Managing blasting operations in four mines surrounded by villages is a challenging task. This study aims to compare the blast-induced flyrock as observed in opencast mines by predicting flyrock distance using a soft computing approach known as an adaptive neuro-fuzzy inference system (ANFIS) and Flyrock Predictor software based on empirical studies. Blast design and geotechnical variables such as linear charge concentration, burden, stemming length, specific charge, unconfined compressive strength and rock quality designation were selected as independent variables, while flyrock distance was used as a dependent variable. Out of 120 blasts, data from 100 blasts was used to train and test the ANFIS model. Flyrock distance for the remaining 20 blasts was used to evaluate the model. Flyrock distance has also been predicted using Flyrock Predictor software based on an empirical approach using the same data sets as those used for ANFIS. Performance indices like root mean square error (RMSE), mean absolute error and coefficient of determination (R2) were evaluated to compare the reliability of ANFIS and Flyrock Predictor quantitatively. ANFIS was found to be a better predictive tool, with 1.75 m RMSE and 0.966 R2 compared to Flyrock Predictor, which returned RMSE and R2 values of 2.48 m and 0.932 respectively.
Trivedi, R, Singh, T N, Gupta, N and Bhandari, S, 2015. Soft computing approach to predict blast-induced flyrock, in Proceedings 11th International Symposium on Rock Fragmentation by Blasting, pp 455–462 (The Australasian Institute of Mining and Metallurgy: Melbourne).