Abstract— Structural identification based on experimental data coming from rapid and/or permanent monitoring is continuously increasing its relevance in the process of the structural safety assessment. The paper discusses the direct and inverse problem of structures subjected to base excitation. Recent results in time domain modal identification are summarized for their relevance in the treatment of data coming from either rapid testing or permanent monitoring activities realized after the 2009 L’Aquila earthquake for two case studies. In the first one, the vibration measurements acquired in a two-day testing campaign on a damaged reinforced concrete building belonging to the Engineering Faculty of the University of L’Aquila have been used to directly compare the identified modal features with a modal model constructed by finite elements, representative of the structural behavior of the building in the early post-earthquake conditions. In the second one, the seismic response obtained through a permanent structural health monitoring system deployed on a monumental masonry building, the Basilica of S. Maria di Collemaggio, has been used to extract the modal parameters which compared with results coming by a finite element model, have furnished valuable information to evaluate the effectiveness of the temporary scaffolding structures. The study evidences that the reference-based combined deterministic-stochastic subspace identification of structures heavily damaged by earthquakes is a valuable tool to insight into noisy data acquired in difficult operational conditions. Finally, the specific technology developed along the research, for easy structural response measurements, is also presented, and its peculiar characteristics and drawbacks are finally discussed. I. INTRODUCTION In recent years the modal identification has played an increasingly important role in the large activities and design of structural health monitoring. The identification process is required to finely calibrate a parametric model in order to match the experimental data [1]. Methods for the estimation of the modal parameters have been studied over past years to solve the so- called system identification problem [2]. The research in the field of the structural identification has invested a lot of effort in the development of reliable system identification algorithms for the recognition of the main modal parameter. The main analysis techniques can be divided in two group: Experimental Modal Analysis (EMA) and Operational Modal Analysis (OMA). In the first case the structure is forced by a known and/or measured external action and the structural response is also measured. Subsequently the modal model can be identified from these input-output data. One of the first EMA methods was the Peak Picking PP [3]. Some of the main drawbacks of these procedures are the following: in large civil engineering structure the contribution to the dynamic response due to the artificial force may be quite low with respect to that produced by other types of environmental excitation, such as wind or traffic. Another problem is that sometimes the lowest fundamental structural frequencies may fall outside the range of frequencies of the excitation. For these reasons in the last years procedures based on output-only measurements (OMA) have been developed. In general, methods for the vibration-based identification of modal parameters can be classified into two categories, depending on whether they operate in the time domain or in the frequency domain. However, the amount of information available is absolutely independent of whether the data are represented in the time or the frequency domain. Even if different information can be derived on the robustness of the identified modal model, recent scientific studies have shown that in controlled laboratory condition it is possible to identify very similar models from the two approaches [4]. The choice of one or the other approach should be based on the objectives to be achieved and on the case of study. Two of the primary methods for output-only vibration-based identification of modal parameters are the Enhanced Frequency Domain Decomposition (EFDD), operating in the frequency domain and the Stochastic Subspace Identification (SSI), operating in the time domain. The OMA techniques also have their shortcomings such as the impossibility of determining the modal scaling factors, even if they can determined by performing two different OMA tests, adding a known mass to the structures during the second test [5]. Indeed, sometimes for the damage identification is required the scaled modal shape. For these reasons methods based on combined deterministic-stochastic system identification procedures, in which the total structural response is due to measured and unmeasured force, have been developed. These specific techniques fall in a category called Operational Modal Analysis with eXogenous input (OMAX). Among these methodologies a suited procedure, called Combined deterministic-stochastic V. Gattulli and F. Potenza are with the Department of Civil, Construction-Architectural and Environmental Engineering, University of L’Aquila, L’Aquila, 67100 Italy (the first is the corresponding author to provide phone: +39-862-434511; fax: +39-862-434548; e-mail: vincenzo.gattulli@univaq.it, francesco.potenza@cc.univaq.it). F. Graziosi, F. Federici and A. Colarieti are with the Department of Information Engineering, Computer Science and Mathematics, University of L’Aquila, L’Aquila, 67100 Italy (e-mail: fabio.graziosi@univaq.it, fabio.federici@graduate.univaq.it, andrea.colarieti@univaq.it) M. Lepidi is with the Department of Civil, Chemistry and Environmental Engineering, University of Genoa, Genoa, 16145 Italy (e-mail: marco.lepidi@unige.it). Role and perspectives of modal identification in rapid and permanent structural monitoring after an earthquake Vincenzo Gattulli, Fabio Graziosi, Fabio Federici, Francesco Potenza, Andrea Colarieti, Marco Lepidi