Abstract
Purpose: Hardgrove grindability index (HGI) is an important physical parameter used to demonstrate the relative hardness of coal particles. Modeling of HGI based on coal conventional properties is a quite complicated procedure. The paper aims to develop a new accurate model for prediction of HGI that is called optimized evolutionary neural network (OPENN). Design/methodology/approach: The procedure for generation of the proposed OPENN predictive model was performed in two stages. In the first stage, as the high dimensionality involved in the input space, a correlation-based feature selection (CFS) algorithm was used to select the most important influencing variables for HGI prediction. In the second stage, a combination of differential evolution (DE) and biography-based optimization (BBO) algorithms as a global search method were applied to evolve weights of a multi-layer perception neural network. Findings: The proposed OPENN was examined and compared with other typical models using a wide range of Kentucky coal samples. The testing results showed that the accuracy of the proposed OPENN model is significantly better than the other typical models and can be considered as a promising alternative for HGI prediction. Originality/value: As HGI test is relatively expensive procedure, there is an economical interest on HGI modeling based on coal conventional properties (proximate, ultimate and petrography); the proposed OPENN model to estimate HGI would be a valuable and practical tool for coal industry.
Original language | English |
---|---|
Pages (from-to) | 1003-1048 |
Number of pages | 46 |
Journal | Engineering Computations (Swansea, Wales) |
Volume | 35 |
Issue number | 2 |
DOIs | |
State | Published - 2018 |
Bibliographical note
Publisher Copyright:© 2018, Emerald Publishing Limited.
Keywords
- Biography-based optimization
- Coal
- Differential evolution
- Hardgrove grindability index
- Neural networks
ASJC Scopus subject areas
- Software
- General Engineering
- Computer Science Applications
- Computational Theory and Mathematics