Skip to main content
No Access

Development of image-based analyser for sizing of raw materials in the steel industry

Published Online:pp 1-18https://doi.org/10.1504/IJIT.2021.123071

Quality coke as one of the raw materials is very important to produce hot metal in a blast furnace. Coal with required size fractions is fed to the coke oven for efficient coking operation after stamping. The determination of size fraction of coal particles is very much essential for the efficient stamping process. Offline sieve analysis is generally used for this purpose irrespective of highly time consuming, non-representative, maintenance prone. Real time machine vision system which robustly determines the size, distribution of free falling coal particles at the end of conveyor belt was developed to eliminate drawbacks of sieving. The new feature extraction algorithm was developed to obtain features, a subset of which is statistically chosen on their relative relevance for classification using minimum distance classifier (MDC), Bayes Gaussian network and probabilistic neural networks (PNN) with PNN outperforming the others for 17,000 training samples, and 15,000 testing samples. Additionally, this size analysis result was compared with that of sieve analysis. The outcome of this experiment showed slight tilt toward finer sizes which is desired since one of its application lies in minimising the fine generation to optimise the operation of the plant.

Keywords

chemical leaching, SCADA, FTV, PLC, automation