Blast Explosive Prediction Using Supervised Prediction Model

Accurate prediction of blast peak overpressure is crucial for safety and mitigation efforts. While extensive research has focused on characterizing blast waves, existing studies often prioritize spherical explosions originating from a central point, potentially overlooking the influence of factors such as explosive type, shape, and detonation position. This research addresses this gap by investigating the impact of these variables on blast peak overpressure prediction. Using experimental data from the detonation of 500 grams of PE-4 and Emulex explosives at distances ranging from 0.5 to 4.0 meters, we evaluate the performance of several supervised learning models, including K-Nearest Neighbors (KNN), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Principal Component Analysis (PCA), and Decision Trees (DT). While the relatively small dataset size presents a limitation, our results suggest that KNN outperforms the other models, achieving a mean squared error (MSE) of 0.31 and a regression score of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{0. 8 9}$</tex>. This highlights the potential of KNN for predicting blast peak overpressure when considering multiple variables, warranting further investigation with larger and more diverse datasets.