Title: Evolving neural network using genetic algorithm for mining method evaluation in thin coal seam working face
Authors: Wang Chen; Tian Shixiang
Addresses: School of Mining, Guizhou University, Guiyang, Guizhou 550025, China ' School of Mining, Guizhou University, Guiyang, Guizhou 550025, China
Abstract: Mining method selection is one of the non-linear decisions made by mining engineers. The artificial neural network (ANN) is a commonly used method for this decision-making. This paper investigated the effectiveness of ANN for mining method evaluation. The back-propagation (BP) algorithm was selected. The input variables are the geological conditions and face parameters. The output ones are the mining method and the production. Synthesising iterative efficiency and MSE, BP with VSS algorithm by appending MT was the priority to the evaluation. For better results, ANN optimised through genetic algorithm (GA) was also applied and tested. As a result, the mean square errors (MSE) for ANN and GA-based ANN at testing stage are 0.54 and 0.08, respectively. Moreover, the correlation coefficient R2 values are 0.99 and 0.96. The gained results indicated that GA-based ANN was more promising for mining method evaluation.
Keywords: ANN; artificial neural network; genetic algorithm; mining method; back-propagation; thin seam.
International Journal of Mining and Mineral Engineering, 2018 Vol.9 No.3, pp.228 - 239
Available online: 10 Nov 2018 *Full-text access for editors Access for subscribers Purchase this article Comment on this article