Development Of A Machine Vision-based Online Analysis System For Coking Coal Particle Size

During the coking process,the particle size composition of the incoming coking coal has an important influence on the stack density in the coke oven,the adhesion index,the porosity in the coke and the wear condition of the coke oven.A reasonable mix of coking coal particle sizes is essential to ensure the strength of the coke and the stable and safe operation of the coke oven.The manual sieving method currently used by coking companies is cumbersome and time-consuming,and seriously lags behind industrial production,making it difficult to guide production operations in a timely manner.Machine vision-based coking coal particle size analysis has the advantages of easy operation and real-time online,but there is currently no online analysis equipment specifically for coking coal particle size in the market,and there is an urgent need to develop independent intellectual property rights of machine vision coking coal particle size online analysis technology and equipment.This paper focuses on the major demand for coking coal particle size control of incoming coal in the coking coal making process,and aims at the clean and efficient utilization of industrial coking coal,and develops the online analysis technology of coking coal particle size based on machine vision image processing and integrates complete sets of equipment,focusing on solving the image segmentation problems caused by the wide range of coal sources,variable coal types and fine particle sizes in the coking industry,and achieves the following key technologies in the online analysis of coking coal particle size breakthroughs.(1)A secondary segmentation image processing method based on convex packet detection is proposed.The image segmentation algorithm based on a combination of distance transformation,marker-controlled watershed algorithm and morphological operations is used to segment the coking coal particles for the first time,and then the convex packet detection technique is used to screen out the adhering particles,and when the convex packet rate of the particles is lower than a selected threshold value of 0.75,the particles are segmented for the second time,thus ensuring the particle size detection accuracy.(2)A stable online analysis technique for coking coal particle size has been developed.Ultra-high pixel imaging was used to improve image resolution to focus on coking coal particle sizes <0.5 mm and 0.5 to 3 mm.A levelling scraper assembly and coal flow switch were designed to solve the problem of peaked fuel distribution on the belt which tends to distort the image.The mean measurement error and mean uncertainty of the splitting algorithm were controlled at 3.56% and 2.31% respectively.(3)A machine vision online coking coal particle size analysis system is integrated into the optical and mechanical integration and applied to coking enterprises.The system consists of hardware testing equipment and data management and analysis software,with a single test time of 1 minute,output data types including particle size distribution,fineness,average particle size and particle mass ratio,self-cleaning and dust removal,data telemetry,multi-client display and history query,with a repeatability and accuracy of 6.54% and 3.10%respectively.