Performance of fuzzy logic and artificial neural network in prediction of ground and air vibrations

Abstract

The prediction of air and ground vibrations is an important problem inrock blasting activities. The aim of this study is to evaluate the predictionof ground and air vibrations by using intelligent networks and traditionalregression model. So, fuzzy logic and artificial neural network (ANN)models have been constructed to predict peak particle velocity and airoverpressure induced by blasting in Assiut Cement Company. For thispurpose, the peak particle velocity, air vibrations, and charge weight perdelay were recorded for 136 blast events at various distances and usedfor the training of the predictor models. About new 26 data sets havebeen used to test and validate the models. The performance, validity andcapability of these models to predict were proved to be successful bystatistical performance indices. These indices are variance-accounted for(VAF) and root mean square error (RMSE). The results from thesemodels asserted that, intelligent networks technologies can be preciselyand effectively used for predicting the air and ground vibrations incomparison with traditional regression analysis. Also, the comparisonindicated that the fuzzy logic model exhibited slightly better predictionperformance and generalization than the artificial neural network inground and air vibration prediction.