An improved estimation of coal particle mass using image analysis

• Published in: Powder technology 2012-10, Vol.229, p.178-184
• Main Authors: Zhang, Zelin, Yang, Jianguo, Ding, Lihua, Zhao, Yuemin
• Format: Article
• Language: English
• Subjects: Applied sciences; Chemical engineering; Coal; Coarse coal particle; Density; Errors; Exact sciences and technology; Image analysis; Image processing; industry; Mass model; Mathematical analysis; Mathematical models; Miscellaneous; Particle mass; Particle mass model; Powder technology; Solid-solid systems
• ISSN: 0032-5910; 1873-328X
• DOI: 10.1016/j.powtec.2012.06.027

An improved mass model with only two parameters for coarse coal particles using image analysis was established. Five variables—density, projected area, perimeter, length and breadth of the best-fit-rectangle, were used for the initial mass model. Due to the discovery of some relationships between the variables, the improved mass model only consisted of projected area and density and was established with a satisfied accuracy. Simultaneously, the computation speed of improved mass model is 5 times more than that of initial mass model. When tested 50 times against a different batch of modeling sample, the total relative errors of the improved mass model were less than 6%, however, the relative errors of 25–50mm were less than 7% and that of other size fractions were less than 5%. Experimental results revealed that only projected area and density were required to develop such model with satisfactory results. A promising application of this mass model will improve the on-line control of coal industry. We proposed an improved mass model for the estimation of coarse coal particles using image analysis, which was only consisted of two variables, i.e. projected area and density of coal particles. A series of experiments was designed to establish and test the improved mass model, and some conclusions were acquired. [Display omitted] ► Developing an effective thickness model referring to the previous model parameters by MLR. ► Discovering three relationships between projected area and other parameters. ► Developing the final mass model with high accuracy, fast calculation and no assumptions of coal sample based on two variables.