首页|Proactive measures to prevent conveyor belt Failures: Deep Learning-based faster foreign object detection
Proactive measures to prevent conveyor belt Failures: Deep Learning-based faster foreign object detection
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Elsevier
Deep learning-based faster identification and classification of foreign objects entrapped in coal conveyor belts is the research topic of this paper. The presence of foreign objects poses a serious threat to the service life of the conveyor belt and the safe transport of coal mines. The paper aims to detect foreign objects entrapped in the conveyed material in advance and provide early warning, thereby safeguarding the safe operation of the belt conveyor and preventing tearing accidents from occurring. Firstly, a new and large conveyor belt foreign objects dataset is established for train and test. Then, based on the higher and faster requirements for camera acquisition frequency and real-time algorithm processing that put forward by high-speed belt conveyors, the paper implements a deeply lightweight target detection network based on the Yolov4 by improving its backbone and neck to ensure the detection speed and accuracy. Exper-iments show that the improved network proposed in this paper has achieved a highest detection accuracy of 93.73% with the fastest detection speed of 70.1 FPS (frames per second) on proposed foreign objects dataset, which is an improvement of 1.72% in accuracy and 211.56% in detection speed compared to the original Yolov4; Furthermore, by exploring the influence of motion blur on the detection results, the necessity of shortening the exposure time and improving the video frame rate is proved. All this work is of great significance to improve the detection speed of foreign objects identification and is helpful to promote the application of it in the field of edge computing equipment, so as to ensure the safe and efficient operation of belt conveyors.
Belt conveyorForeign objects detectionDeep learningLightweight CNNssmart mineMotion blurCOALRECOGNITIONEXTRACTION
NSTL
Elsevier
