| As the key core equipment in the coal mine transportation system,the conveyor belt is prone to longitudinal tearing during long distances,full load,and frequent work.Once the longitudinal tearing accident of the conveyor belt occurs,it will seriously affect production and even cause personal injury.Therefore,it is extremely important to strengthen the real-time and reliable detection of the longitudinal tear of the mining conveyor belt.In view of the shortcomings of many current conveyor belt longitudinal tear detection methods,and considering the special situation of the complex and changeable coal mine production site environment,the paper introduces sound detection into the conveyor belt longitudinal tear detection,and proposes a weighted decision fusion based on audiovisual information The audio-visual detection method for longitudinal tear of the conveyor belt realizes the analysis and judgment of the normal,damaged and longitudinal tear state of the conveyor belt,and designs and completes the online real-time detection system platform to form a complete conveyor belt longitudinal tear audio-visual detection technology operation system.The thesis first analyzes the cause of the conveyor belt damage and the longitudinal tear of the conveyor belt from the actual operation of the mine conveyor,and analyzes the characteristics of the collected conveyor belt normal,damaged and longitudinal tear images,and the collected conveyor belt is running normally.Contrast and analyze the abnormal sound from many aspects such as frequency spectrum and waveform.Research the mechanism of image detection and sound detection of longitudinal tear of conveyor belt,explore the image and sound characteristics of longitudinal tear of conveyor belt,and analyze the feasibility of audiovisual detection of longitudinal tear of conveyor belt and the judgment theory of audiovisual detection.According to the characteristics of the audio-visual signal of the longitudinal tear of the conveyor belt,the paper constructs the audio-visual detection model of the longitudinal tear of the conveyor belt,and studies the realization method of the audio-visual detection.Build the conveyor belt detection image model,sound model and audiovisual decision fusion model respectively.The image of the conveyor belt is processed and the connected area is analyzed,the structural features of the area are extracted,and the K-nearest neighbor classifier is used to realize the image-based judgment of the conveyor belt operation status.In the sound detection model,the 13-dimensional combined features of the sound are extracted,and the Gaussian mixture model method based on the general background model is used to train and recognize the running sound of the conveyor belt,so as to realize the judgment of the running state of the conveyor belt based on the sound.Finally,the detection accuracy of the image model and the sound model is used as the weight,and the weighted decision fusion analysis is carried out on the detection results of the image model and the sound model,and finally the decision-making judgment on the normal,damaged and longitudinal tear of the conveyor belt is realized.The thesis is divided into modules to build a conveyor belt longitudinal tear detection system platform based on the weighted decision fusion of audiovisual information to realize the audiovisual information collection of the conveyor belt operation,real-time audiovisual information display,data analysis,data management,system settings and control operations.A relatively complete operation plan of the conveyor belt longitudinal tear detection system.At the same time,in order to verify the accuracy and reliability of the proposed conveyor belt longitudinal tear detection method,an experimental system for audiovisual testing of the conveyor belt longitudinal tear was built on the existing conveyor device in the laboratory to simulate the on-site conveyor belt operation status and conduct data collection.analysis.After analysis,it can be found that the average detection accuracy of the audio-visual detection system for the longitudinal tear of the conveyor belt in this paper is 93.27%,and the accuracy of the recognition of the longitudinal tear of the conveyor belt is more than 90.82%.Compared with the image detection method alone,the audiovisual decision fusion detection method is at least 11.90% higher than the image detection method.The audio-visual weighted decision fusion detection method for longitudinal tear of the conveyor belt in this study can accurately and reliably obtain the real multi-dimensional information when the longitudinal tear of the conveyor belt occurs,and realize the accurate real-time identification of the longitudinal tear of the conveyor belt.It is an online detection and early warning system for the longitudinal tear of the conveyor belt.The key technology research.The research results will promote the development of coal mine production safety early warning detection theory,enrich coal mine safety detection methods,promote the diversified development of coal mine safety detection,and provide strong technical support for stable and efficient production and safe dispatch decision-making in coal mines.It has important theoretical value and practical significance. |
PDF Full Text Request