Prediction of air-overpressure induced by blasting using an ANFIS-PNN model optimized by GA

Abstract

Blasting operations typically have several negative impacts upon human beings and constructions in adjacent region. Among all, air-overpressure (AOp) has been persistently attractive to practitioners and researchers. To control the AOp-induced damage, its strength should be predicted before conducting a blasting operation. This paper analyzes the AOp consequences through the use of the Fuzzy Delphi method (FDM). The method was adopted to identify the key variables with the deepest influence on AOp based on the experts’ opinions. Then, the most effective parameters on AOp were selected to be used in developing a new hybrid intelligent technique, i.e., adaptive neuro-fuzzy inference system (ANFIS)-polynomial neural network (PNN) optimized by the genetic algorithm (GA), called ANFIS-PNN-GA. From FDM and experts’ opinions, four parameters, i.e., amount of explosive charge, powder factor, stemming, and distance from the blast-face were identified as the most effective ones on AOp. In fact, in ANFIS-PNN-GA system, GA was used to optimize the ANFIS-PNN structure. The new framework of ANFIS-PNN-GA was developed, trained, and tested on actual datasets collected from a total of 62 blasting events. To show capability of the newly-proposed model, the ANFIS and PNN predictive models were also constructed to estimate AOp, and the performance prediction of the proposed models were evaluated through the use of several performance indices, e.g., correlation coefficient (R) and mean square error (MSE). R values of (0.94, 0.72, and 0.84) and (0.92, 0.58, and 0.77) and MSE values of (0.003, 0.03, and 0.021) and (0.005, 0.066, and 0.05) were obtained for training and testing datasets of ANFIS-PNN-GA, PNN, and ANFIS models, respectively. Accordingly, because of the role of GA as a practical optimization algorithm in improving the efficiency of both PNN and ANFIS models, results obtained by the ANFIS-PNN-GA model are more accurate compared to other implemented methods.