An ANN-adaptive dynamical harmony search algorithm to approximate the flyrock resulting from blasting

Blasting is the cheapest and most common method of rock excavation. The basic purpose of blasting is to breakage and displacement of rock mass and, on the other hand, it has some undesirable and inevitable effects such as flyrock. In this study, a novel hybrid artificial neural network (ANN) based on the adaptive musical inspired optimization method is proposed for accurate prediction of blast-induced flyrock. The dynamical adjusting process was adaptively introduced to enhance the ability of harmony search algorithm to obtain the optimum relationship between input variables, i.e., spacing, burden, stemming, powder factor and density of rock and output variable, i.e., flyrock. Two adjusting processes were used to update the new position of particles. The statistical information of the harmony memory was implemented in the proposed hybrid ANN coupled with adaptive dynamical harmony search (ANN-ADHS). The capacity for agreement, tendency, and accuracy of the proposed ANN-ADHS was compared with that of the ANN and two hybrid ANN models coupled by harmony search (ANN-HS) and particle swarm optimization (ANN-PSO) models using comparative statistics such as root mean square error (RMSE). The results confirmed viability and effectiveness of the ANN-ADHS model (with RMSE = 17.871 m and correlation coefficient (R²) = 0.929) and showed its capacity for better predictive performance compared to ANN-HS (with RMSE = 22.362 m and R²= 0.871), ANN-PSO (with RMSE = 24.286 m and R²= 0.832), and ANN (with RMSE = 24.319 m and R²= 0.831).