Dynamic Identification of confined masonry for existing buildings via AI techniques

Regarding the conservation and retrofitting of existing structures, several methodologies have been studied and new techniques have been implemented over the last decade. Nowadays the improvement in Structural Health Monitoring (SHM) technologies, sensor data collection and specific survey procedures lead to a deep knowledge of the status of existing structures and allows to provide accurate modelling. Specifically, the assessment of the dynamic behaviour of existing buildings can be considered as one of the main tasks for researchers. Within this context, two confined masonry buildings located in the south of Italy were adopted as case studies with the aim of conducting the dynamic identification and assessing their dynamic behaviour. Different configuration layouts of the sensors have been implemented to perform the monitoring phase using the optimal sensor placement procedure in order to detect the modal parameters. The in situ experimental data has been analysed using Operational Modal Analysis (OMA). Additionally, clustering unsupervised machine learning has been performed for processing the experimental data and identifying main modal parameters. To obtain a feasible calibration of the structural parameter of the models, AI techniques have been adopted and an optimization procedure has been implemented with the aim to reduce the discrepancy between the numerical frequencies obtained from the models and the experimental ones. Thus, a F.E. model for each case study has been prepared by using the SAP2000 F.E code taking into account the mechanical and geometrical properties of the buildings derived from the survey. Then, Genetic Algorithm has been used for the implementation of the optimization procedure in MATLAB. At each step of the process, the structural parameters of the numerical models have been dynamically changed until the fulfilment of the defined stopping criteria. The result of this study points to a standard procedure to approach the inspection, identification, and calibration phases of FE models leading to obtaining a feasible procedure for the dynamic identification of confined masonry buildings. The thesis presents the following organization: in Chapter 1, the most promising techniques in the field of model updating and the most common calibration problem will be discussed. In Chapter 2 information regarding confined masonry, its behaviour and configuration is presented. For Chapters 3 and 4, a generic description of both case studies as well as the results obtained by the survey campaign will be introduced, respectively. Chapter 5 shows an introduction to OMA and the sensor layout configuration adopted during the acquisition phase. Chapter 6 presents the processing and interpretation of the dynamic test. Chapter 7 introduces some details regarding the Finite Element model and the preliminary consideration of their accuracy. In Chapter 8, the outcomes obtained by the modal updating procedures complemented with the optimization process will be discussed and details about the setting criteria of the F.E model initial calibration will be clearly defined. In conclusion, the main results of the overall thesis will be summarized, and new frontiers of this research will be suggested.