Prediction of Ground Vibration Using Various Regression Analysis
S. K. Bisoyi, B. K. Pal
National Institute of Technology, Rourkela, Odisha, 769008 India
Keywords: Сейсмические колебания, взрывные работы, пиковая скорость колебаний, эмпирические зависимости, модели статистической регрессии, регрессия на основе гауссовского процесса, Ground vibrations, blasting, peak particle velocity, empirical formulas, statistical regression models, Gaussian Process Regression
Abstract
Blasting still dominates the most suitable and economic processes of exploitation of minerals from the ground. Although there have been many advancements to optimize blasting to inhibit the impacts due to ground vibration caused by it, still there is a long way to go. Some empirical formulas from the past have helped in designing the mining process and served us well in configuring the blast design to minimize the adverse impacts on the surrounding environment. A couple of empirical formulas taken in this study have also proven worthy for predicting the ground vibration with good accuracy, but the reliance of the empirical formulas on only two parameters is their limitation since the beginning. This study aims to find alternatives with the help of various regression models and comparing their competence against the more traditional predictors existing today. The findings of this study suggest that the regression methods can have a better insight into the prediction of the PPV corresponding to the input parameters. The GPRs (Gaussian Process Regressions) were able to predict the ground vibration with much better precision compared to the linear regression methods and also the empirical predictors.
|
