Research On Conveyor Belt Abnormal State Monitoring System Based On Machine Vision

The belt conveyor is a machine that relies on friction to achieve continuous transportation.It has the characteristics of strong transportation capacity,long transportation distance,simple structure and easy maintenance.It is widely used in metallurgy,power,coal and port transportation.Conveyor belt is an important component of belt conveyor,which is mainly used to carry materials and realize long-distance transportation under the traction of driving roller.Due to the harsh working environment of the conveyor belt and long-term load operation,it is very prone to failure.If the conveyor belt failure is not detected and handled in time,it will cause the conveyor belt to spatter,break or even cause a fire,resulting in a major safety accident.Therefore,the monitoring of the abnormal state of the conveyor belt is an urgent problem to be solved.This paper analyzes the current research situation of conveyor belt abnormal state detection technology at domestic and foreign.This paper combines the characteristics of the working environment of the conveyor belt,proposes a machine vision-based conveyor belt abnormal state monitoring system.The monitoring system proposed in this paper realizes the non-contact detection of the abnormal state of the conveyor belt.In this paper,the principle of online detection of abnormal state of the conveyor belt is studied in detail,the software and hardware of the system are selected,and the overall design of the conveyor belt abnormal state monitoring system is completed.This paper analyzes the characteristics of image motion blur and deduces the degradation model of conveyor belt motion blur image.This paper proposes a method of estimating the point spread PSF function based on the principle of Radon transform,and uses the least constrained square method to restore the blurred image.Several image preprocessing algorithms were analyzed,combined with the characteristics of the conveyor belt image,using histogram equalization and median filtering to enhance the image quality;The image segmentation algorithm is studied.Based on the simplified PCNN model and the Renyi cross-entropy image threshold selection method,an improved PCNN image segmentation method is proposed,which uses information entropy,regional contrast and global consistency error to objectively evaluate the segmentation effect;The gray-scale template matching algorithm is studied,and the SSDA constant threshold Tk is introduced into the NCC template as the template matching threshold,which realizes the rapid recognition of the conveyor belt crack image.The moving target recognition and tracking methods is researched.Combined with the characteristics of the operating environment of the conveyor belt,the background difference method and the two-frame difference method are used to realize the detection of the moving target,and the Kalman filter is used to track the moving track of the target;The relationship between the video image sequence and the actual distance is analyzed,a mapping table of video sequence image tracking points and spatial distance is established,the running speed of the conveyor belt is calculated,and the running speed of the conveyor belt is monitored.Designed a conveyor belt control module based on the STM32F103VET6 development board.The monitoring system is developed based on the Visual Studio2017 platform using the OpenCV3.4.1.A conveyor belt experimental platform is built in the laboratory,and the overall scheme is tested.Experiments show that the machine vision-based conveyor belt abnormal state monitoring system can realize automatic monitoring of conveyor belt abnormal state.