An adaptive neuro-fuzzy inference system and gene expression programming models for the prediction of blast-induced ground vibration in limestone quarries

초록

The impact of blasting activities on natural and man-made structures have received tremendous interest due to the growing danger as the quarries are been encroached by the building as a result of population increase. Owing to the imperativeness of the blast-induced ground vibration, various methods such as field measurement, empirical model, multilinear regression, and soft computing have been proposed to quantify its magnitude in term of peak particle velocity (PPV). Due to the high cost of obtaining the measuring equipment, safety and expertise required in operating the equipment, the blasters depend largely on the empirical models. The most commonly used method is the empirical models and multilinear regression (MLR) method. However, the traditional empirical methods and the MLR model have been found to be unreliable due to large error margin and low correlation coefficient, while the soft computing method such as artificial neural network (ANN) and the hybrid form of ANN has proven to be more reliable as evident in the literature. Therefore, adaptive neuro-fuzzy inference system (ANFIS) and gene expression programming (GEP) has been proposed in this study to predict the blast-induced ground vibration in limestone quarries. The charge per delay, distance from the measuring station to the blasting point and some controllable blasting parameters were the input parameters, while the PPV was the targeted output. About 23 blasting cases were used to develop the model. The datasets used in developing the model were divided into training and testing. The performance of the proposed model was evaluated using the coefficient of determination (R2), mean absolute error (MAE), and root-mean squared error (RMSE). The ANFIS performed better than the GEP model.