Research On Belt Conveyor Roller Fault Detection Based On Machine Visio

Belt conveyor is a continuous transportation equipment in modern production.It has the advantages of long transportation distance,low freight,large transportation volume,convenient loading and unloading,low energy consumption,and suitable for bulk transportation,etc.It has been widely used in coal,mines,and ports,electric power,metallurgy,chemical industry and other fields.Together with automobiles and trains,it has become the three major industrial transportation tools.The belt conveyor is a large-scale rotating mechanical equipment that contains a large number of roller groups.The rollers rotate by friction with the conveyor belt,and often malfunctions,causing the rotational line speed of roller to decrease or jam.When the rollers fail,the friction between the conveyor belt and the rollers will increase,which will increase the power consumption of the belt conveyor,which will cause major accidents such as fire and longitudinal tearing,which will seriously affect safe production.In order to ensure the safe and reliable operation of the belt conveyor,it is necessary to detect the failure of the belt conveyor rollers.Due to the large number of rollers,many sensors are needed for detection,and the cost is high.The rollers rotate on the lower surface of the conveyor belt and rotate during operation.The site environment is harsh,the space is small,and the roller failure detection is difficult.Existing belt conveyor roller failure detection methods have the problems of high cost,poor reliability and accuracy.This thesis studies the failure detection technology of belt conveyor roller based on machine vision,and proposes a machine vision-based belt conveyor roller failure detection method.The method uses inspection robot equipped with industrial cameras to collect roller image sequences.Through the belt conveyor roller failure detector,the belt conveyor roller rotation linear speed detection algorithm is used to process and analyze the collected roller image sequence to obtain the rotational line speed of roller,and realize the detection of the roller failure according to the rotational line speed of roller,and the detection result uploads to the upper computer.Carry out alarm processing when the roller failure is detected;A belt conveyor roller rotation linear speed detection algorithm is proposed.The algorithm uses the YOLOv4-Tiny target detection algorithm to extract the roller region of interest(ROI),and uses the roller segmentation algorithm to extract the surface area of the roller.The image processing algorithm for the surface area of the roller of two adjacent frames using the difference between frames determines whether the roller is jammed.If the roller jam is detected,an alarm will be issued.If the roller is detected to be rotating,the dense sampling algorithm is used to extract the characteristic points of the roller surface area.Using pyramid LK sparse optical flow method to track feature points.According to the positions of multiple corresponding feature points of two adjacent frames of images,multiple the rotation angle values of roller are calculated.Use statistical methods to determine the rotation angle of the roller between frames for multiple the rotation angle values of roller.Then the rotating angle of the roller between frames and the time between frames are used to calculate the rotational line speed of roller;The hardware of the belt conveyor roller failure detector is designed by Jetson TX2 development board,and the software of the belt conveyor roller failure detector and the upper computer software are designed by using C++ language in the Linux operating system;The communication software between the belt conveyor roller failure detector and the upper computer is designed by using TCP/IP protocol;An experimental platform for the failure detection system of belt conveyor roller based on machine vision is built,and experiment is carried out.The experimental result shows that the average accuracy of the roller jam detection by this method is 96.6%,and the maximum error of rotational line speed detection does not exceed 5.3%.It can be used for the failure detection,has the advantages of low detection cost,high reliability and accuracy,and has a wide range of application prospects in coal,mines,ports and other fields.