An intelligent approach to predict pillar sizing in designing room and pillar coal mines

Various empirical procedures have been developed to determine pillar sizing based on back analysis of failed and successful case histories and statistical analysis techniques. Artificial intelligence techniques are now being used as an alternate to statistical techniques. In this study, the fuzzy logic was applied to predict safe pillar sizing in room and pillar coal mines. The model predicts pillar length and width using depth of cover, mining height, panel width, roof strength rating and loading conditions. The predictive fuzzy model was implemented on fuzzy logic toolbox of MATLAB using the Mamdani algorithm and was developed based on a database including 399 datasets from US room and pillar coal mines. Eighty datasets of this database were used to assess the performance of this fuzzy model. The coefficient of determination (R2), the variance account for (VAF) and the root mean square error (RMSE) were calculated to check the prediction performance of the model. The R2, VAF and RMSE values were obtained as 89.3%, 89.27 and 1.39 for the pillar width, and 86.6%, 86.4 and 2.77 for the pillar length. These indices revealed that the developed model is suitable for practical use at mines. In addition, the strength of the relationship between the pillar sizing and the five input parameters were evaluated by the cosine amplitude method and the results showed that the most effective parameter on pillar sizing is loading conditions.

• The paper presents a fuzzy model to predict pillar sizing.
• The model was developed based on a database from US room and pillar coal mines.
• The results show that fuzzy model is an efficient tool for predicting pillar sizing.
• The output of the model provides a good estimation for reducing pillar failure risk.
• Loading conditions was found to be the most effective parameter on pillar sizing.