Research And Implementation Of Coal Dust Detection Algorithm For Chute Discharge Based On Deep Learning

The Dust will be generated in the process of unloading coal from the chute of ship loader in coal port.In order to solve the problem of dust,the coal port manually controls the dust removal device according to the size of dust for a long time or opens the dust removal device in the whole process.Among the two methods,the former needs to be handled by special personnel,which has the problem of low production efficiency.The latter will affect the economic benefits of the port.In order to solve this problem,this paper studies the coal dust detection algorithm based on deep learning,and builds a detection system based on the algorithm.The detection system can replace the manual real-time detection of dust.When the dust is detected,the system will send alarm signal to the PLC controlling the dust removal device to realize the automatic treatment of the dust discharged from the chute.The innovation and main work of this paper are as follows:(1)Aiming at the problem of real-time and accuracy of dust detection,an improved dust detection algorithm is designed based YOLOv4-tiny network.In this paper,the SERes module is proposed to enhance the information interaction between the detection algorithm network channels;the XRes module is proposed to increase the depth and width of the algorithm network;and the SPP module and PRN module are added to enhance the feature fusion ability of the algorithm,which can significantly improve the network performance without affecting the network detection speed.Then the dust data set of chute discharge is constructed and trained by the improved algorithm to realize the real-time detection of dust.The real-time and accuracy of the dust detection algorithm are verified by simulation experiments and practical application.(2)In order to solve the problem that the object detection algorithm based on deep learning can only judge whether there is dust or not,and cannot obtain the dust distribution information.In this paper,a total of 16378 coal dust pictures of chute discharge were collected,and the coal dust was divided into four categories for detection.The first is the bright white suspected dust,which is often disturbed by the coastal fog or hot steam emitted by coal;the second type is blue or gray black dust,which has high identification;the third type is all diffused dust except the above two types,which is usually gray white;the fourth type is non diffused dust,which is small but also included in the detection range in order to better grasp the dust change trend.(3)The dust detection software is designed based on MFC,which is the basic class library of Microsoft,in the integrated development environment of Microsoft Visual studio 2015.The software adopts multithreading technology,including video acquisition and display as the main thread,image video processing,detection result preservation and communication with PLC controller through OPC protocol as sub thread.The detection software can read and display the running status of the device in real time,and allow the operators with corresponding permissions to issue operation instructions to the equipment operation.