The longwall mining method is often affected by the out-of-seam dilution （OSD）. Therefore, predicting and controlling of dilution are important factors for reducing mining costs. In this study, the fuzzy set theory ...The longwall mining method is often affected by the out-of-seam dilution （OSD）. Therefore, predicting and controlling of dilution are important factors for reducing mining costs. In this study, the fuzzy set theory and multiple regression models with parameters, including variation in seam thickness, dip of seam, seam thickness, depth of seam, and hydraulic radius as inputs to the models were applied to pre- dict the OSD in the longwall coal panels. Field data obtained from Kerman and Tabas coal mines, lran were used to develop and validate the models. Three indices including coefficient of determination （R2）, root mean square error （RMSE） and variance account for （VAF） were used to evaluate the perfor- mance of the models. With 10 randomly selected datasets, for the linear, polynomial, power, exponential, and fuzzy logic models, R2, RSME and VAF are equal to （0.85, 4.4, 84.4）, （0.61, 7.5, 59.6）, （0.84, 4.5, 72.7）, （0.80, 4.1, 79.6）, and （0.97, 2.1, 95.7）, respectively. The obtained results indicate that the fuzzy logic model predictor with R2 = 0.97, RMSE = 2.1, and VAF = 95.7 performs better than the other models.展开更多
文摘The longwall mining method is often affected by the out-of-seam dilution （OSD）. Therefore, predicting and controlling of dilution are important factors for reducing mining costs. In this study, the fuzzy set theory and multiple regression models with parameters, including variation in seam thickness, dip of seam, seam thickness, depth of seam, and hydraulic radius as inputs to the models were applied to pre- dict the OSD in the longwall coal panels. Field data obtained from Kerman and Tabas coal mines, lran were used to develop and validate the models. Three indices including coefficient of determination （R2）, root mean square error （RMSE） and variance account for （VAF） were used to evaluate the perfor- mance of the models. With 10 randomly selected datasets, for the linear, polynomial, power, exponential, and fuzzy logic models, R2, RSME and VAF are equal to （0.85, 4.4, 84.4）, （0.61, 7.5, 59.6）, （0.84, 4.5, 72.7）, （0.80, 4.1, 79.6）, and （0.97, 2.1, 95.7）, respectively. The obtained results indicate that the fuzzy logic model predictor with R2 = 0.97, RMSE = 2.1, and VAF = 95.7 performs better than the other models.
