A Combination of Fuzzy Delphi Method and ANN-based Models to Investigate Factors of Flyrock Induced by Mine Blasting

The identification of parameters that affect mining is one of the requirements in executive work in this field. Due to the dangers of flyrock, studying the role of the factors that affect it will be useful to control this serious environmental issue of blasting. In this research, using hybrid intelligence techniques, a new guide to investigate the parameters that affect the occurrence and characteristics of flyrock is presented. Hybrid models were improved based on five types of optimization algorithms, namely particle swarm optimization, artificial bee colony, the imperialist competitive algorithm, firefly algorithm (FA), and genetic algorithm. The process of designing the structure of the models was controlled under the fuzzy Delphi method. This filter helps to determine the most important factors that play a key role in the flyrock phenomenon and its accurate prediction. The best optimization technique was selected based on applying two popular performance indices, i.e., the root-mean-square error and coefficient of determination (R-2). As a result, the best combination obtained was the FA-artificial neural network (ANN), which was able to provide the best optimization of the weights and biases of the ANN among all the proposed models. In addition, this system showed the lowest network error in the prediction of flyrock compared to other ANN-based models. The new combination (FA-ANN) can be used as a powerful and practical technique to predict the flyrock distance prior to blasting operations.

Automated summary

The study used hybrid intelligence techniques to examine the parameters affecting flyrock, optimized models with various algorithms, and ultimately selected the FA-ANN combination model as the best choice for accurate prediction of flyrock distance.