Machine learning model for predicting structural response of RC slabs exposed to blast loading

Abstract

Considering the risk of exposure of diverse structures to detonations and explosions, the need for understanding the structural behavior under such events and enhancing blast resistance is a growing topic of importance. The present study introduces a machine learning model to predict the maximum displacement (output) of reinforced concrete slabs exposed to blast loading using ten (input) features including length, width, and thickness of the slab, concrete compressive strength, reinforcing steel yield strength, steel reinforcement ratio, reflected impulse, blast scaled distance, type of slab, and slab support. The dataset used in this study consists of 150 data points compiled from studies retrieved from the open literature. The effect that each input feature has on the output was investigated using the variable importance measure, Permutation Feature Importance, in which the effect of features is compared to parametric studies found in the literature. The Random Forests algorithm was used to develop the learning model and its performance was compared to that of other learning algorithms. Additionally, a hybrid classification-regression Random Forests algorithm was implemented for the development of the final model. The performance of the machine learning model in predicting maximum slab displacement under blast loading was satisfactory with a MAE value of 4.38, a VEcv value of 94.4%, and an R2 value of 96.2%, while being computationally more effective than existing numerical approaches.