Multi-scale image-based damage recognition and assessment for reinforced concrete structures in post-earthquake emergency response

Abstract

As cities continue to develop, the significance of resilience and intelligence is increasing. In post-earthquake emergency response, there is a continuous demand for efficient and accurate methods of structural state assessment. This study employs deep learning (DL) techniques to propose a multi-scale image-based damage recognition and assessment method for reinforced concrete (RC) structures. First, an RC structural damage recognition task framework and a structural mechanical damage image dataset are established. Second, a DL model is selected to conduct the experiments and enhance its performance through transfer learning. Then, a multi-scale correlated structural state assessment procedure is introduced where local, component, structural, and regional scales are linked. Finally, an engineering case is presented to describe the application steps of the method in real-world scenarios, demonstrating its feasibility. Nevertheless, the proposed method lacks comprehensive validation across various geotechnical conditions and detailed structural configurations, which may limit its generalizability. This study has the potential to enhance the efficiency and scope of post-disaster emergency response and contribute to the development of sustainable cities.