**1. Introduction**

Deriving economical, sustainable, and practical solutions without a compromise in infrastructure safety and integrity is a broad challenge in civil and structural engineering disciplines. The unpredictable nature of hazardous events combined with limited resources lead the current practice to inherit performance-based criteria in structural design and evaluation. Therefore, controlling the extent of structural damage rather than exclusively avoiding it, is the trending principle in up-to-date engineering codes and regulations [1,2].

Observing the changes in vibration characteristics of structures with the state-of-the-art sensing and processing tools, structural health monitoring (SHM) technologies attract significant attention in research and industry in the last three decades [3–6]. On the other hand; instrumentation, cabling, operation, and maintenance of SHM systems require labor work, knowhow, and financing; declining the growth rate of SHM use in practice. Especially in the past decade, these drawbacks lead researchers to focus on innovative methods such as noncontact vibration measurement techniques [7–9], wireless

sensor network (WSN) and distributed sensor network (DSN) systems [10–16], as well as smart [17–19], mobile [20–23], and multisensory [24–27] sensing platforms. Eventually, smartphones are adopted into SHM such that their built-in sensors, operating systems, computation, and wireless communication capabilities can perform as structural vibration measurement devices [28–31].

The authors' previous works present the first vibration-based SHM system (CS4SHM) using crowdsourcing power [32] and o ffer multisensory solutions to citizen-induced errors by considering spatiotemporal [33] and directional [34] uncertainties. Without any prior engineering education and background, citizens as uncontrolled SHM device operators can provide a central server system with ubiquitous vibration data. The acquired data is autonomously processed for modal identification which is an important indicator of structural vibration characteristics. Unlike conventional SHM systems, CS4SHM points out unorthodox monitoring issues which are concurrently discussed in the upcoming technological boom "Industry 4.0", the latter phase of digital revolution [35,36]. Collecting the distributed crowd sensed information through a central server and conducting modal identification autonomously, civil infrastructures as physical objects are connected with server-side computing in a massive scale forming a CPS [37–41], or in some cases, an Internet of Things system [42–45]. This highlights a significant potential to evolve from pure theoretical structural response simulation (FEM) to experiment-aided and calibrated models (model updating) in massive scales. In other words, with the help of autonomous, connected, scalable cyber networks; citizen-engaged sensing; digital (FEM predictions) and physical (field measurements) civil infrastructure representations; monitoring systems can be adopted to the upcoming technological innovations.

Aforementioned hybrid models can be used for large-volume analysis to retrieve quick evaluation of structural status. This can be performed by utilizing the modal identification results, calibrating mathematical models, and obtaining the probabilistic failure distribution under a wide range of strong ground motions. Eventually, using identification results as model calibration tools, civil infrastructures' seismic response and structural reliability can be estimated [46,47] to provide the decision makers with the necessary information.

This study presents thelast section of a PhD dissertation dedicated to cyber-physical system applications from structural engineering perspective [48]. The paper extends the outcomes of a crowdsourcing-based modal identification platform by modifying FEMs constructed with limited information. FEMs as cyber representatives of building behavior are updated to minimize the error between the simulated models and the identification results obtained from the "physical objects". Then, the updated models, which represent the actual vibration characteristics to a better extent, are used to simulate structural response under different earthquake motion scenarios. Finally, collecting the simulation results obtained from numerous time history analyses, the demand distribution is evaluated according to the exemplary code and regulation criteria. To summarize, the developed platform presents an innovative mobile CPS by converting the very initial physical vibrations into highly abstracted decision-making information (i.e., service close, retrofitting, and reconstruction) through a digital and multiphase mathematical information processing framework.

Section 2 discusses the experimental and theoretical phases of the methodology followed throughout the study. These phases include information about the testbed bridge structure, the CPS adaptation, model updating, and reliability estimation methodologies. Section 3 presents the application to the testbed and presents the monitoring results including objective function minimization and determination of structural reliability. Finally, Section 4 reviews the overall work, introduces the future goals, and highlights the concluding remarks.

## **2. Materials and Methods**

The methodology presented in this study connects the experimental data obtained from civil infrastructures with the advanced mathematical modeling and analysis procedures. The following subsections introduce the testbed structure, modal identification, FEM updating, and reliability estimation processes as a CPS framework. From sensing to decision making, Figure 1 represents an idealized cyber-physical information processing scheme, with a comparison of the current CS4SHM system. The up-to-date platform is capable of receiving vibration measurements from citizens and conduct modal identification on the server-side. Then, the identification results are collected to set the reference modal analysis values for FEM updating and reliability estimation procedures. These phases are currently conducted through a scripted Matlab and OpenSees [49] loop, and the ultimate goal is to handle these cyber procedures through cloud computing. Nevertheless, in both cases, the decision makers can be provided with the quantitative information regarding structural status. Depending on the changes made to the structural system, the effects will be reflected on the future vibration characteristics which completes the cyclic information processing scheme.

**Figure 1.** Cyber-physical processes of CS4SHM system.

It should be noted that the probabilistic nature given here is mainly concentrated on input ground motion and FE model updating. In fact, the effect of participatory sensing is an indispensable aspect of the system proposed in this paper since citizen engagemen<sup>t</sup> brings numerous uncertainties into the measurements. Previously, it has been shown that actual crowdsourcing results (results from uncontrolled citizens) matched well with the reference identification results [32]. These uncertainties are extensively studied in [48] including spatiotemporal variation of a citizen sensor [33], phone orientation which is subjected to change before, during, and after the measurements [34], and biomechanical distortions caused by human nature [50]. Therefore, in this paper, the main focus is on uncertainties induced by ground motions and FE model parameters.
