Title: Study the relationship between coal properties with Gieseler plasticity parameters by random forest
Authors: Saeed Chehreh Chelgani; Sahar Soleimani Matin
Addresses: Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI 48109, USA ' Young Researchers and Elite Club, Islamshahr Branch, Islamic Azad University, Islamshahr, Iran
Abstract: Gieseler fluidity provides thermoplastic information and the compatibility of blended coals for the cokemaking. A novel soft computing method, random forest (RF), for prediction of the softening temperature (Ts), the temperature of maximum fluidity (Tf), resolidification temperature (Tr) and maximum fluidity (MF) [Gieseler parameters (Gp)] was conducted based on the coal proximate analysis. Variable importance measurements were performed by RF to select the most effective variables for the prediction of Gp. Selected variables have been used as an input set of RF model for the modelling and prediction. Results of models indicated that RF can provide a satisfactory prediction of Gp with the correlation of determination R2: 0.64, 0.82, 0.90, and 0.86 for Ts, Tf, Tr and MF, respectively. Based on these results, it can be proposed that RF as a reliable non-parametric reliable predictive tool can be used for modelling of complex relationships in the fuel and energy investigations. [Received: January 28, 2017; Accepted: March 28, 2017]
Keywords: Gieseler; coal pyrolysis; coke; proximate analysis; random forest; variable selection.
DOI: 10.1504/IJOGCT.2018.089345
International Journal of Oil, Gas and Coal Technology, 2018 Vol.17 No.1, pp.113 - 127
Received: 28 Jan 2017
Accepted: 28 Mar 2017
Published online: 19 Jan 2018 *