Research On Key Technologies Of Intelligent Loading In Quick Quantitative Loading Station

At present,coal mine enterprises have an urgent need to improve the efficiency of coal transportation and improve the loading effect.Realizing the quick quantitative loading of coal mines at smart stations has gradually become one of the hot research issues.With the continuous maturity of computer vision technology,the application of computer vision technology to quick quantitative loading stations solves the problems of poor loading effects and high labor intensity of quick quantitative loading stations,and the realization of intelligent loading is of great significance.The three key technologies of railway carriage number identification,railway carriage position detection and loading chute control at the station are studied,aiming to achieve rapid quantitative loading at the station with accurate loading,improve loading efficiency and improve loading effect.Aiming at the problems of various sizes of railway carriage numbers,different spraying effects and uneven environmental illumination,a railway carriage number recognition method based on improved Faster R-CNN network was proposed.Firstly,Retinex image enhancement algorithm was introduced to enhance the low illumination railway carriage number image to reduce the influence of illumination on the image;Secondly,cluster the Anchor box on the railway carriage number data set to obtain the reasonable anchor Finally,through the use of hierarchical multi-scale regional recommendation network,and the introduction of GIo U bounding box regression loss function,improve the accuracy of bounding box number recognition regression box and small number detection ability.The results show that the speed is 4FPS and m PA is 95.89%,which can realize the accurate and real-time identification of bounding box number in the loading process.Aiming at the problems of sprinkling,eccentric loading and pit retention caused by the current judgment of the railway carriage position based on manual experience,the upper frame division method of the railway carriage was designed to provide control signals for the control of the loading chute lifting timing and the opening timing of the quantitative bin gate.The image segmentation algorithm of the upper frame of the carriage was based on the FPN network to extract and fuse features of different resolutions and different semantic strengths;by introducing an attention mechanism based on the expectation maximization algorithm to filter noise and improve the accuracy of the image segmentation boundary,the model m Io U is 81.21 %,m PA is 88.64%,the railway carriage position detection experiment is carried out on the site of the quantitative loading system,and the detection accuracy of the railway carriage position detection model meets the requirements of the railway carriage position detection task during the loading process.Finally,the control strategy of loading chute was obtained by experiment,and the discrete element model of loading process was established to verify the control strategy of loading chute.The loading chute control system selects PLC as the core controller,receives the vehicle type information and railway carriage position information output from the visual inspection system,and completes the loading chute trigger control and height control by matching the vehicle type information with the loading chute control strategy,so as to realize the quick quantitative loading station intelligent loading.This paper has 75 pictures,15 tables,and 79 references.