Intelligent SHM-GIS Cloud-Based Bridge Monitoring System

In this section, an intelligent SHM-GIS cloud-based framework is introduced and discussed. Individually deployed GIS and SHM tools discussed above may have their


Comparison of Different Bridge Assessment Techniques




Structural Damage Assessment







Risk Assessment Feature












/* indicates in limited scenarios given the use case of the application.

benefits in bridge monitoring, specifically for small-scale applications. However, in large-scale deployments considering the size and the complexity of the application, it may result in higher overall cost and less accurate results. It is, therefore, recommended that combined use of the tools on a cloud-based platform could enhance their performance. Furthermore, cloud-based platforms allow collaboration of different stakeholders enabling each party to add and edit on top of the existing data, and performing simultaneous analysis.

The nature of SHM systems in terms of sustainability already considers the three components of a sustainable approach, i.e., economic, social, and environmental. These include a reduction in traffic delays and downtimes, which subsequently lead to lower carbon dioxide emission and, lastly, the expected economic loss. These are mainly related to the network aspect, but the same can be said to the structure itself, i.e., bridge. Bridge monitoring can provide useful information in terms of the remaining helpful time and any maintenance that may be necessary for the future, which can help minimize the costs and maximums the life expectancy by early retrofitting or reconstructing.

Similarly, GIS can help bridge managers to have a better understanding of the structures and their behavior under different conditions. It can deliver a decisionmaking platform (Bane§ et al., 2010) for the risk assessment of bridges and their cascading failures on the network, thus offering a complete management tool that could provide sets of strategies depending on the application use.

The new paradigm shift to cloud computing and web-based applications marks the SHM-GIS cloud-based platform a necessity in today’s technological world. Not only it provides the core functionality of the tools, but instead, it goes further to expand its roots for even more cost-effective, efficient, and sustainable solution in bridge prognosis and diagnosis. Synergistic use of SHM and GIS can develop or update earthquake models on the fly and provide a more accurate damage estimate of the bridge and its effect on the network.

The proposed SHM-GIS cloud-based system architecture is, therefore, presented in Figure 7.5. The sensory subsystem layer acts as the data acquisition where it collects the data from bridges. The collected information is then transferred to a server via a different form of communication standards such as Wi-Fi, Bluetooth, and cellular and later uploaded on the cloud. Due to the enormous size of the acquired data for any given time history chiefly in the extensive application of bridge monitoring, storage methods need investigation. The issue of big data and storage has led to the creation of different file structure format. Standard file formats for storing large amounts of data are (1) HDF4 (Hierarchical Data Format) and (2) netCDF5 (Network Common Data Format). However, due to the file structure of these formats, they are not ideal in cloud computing. Many alternatives with their strengths and weaknesses are present in Matthew Rocklin (2018) webpage. HDF5 (Hierarchical Data Format version 5) can be an ideal solution in this case for storing multi-dimensional data. Bridge information such as geometry and location, and network description such as highway information and traffic information are stored in an object-related database management system (ORDBMS). PostgreSQL, with the extension, PostGIS for handling spatial data, is the common database management system (DBMS) for

Framework of the system architecture

FIGURE 7.5 Framework of the system architecture.

SHM applications. PostgreSQL is an open-source DBMS that is well developed and intuitive. The relationship between the sensor data and the structural/network elements is also a one-to-many relation.

The cloud service for this system relies on infrastructure as a service (IaaS) type. IaaSs are often low-cost, more accessible, and faster options over different cloud services enabling storage resiliency, frequent backup, and high level of automation. Deciding which cloud provider to use depends on the performance and uptime required from the provider. A typical solution for cloud computing is Google Cloud and Compute Engine. Other services, such as Microsoft Azure and Amazon Web Services (AWS), are also available. These data then proceed into performance analysis and monitoring of the bridges. Depending on the data type (vibration, displacement, image, etc.), different algorithms can define the damage state in the given earthquake scenario.

Incorporating the network data such as traffic delay and routing info into the database can enable the employment of a cloud GIS platform capable of visualizing, analyzing, managing, and monitoring bridges and the effects of failure of them on the transportation network. Using this information and a simple risk formula that includes direct costs such as structural loss, network loss, and indirect loss, it can provide a decision-making platform for pre- and post-earthquake disaster scenarios. The advantages of this SHM-GIS cloud-based system are as follows:

  • • Utilizing open-source and free software and system providers
  • • Ability to add/remove or change any information without the problem of proof checking for errors
  • • The flexibility of the system in any application use (using a small or large number of sensors)
  • • An intuitive and low-cost solution for bridge monitoring (especially for bridges owners)
  • • The scalability of the system in terms of the location and the size of the application.

Moreover, risk assessment based on dynamic changes in the model can also serve in the system. As parameters of the model change throughout time, real-time risk assessment can assess the performance of the bridge under future loads. The data from traffic and future loading can predict the future state of the bridge, aiding bridge owners to decide about retrofitting or reconstructing all or some parts of bridge elements.

The whole system, from the data acquisition, DBMS, and user interface, can be programmed with the open-source Python programming language. Web applications, as well as mobile applications for viewing and extracting information, can also be implemented for easier and faster utilization of the data. The ability of information exchange and information sharing with other software and services is another advantage that distinguishes this from other similar systems (Ellenberg et al., 2015; Eschmann et ah, 2012; Sankarasrinivasan et ah, 2015). A summary of the traditional SHM-GIS damage assessment and the cloud-based variant is tabulated in Table 7.2. The next section brings a recent technological implementation and aerial devices, which provide an efficient synthesis of GIS and SHM domains.

Machine Learning in SHM Application: A Complementary Addition

Given the amount of data gathered from many different things, it is important to understand the pattern that underlines it. Day by day with increase in complexity of structures, without automatic (sometimes semiautomatic) processes to discover patterns using computer, such tasks would be infeasible and impractical. ML is considered as tool to recognize/classify information based on a learned pattern through the use of different algorithms. In general, ML algorithms are based on either (1) statistical, (2) neural, or (3) synthetic approaches. The first two approaches are generally considered as the main pattern classifiers for SHM (Bane§ et al., 2010). There are many works utilizing ML. For example, Cao et al. (2018) developed a piezoelectric impedance measurement for an effective structural damage identification through an inverse analysis. Similarly, in a study by Moore et al. (2012), crack identification in a thin plate was achieved by model updating.


Summary of the Traditional and New Novel SHM System








Mobility (Easy Accessibility)

Maintenance and Management

Open and Interoperable



Jeong et al. (2017) and Shi etal. (2002)







This study framework







With the advent of ML and statistical pattern recognition algorithms, a new level can be added to the Rytter (1993) four-stage damage identification. Type of damage or classification of damage is the level that is possible through the use of ML algorithms. This new step lies between step 2 and step 3 introduced by Rytter. To illustrate this, Figure 7.6 depicts the five-stage damage identification in SHM application given the domain and level of difficulty.

Given that both damage and undamaged information are available, a supervised learning algorithm can effectively go through all five levels of damage detection. This requires many data to be readily available from the sensing systems or the physical-based models and the experiments. This is not possible in many applications, and the current information for damage sate is limited, if not, unavailable. For such situations, there exists a method called unsupervised learning. In this mode, instead of learning the models and train based on the data, a rather simple approach, novelty, or outlier detection is applied (Casas and Cruz, 2003). Figure 7.7 illustrates a statistical pattern recognition model for a typical damage assessment scenario utilizing ML. Moreover, Table 7.3 shows the current reviews on ML utilization on SHM application.

ML can augment SHM in many aspects which the old system is incapable of. For example, environmental and operational variabilities oftentimes are not considered but have proven that they can greatly influence in-service structures (Sohn, 2007). Including these effects by leveraging the power of ML can definitely help SHM application achieve better level of detection. Moreover, ML and deep learning can be particularly useful in bridge monitoring applications which are combined with GIS and remote sensing tools that utilize machine vision for anomaly detection or as tools in data analytics inside the GIS package.

A typical ML model

FIGURE 7.7 A typical ML model.


Works on ML Utilization in SHM






Application of ML/Deep Learning



Machine Vision Consideration


Applications (UAV, VR, AR, etc.)

Fan and Qiao. (2010)and Gomes et al. (2019)

Ye et al. (2016)

An et al. (2013)

Moughty and Casas (2017)

Kerle et al. (2019)

UAV Only

/* indicates a little information.

< Prev   CONTENTS   Source   Next >