Abstract
Reinforced concrete (RC) slabs as the primary force member in the engineering structure is often subjected to the threat of terrorist attacks or industrial gas explosions. Therefore, the dynamic response of the RC slab under the action of the explosion is an important index of the impact of anti-blast in engineering structures. Thus, it is necessary to investigate and predict the structural response of RC slabs under blast load that can improve the proactive measures to alleviate life and economic losses. In this study, a machine learning model was introduced to predict the maximum displacement of reinforced concrete slabs under blast loads using the ten features or parameters related to reinforced concrete slabs and explosions. The dataset used in this study consists of 260 data examples selected from the existing experimental, numerical, and analytical studies in the open literature. The support vector machine, Gaussian process regression, random forest, and BP neural network algorithms of machine learning regression algorithms are used to calculate the dynamic response of RC slabs, and the robustness of the four machine learning algorithms was validated by statistical performance metrics and compared with existing methods and feature importance analysis. Finally, the reasons for model errors, improvements, and applications of the model are discussed and analyzed. The results show that the machine learning models have high prediction performance, and the Gaussian process regression algorithm has the highest prediction accuracy in calculating the dynamic response of RC slabs. It is found that the Mean Absolute Error (MAE) value and the Root Mean Square Error (RMSE) value of the test set are 4.07 and 5.79, and the R2 value is 0.96. Simultaneously, the machine learning method with higher prediction accuracy and computational efficiency is better than the existing prediction methods. Furthermore, the influence of different input parameters on the output results of the model is analyzed and obtained, which realizes the interpretability of the output results and increases the reliability of the model.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have