Modeling of Bed Depth Profile of Materials in Rotary Drums Using Support Vector Regression (SVR): Comparison with a Well-Structured Parametric Model

JOURNAL OF CHEMICAL ENGINEERING OF JAPAN

Mohammad Reza Yousefi¹  Mansour Shirvani¹ 

[1] School of Chemical Engineering, Iran University of Science and Technology

关键词: Rotary Drum;    Steady-State Modeling;    SVR Method;    Material Flow;    Bed-Depth Profile;   

DOI  :  10.1252/jcej.14we273

来源: Maruzen Company Ltd

【 摘 要 】

References(28)Based on experimental data gathered from a pilot-scale rotating drum, some improvements in the steady-state modeling of the bed depth of the material flow in a rotary drum are presented. These improvements were achieved in two ways: first, through the application of a support vector regression (SVR) method on the experimental bed depth data collected for the entire length of the drum including the critical point of the discharge end, and second, by incorporating four additional parameters into an existing well-known steady-state model and further fitting the modified model to the experimental data using a genetic algorithm (GA). As a boundary condition of the modified model, the discharge-end bed depth data were also modeled using SVR during the parameter estimation of the improved model. The bed depth of the material was measured through image processing. The new models were shown to fit the experimental data much better than the existing traditional model, particularly at a higher axis inclination and lower rotational speed. As the main conclusion, no large deviations were observed for the improved model in comparison with the original model when predicting the real depth at the discharge end of a drum.

【 授权许可】

Unknown