LAUSR

Abstract In the present study, principal component analysis (PCA) and stepwise selection and elimination (SSE) techniques were used to establish significant parameters (blasting design, rock and explosive) in surface coal mines by reducing dimensionality and variables from a host of blasting parameters. Mean fragment size (MFS) prediction models were subsequently developed using multiple linear regression (MLR) analysis technique. The two constructed and proposed models adequately selected relevant blast design, explosive and rock mass parameters. The performances of these models were assessed through the determination coefficient (R 2 ), F-ratio, standard error of estimate and root mean square error (RMSE). The PCA technique has shown good promise in eliminating the redundant parameters and in selecting relevant blast design parameters. Hierarchical cluster analysis technique was used for confirming the similarity of blasting design parameters in two trial blasting data set. The results were tested and validated with the 19 actual blast data set at acceptable correlation levels and have been illustrated in the form of figures, tables and graphs. MFS prediction equations based on PCA and SSE techniques were simple and suitable for practical use in overburden bench blasting of Indian coal mines.