show that the system can realize the accurate and rapid on-site detection of elevator traction wheel,and provide technical guarantee for elevator public safety.

关 键 词：电梯 曳引轮 轮槽磨损 定量测量 神经网络 

分 类 号：X924.2[环境科学与工程—安全科学]