Monitoring buildings performance using FEA and ML based on the data acquired by InSAR. A case study of Vittoriano building, Rome

Monitoring the performance of buildings and monuments with high cultural and artistic importance is critical for preserving the built heritage. Installing measurement devices is generally prohibited on such structures, and their behavior is assessed by analytical methods. This paper presents a novel methodology for monitoring structural performance through advanced Finite Element Analysis (FEA) and Machine Learning (ML) using the data acquired by InSAR. A cracked wall of Vittoriano building located in Rome, Italy is considered for applying this approach. The paper is organized in three main parts: (i) an advanced 3D numerical model of the wall was developed in ABAQUS. The Concrete Damage Plasticity (CDP) model was utilized for defining the masonry material based on the macro-modeling approach. A nonlinear dynamic analysis was performed to determine the wall’s response to the ground settlement reported by InSAR. The model was verified by comparing the stress concentration with the real cracks observed on the wall. (ii) In the second part, an ML-based technique was implemented to predict ground movement in both short- and long-term periods. (iii) Finally, the process of developing a digital twin for monitoring wall’s performance using FEA, ML and InSAR data is clearly described. Overall, the results exhibited a high accuracy of both FE model and ML-based technique for evaluating buildings’ behavior over time.