基于机器视觉的口岸车道闸机故障远程检测方法
2022,30(3):19-24
摘要：针对口岸车道闸机运行时间的延长，噪声信号会逐渐掩盖真实信号，从而造成信号混合行为的出现，导致口岸车道闸机抬杆机械动作故障检测精度较低的问题，提出基于机器视觉的口岸车道闸机故障远程检测方法。利用CCD传感器，最大化扫描复原口岸车道闸机抬杆机械动作故障信号，并对关键应用镜头设备进行选型处理，完成机器视觉检测的硬件结构设计。输入口岸车道闸机的远程故障图像，按照图像配准原则，得到具体的直方图修正处理结果，拼接与预处理远程故障图像。在此基础上，分析口岸车道闸机抬杆机械动作实际故障特征，通过信号参量非均匀采样的方式，对检测盲源进行分离，再联合故障信号输出信噪比数值，实现口岸车道闸机抬杆机械动作故障远程检测。实验结果表明，基于机器视觉的检测方法的口岸车道闸机抬杆机械动作故障检测准确率可达90.4%，IMF分量值较大，可有效抑制噪声信号对真实信号的覆盖影响，提高口岸车道闸机抬杆机械动作故障检测精度
关键词：机器视觉；口岸车道闸机；闸机抬杆；故障检测；直方图修正；检测盲源；信噪比；
Remote detection method of port lane gate fault based on machine vision
李军
,518045,yhbgv690250@sina.com
,518045,yhbgv690250@sina.com
Abstract：With the extension of the operation time of the Port Lane gate, the noise signal will gradually cover up the real signal, resulting in the emergence of signal mixing behavior, resulting in the low accuracy of the mechanical action fault detection of the lifting rod of the Port Lane gate, a remote fault detection method of the Port Lane gate based on machine vision is proposed. The CCD sensor is used to maximally scan and recover the mechanical action fault signal of the lifting rod of the Port Lane gate, select and process the key application lens equipment, and complete the hardware structure design of machine vision detection. Input the remote fault image of the Port Lane gate, obtain the specific histogram correction processing results according to the image registration principle, and splice and preprocess the remote fault image. On this basis, the actual fault characteristics of the lifting rod mechanical action of the Port Lane gate are analyzed. The detection blind sources are separated by means of non-uniform sampling of the signal parameters, and then combined with the fault signal to output the signal-to-noise ratio value to realize the remote detection of the lifting rod mechanical action fault of the Port Lane gate. The experimental results show that the detection accuracy of the mechanical action fault detection of the lifting rod of the Port Lane gate based on the machine vision detection method can reach 90.4%, and the IMF component value is large, which can effectively suppress the influence of the noise signal on the coverage of the real signal and improve the fault detection accuracy of the mechanical action of the lifting rod of the Port Lane gate.
Key words：Machine vision; Port Lane gate; Gate lifting rod; Fault detection; Histogram correction; Blind source detection; Signal to noise ratio;
收稿日期：2021-08-06
基金项目：国家重点研发计划课题（2018YFC0809105）
