Study On The Key Technology Of Foreign Object Detection Of Coal Mine Belt Based On Edge Computing

Belt conveyor is a key part of underground coal mine transportation system.Due to the long transportation distance and complex geological conditions,various foreign objects occur frequently on the belt,which seriously affects coal mine safety production.Therefore,belt foreign object detection has become an important research content of mine intelligent information processing.The current underground coal mine belt monitoring system is mostly based on sensors to monitor voltage,current and other operating parameters,unable to identify the belt foreign object effectively,this thesis aims to conduct a comprehensive visual detection of foreign objects that may damage belt conveyers,so as to make early warning and treatment of accidents in advance.The conventional centralized processing schemes based on video server have high computational cost and heavy network traffic load.Aiming at the above problems,this thesis proposes a belt video monitoring method based on edge computing and an optimization algorithm of target detection adapted to edge devices by analyzing the characteristics of data transmission and target detection.The main works are as follows:(1)Aiming at the problems of high computing cost and heavy network traffic load of centralized processing scheme based on servers,this thesis proposes a visual detection method for foreign objects of underground belt based on edge computing.Firstly,the embedded computing cards are deployed in the belt monitoring terminal,and the target detection model deployed on the embedded computing cards is used to detect the foreign objects in the belt in real time,so as to reduce the computing pressure on the server and improve the real-time response of the system.Secondly,the detection task on the edge computing device is split,and only the target recognition part is focused on,so as to reduce the time consumption of the computing task and further improve the real-time processing performance,so as to meet the needs of detection of foreign objects in underground belts.(2)Aiming at the problems of long detection time and low accuracy of the algorithm YOLO-V3-Tiny in foreign object detection of belt in underground coal mine,this thesis proposes a detection algorithm named Fire-Dense-YOLO that is more suitable for edge computing devices.Through in-depth analysis of the network structure of the algorithm and the study of the number of parameters in the model and the impact on accuracy,8-layer fire modules were added to reduce the number of parameters and the amount of calculation,and reduce the size of the model.At the same time,dense connection and pass-through layer were introduced to improve the detection accuracy.The experimental comparison between two models on VOC data set and belt foreign object data set shows that the optimization algorithm proposed in this thesis is superior to YOLO-V3-Tiny in model space occupier,calculation amount,real-time performance and precision,and Fire-Dense-YOLO realizes a smaller,faster and more accurate belt foreign object detection in underground coal mine.At last,this thesis constructs the underground coal mine belt foreign object detection prototype system based on edge computing,the system can detect the foreign object in the belt in real-time by video.The results show that the system has good realtime performance and high accuracy,which can provide effective guarantee for the safety production of coal mine enterprises.In this thesis,there are 37 figures,14 tables and 86 references.