Efficiency of deep neural networks for reinforced concrete shear walls failure mode detection

After a severe earthquake, it is critical to check the health of hospitals, buildings, fire stations, to name but a few. New technologies such as deep learning algorithms, image processing, and signal processing methods have become more appealing in such quick applications because of their higher reliabilities and efficiency than traditional visual methods. On the other hand, reinforced concrete shear walls are the key structural elements of the most buildings as lateral force resisting systems in seismic areas. Determining the failure mode of reinforced concrete shear walls is a crucial task that can help engineers and designers choose the best retrofitting options. The goal of this research is to assess the ability of deep neural networks for predicting the failure mode of the reinforced concrete shear walls. Furthermore, the SHapley Additive exPlanation approach is used to demonstrate the relevance and contribution of the factors that influence the reinforced concrete shear walls failure mode determination.