基于深度学习与计算机视觉的皮带跑偏检测研究OA

In order to solve the problem of the belt deviation faults that may occur in the transportation of the belt conveyor,a belt devia-tion detection method based on the combination of deep learning target detection and computer vision is proposed.First,the video image information of the belt conveyor operation is captured by using the camera,and the YOLOv5 target detection method is used to find the belt rollers,which are used to mark the edge limit of the maximum belt displacement.Then,the image information is subjected to Canny operation and Hough transform using OpenCV to find the belt edge.Finally,multiple sections of belt edge data are labeled with multiple regression methods to obtain the real-time offset between the belt edge and the edge limit,which is used for belt deviation detection.The test results show that the method proposed has strong model generalization ability,is not easily affected by ambient light,and can effec-tively detect belt deflection of different types of belt conveyors.The detection efficiency of the real-time running situation of the belt is high,and the frame rate of the algorithm can be higher than 32 FPS,which can accurately make an effective judgment on the belt deflec-tion.The detection information obtained can assist staff to carry out better safe production,thus reducing the occurrence of belt conveyor accidents.