A new paradigm for intelligent status detection of belt conveyors based on deep learning

Abstract

All-round visual monitoring of conveyor belt running status is an important component of the belt conveyor operation and maintenance system. Intelligent analysis methods will help improve the detection in real-time and effectiveness and help make reasonable decisions. Aiming at the current situation that the existing technology cannot achieve integrated detection, this paper proposes a new conveyor belt running status monitoring paradigm based on deep learning for the first time. By grafting the segmentation network into the target detection network Yolov5, the integrated detection of various tasks, including fuzzy load measurement, deviation status detection, large foreign objects, and belt damage detection, etc., is realized in a real sense. Experiments show that the proposed method balances detection accuracy and speed well, with a detection accuracy of up to 97%, a segmentation accuracy of up to 100%, and a detection speed of up to 90 frames per second (FPS). The proposed integrated detection work is of great significance for improving the intelligent analysis capability of the operation and maintenance system, thereby ensuring the safe and efficient operation of the transportation system. Our source dataset is available at https://github.com/zhangzhangzhang1618/dataset_for_belt_conveyor/tree/master.