【一般講演】 開発機械／資源開発技術 ［9/12(木) AM 第2会場］

[3201-08-04] Integrating Soft Computing Techniques for Accurate Prediction and Assessment of Blast-Induced Ground Vibrations

○Yewuhalashet Fissha Yewuhalashet^{1[博士課程]}, Prashanth Ragam², Hajime Ikeda³, Hisatoshi Toriya¹, Tsuyoshi Adachi¹, Youhei Kawamura⁴ (1. Akita University, 2. VIT-AP University, 3. National Institute of Technology, Asahikawa College, 4. Hokkaido University)

司会：吉光 奈奈（京都大学）

キーワード：PPV, Machine learning, Blasting, Ground Vibration, Regression

Ground vibration resulting from rock blasting is a highly hazardous consequence of blasting. These vibrations have detrimental effects on both local ecology and nearby human populations. Assessing the severity of blasting vibrations necessitates a thorough evaluation of the ((PPV), which is an essential parameter for measuring vibration velocity. The accurate prediction of vibration occurrence is critically important. Therefore, this study employed three machine-learning models to predict the PPV resulting from quarry blasting. This study compares three machine learning models (XGBoost, Gradient Boosting, and Bagging) to choose the most efficient performance model. The performance evaluation of each of the three machine learning models demonstrates that each model achieved a performance of more than 0.90 during the testing phase, and a strong correlation was observed between the actual and predicted models. The analysis of performance metrics shows that the XGBoost regression model demonstrates better performance prediction compared with the other models, with R²=0.982, MSE= 0.000086, RMSE=0.009, MAD=0.004, and MAPE = 24.155 in the PPV prediction. This study will help mining engineers and blasting experts select the best machine learning model and its hyperparameters for estimating ground vibration. By employing machine learning models, this study aims to accurately predict and assess ground vibrations with frequencies resulting from rock blasting.