Research On Fault Diagnosis Technology Of Underground Belt Conveyor In Coal Mine Based On Machine Vision

As an important equipment for coal production and transportation,underground belt conveyor in coal mine has the characteristics of high volume and long distance,but its working environment is harsh,and it is easy to have safety faults,which not only affects the stability of coal mining work,but also poses a threat to the life safety of production personnel.At present,the conveyer fault detection devices used in underground coal mine have single function and poor reliability,so it is difficult to achieve accurate detection under harsh working conditions.In order to make up for the defects of traditional fault detection devices,this thesis proposes a fault diagnosis technology of underground belt conveyor in coal mine based on machine vision,which mainly aims at detecting common faults of conveyor,including belt deviation,coal stacking and smoke faults.The main research work of this thesis is as follows:(1)Based on the actual conditions of underground coal mine,the image acquisition system of underground coal mine belt conveyor is designed.The characteristics of underground image in coal mine are analyzed,and the problems of low brightness,floating dust and water mist and low clarity are obtained.The global adaptive high dynamic range imaging method is selected to enhance the image brightness.At the same time,the dark channel prior method and automatic level adjustment are used to de-fog the image.Proper image segmentation,contrast enhancement and image filtering are selected to improve the quality of downhole images.(2)According to the requirement of fault determination,feature extraction of image is carried out.For the belt bias fault,Canny edge detection and Hough transform algorithm were used to extract the straight line feature of conveyor belt edge.Aiming at the coal stacking fault,the distance between the detecting end and the detecting target is obtained by triangulation method.Aiming at the smoke fault,the moving target region is obtained by using the interframe difference method,and the color,motion and shape features of the moving region are extracted by combining the RGB three-channel component extraction and optical flow method.(3)According to the three fault types,the method of fault analysis and discrimination is proposed respectively.In view of belt bias fault,based on the obtained straight line characteristics of conveyor belt edge,the positioning and judgment method of double reference line was proposed to realize the identification of belt bias fault.Based on the distance value obtained,combined with zero value calibration and threshold setting,the coal pile height detection and coal pile fault identification are realized.Aiming at the smoke fault,combined with the color,motion and shape characteristics of the moving region,the RGB channel value decision rule,the main direction of motion decision rule and the shape change decision rule are adopted.Through the rule fusion decision,the smoke fault identification is realized.(4)In order to verify the effectiveness of the method proposed in this study,an experimental platform of a small belt conveyor was designed.Combined with the existing large belt conveyor and the simulated video,experiments were carried out to identify the belt deviation fault,coal pile fault and smoke fault.The results showed that the three fault detection algorithms proposed in this study all had high accuracy.Combined with monitoring requirements,the interface of fault monitoring platform for underground belt conveyor in coal mine is designed.There are 85 pictures,7 tables and 80 references in this thesis.