Prediction Of Ground Vibration In Open Pit Blasting Operation Using Support Vector Machine

Drilling and blasting are among the most important operations in open pit mining which sometimes are along with inappropriate events causing to some danger and problems. One of these undesirable and dangerous events is ground vibration phenomenon. Therefore, the control of blasting ground vibration has been an important research subject in engineering blasting field and is measured by the peak particle velocity (PPV). In this study, The effect of ground vibration parameters onprediction of PPV of ۳۴ blasting events of Soungun copper mine were used to develop two statistical models of multiple linear regression (MLR) and support vector machine SVM and results were compared with the empirical method of United State Bureau of Mines model (USBM). SVM is a novel machine learning technique usually considered as a robust artificial intelligence method in classification and regression tasks. The root mean square error (RMSE) value was used as the error function to examine the quality of the model. RMSE values for MLR, SVM, and USBM were calculated as ۱.۹۵, ۰.۲۳ and ۰.۸۶ respectively. Different values were set for SVM parameters and the optimal values of C=۱۰۰, =۰.۰۱ and =۰.۰۳۱ based on the model error value for PPV have been obtained. Model statistical parameters showed that there is a very good relation between PPVand the model variables with a R۲=۰.۷۴ for MLR, and ۰.۹۹ for SVM which is comparable with USBM R۲ of ۰.۹۷. Also, the squared correlation coefficient for the prediction set was achieved ۰.۷۳ for MLR, ۰.۹۹ for the SVM and ۰.۸۲ for USBM. The results represent the high ability of SVM method in the prediction of PPV compared to MLR model and empirical method for samples not included in the model building.

Keywords:

Blasting , Ground vibration , peak particle velocity , Support vector Machine , Soungun copper mine.