Development of GA-based models for simulating the ground vibration in mine blasting

Rock blasting is a well-known and common method for the removal of rock masses from an excavation in surface mines and civil projects. Ground vibration is the most hazardous effect induced by blasting operations. Therefore, the level of the blast-induced ground vibration needs to be predicted with a good level of the accuracy. The goal of this paper is to propose two novel practical intelligent models to approximate the ground vibration through genetic algorithm (GA). For comparison aims, the Roy and Rai-Singh empirical models were also employed. The requirement datasets were collected from the Shur river dam, in Iran. Specific charge, distance from the blast face and weight charge per delay were used as the input/independent parameters and peak particle velocity (PPV) was used as the output/dependent parameter. In total, 85 datasets including the mentioned parameters were prepared. Then, the models performance was assessed using statistical indicators, i.e., coefficient correlation ( R 2 ) and root mean squared error. According to the obtained results, it was concluded that GA-based models, with the R 2 of 0.977 and 0.974 obtained from GA-power and GA-linear models, provide relatively closer predictions as compared to Roy and Rai-Singh empirical models, with the R 2 of 0.936 and 0.923, respectively.