Introduction to Monitoring of Bridge Infrastructure Using Soft Computing Techniques

Abstract

More than a billion structures exist on our planet comprising a million bridges. A number of these infrastructures are near to or have already exceeded their design life and maintaining their health condition is an engineering optimization problem. Besides, these assets are damage-prone during their service life. This is due to the fact that different external loads induced by the environmental effects, overloading, blast loads, wind excitations, floods, earthquakes, and other natural disasters can disturb the serviceability and integrity of these structures. To overcome such bottlenecks, structural health monitoring (SHM) systems have been used to guarantee the safe functioning of structures to make satisfactory decisions on structural maintenance, repair, and rehabilitation. However, conventional SHM approaches such as virtual inspections cannot be used for structural continuous monitoring, real-time and online assessment. Therefore, soft computing techniques can be significantly used to mitigate the aforesaid concerns by handling the qualitative analysis of the complex real world behavior. This chapter aims to introduce the optimized SHM-based soft computing techniques of bridge structures through artificial intelligence and machine learning algorithms in order to illustrate the performance of advanced bridge monitoring approaches, which are required to maintain the health condition of infrastructures as well as to protect human lives.

1. Introduction

Infrastructure monitoring is one of the most significant applications of cities [1]. This is likely due to the fact that smooth functioning of cities is a vital need by providing safe and efficient infrastructure. Eventually, complex, large, and expensive engineering assets, i.e., high-rise buildings, long-span bridges, dams, oil platforms, wind turbines, offshore structures, roadways, and rail tracks are designed to last long [2, 3]. For example, bridges are normally built to have a lifespan of 50 years [4]. However, many of them are near to or have already exceeded their design life. For example, according to ASCE 2021 infrastructure report card, 42% of all bridges across the United States are at least 50 years old, and they are considered structurally deficient; they are in poor condition and in need of repair. Besides, these assets are damage-prone during their service life. This is due to the fact that different external loads induced by environmental effects, overloading, blast loads, wind excitations, floods, earthquakes, and other natural disasters can disturb the serviceability and integrity of these structures.

Structural and materials design is a highly iterative process for the optimal design of infrastructures. Even the simplest structures and materials are composed of multiple elementary structural components which can lead to various optimal designs [5]. Likewise, maintaining the health condition of infrastructures is also an engineering optimization problem. This is because it is not easy to find an exact solution [6]. For instance, evolutionary techniques have been applied as a part of the procedure of achieving the exact solution. Therefore, several metaheuristic algorithms such as genetic algorithm, particle swarm optimization, ant colony optimization, and imperial competitive algorithm have been developed to solve a variety of engineering optimization problems in a transdisciplinary field of engineering, so-called structural health monitoring (SHM). Bridge monitoring and optimization are significant areas of SHM and soft computing, respectively.

SHM and soft computing techniques as powerful tools can be significantly used to mitigate the aforesaid concerns by planning scheduled maintenance, control, and management of infrastructures. Based on the above explanations, this chapter aims to introduce the optimized SHM-based soft computing techniques of bridge structures through artificial intelligence and machine learning algorithms in order to illustrate the performance of advanced bridge monitoring approaches which are required to maintain the health condition of infrastructures as well as to protect human lives.

2. Bridge as an iconic infrastructure

Civil engineering is an ancient profession and one of the most noble in the world. Many structures are monuments to civilization and last for centuries, becoming pilgrim and tourist attractions. Many of these huge monumental structures, being one-off in nature, have warranted large realization times. Some of them have taken centuries to build. The above structural attributes, being passed down from generation to generation, necessarily had certain design processes, motivations, and contexts for posterity to appreciate and admire in terms of form, esthetics, and sustenance [7]. Peter Rice [8], one of the most original and influential engineers of the twentieth century was certain that if the method of manufacturing was true to the project’s nature, it would result in a structure capable of producing the emotional response intended by the designer. In the editorial headline “Winning the Emotional Argument,” New Civil Engineer editor Mark Hansford commented in 2017: “civil engineering professionals now, more than ever, need to engagingly present the broader benefits of their infrastructure projects, highlighting the direct impacts they are having on society” [9].

Aside from the implicit recognition that successful infrastructure projects now employ a vast range of complementary disciplines—including engineering, planning, architecture, landscape design, ecology, and many others, and are no longer simply “heavy engineering,” it seems self-evident that the bigger the project, the greater the need to demonstrate the benefits to society. It is understood that value must be added—beyond the base metrics of people movement—in terms of lasting social, economic, and environmental benefits [10].

A report by Beade-Pereda [11] stated that bridges link previously separate geographic areas, defying gravity and transforming the landscape. They bind communities, acting as connectors of people, inviting interaction and integration. Bridges are a paradigmatic case of human transformation of nature, symbols of union, progress, and often innovation. Bridges across physical, cultural, and spiritual barriers, frequently becoming landmarks or even icons. They are much more than a piece of infrastructure and the design of such emotional, prominent, and long-lasting constructions should go beyond their main function as structures that link areas, and should always aspire to improve the quality of the built world. At International Association for Bridge and Structural Engineering (IABSE) “Future of Design” event in London on September 8, 2016, Professor Enzo Siviero, named “The Bridge-man” said: “I always say if a bridge was a woman I could marry it!”. This hilarious example shows the connection between bridges, emotions, culture, and people [12].

Structural and material design is a highly iterative process for the optimal design of infrastructures. Even the simplest structures and materials are composed of multiple elementary structural components, which can lead to various optimal designs [5]. The evolution in bridge design offers an example of this influence between structural components, materials, and boundary conditions on the design (see Figure 1). As can be observed from this figure, the designers have integrated esthetics into the design. Another side of this evolution is dramatic changes in the lighting of bridges in the last decades, which have gained insight into the visual and emotional effects [22].

Figure 1.

Evolution of bridge design: Influence of materials on esthetics and structural design. Roman aqueduct in Spain, 50 A.D. [5]. Ponte Vecchio bridge in Italy, 1345 [13]. Büyükçekmece bridge in Turkey, 1567 [14]. Khaju bridge in Iran, 1650 [15]. Pulteney bridge in UK, 1769 [12]. Maria Pia bridge in Portugal, 1877 [5]. London Tower bridge in UK, 1894 [14]. Chengyang bridge in China, 1912 [16]. Salginatobel bridge in Switzerland, 1930 [17]. Sydney Harbour bridge in Australia, 1932 [14]. Golden Gate bridge in USA, 1937 [5]. Coronado Girder bridge in USA, 1967 [5]. Magdeburg Water bridge in Germany, 2003 [18]. Seri Wawasan bridge in Malaysia, 2003 [12]. Helix bridge in Singapore, 2004 [14]. Millau Viaduct bridge in France, 2004 [19]. Shenyang Sanhao bridge in China, 2008 [7]. Rewa Rewa bridge in New Zealand, 2010 [20] Lower Hatea Crossing in New Zealand, 2011 [10]. Rzeszów circular footbridge in Poland, 2012 [21].

3. Bridge failure statistics: Where? Why? When?

Collapse of bridge structures is one of the worst infrastructural disasters. For example, Figure 2 shows several spectacular bridge catastrophic collapse cases in the world. It has been reported that the most bridge collapses occurred in North America and Europe, especially in the USA (see Figure 3). Extreme conditions under natural hazards, design error and wrong assumption in load combination or ground condition, overloading, impact of vehicles, poor workmanship and inadequate maintenance action, vandalism (e.g., fire and explosion), deterioration, corrosion and fatigue in materials, and limited knowledge due to unknown phenomena are possible causes of bridge collapse. For the sake of clarity, Figure 4 shows a summarization of the causes of the bridge failure in China. Bridge failure can be occurred both during the service life of the infrastructure or in the construction stage. For example, Figure 5 tries to illustrate the number of bridge collapses and casualties between 2009 and 2018 [31]. It should be noted that it was difficult to find out more about the prevalence of bridge collapse in Africa, Asia, and South America. Figure 6 lists a variety of factors, which are essentially not considered in the calculation of the probability of failure.

Figure 2.

Examples of bridge catastrophic failure. (a) Tacoma Narrows Bridge in Washington, USA: (Left) opened on July 1, 1940 and (Right) collpased on Novermber 7, 1940 [23, 24]. (b) Hanshin Expressway bridge in Kobe, Japan: (Left) before and (Right) after collapse on January 18, 1995 [25, 26]. (c) I-35W Bridge in Minneapolis, USA: (Left) before and (Right) after collapse on August 1, 2007 [27]. (d) Morandi bridge in Genoa, Italy: (Left) before and (Right) after collapse on August 14th 2018 [28, 29]. (e) Nan Fang’ao bridge in Taiwan: (Left) before and (Right) after collapse on October 1, 2019 [30].

Figure 3.

Geographical origin of recorded bridge failures. (a) By country. (b) By continent.

Figure 4.

The distribution of bridge failure causes between 2009 and 2018. (a) Percentage of collapse reasons. (b) Proportion of natural factors and anthropic factors leading to bridge failures.

Figure 5.

Number of bridge collapses and casualties between 2009 and 2018 [31].

Figure 6.

Potential factors influencing the observed failure frequency of bridges.

For better understanding, Figure 7 presents the catastrophic bridge failures from 1967 until now in the U.S. along with the cause of failures (e.g., corrosion, scour, human error (i.e., design error and construction error), fire, etc.), number of injuries and fatalities.

Figure 7.

Causes of catastrophic bridge failures in the USA. (a) Silver Bridge in Ohio, collapsed on 1967. (b) Mianus River Bridge, Greenwich, collapsed on 1983. (c) Schoharie Creek Bridge—Albany, NY, collapsed on 1987. (d) Route 69 Tennessee River Bridge—Clifton, TN, collapsed on May 1995. (e) I-35W Bridge in Minneapolis, collapsed on August 2007. (f) I-580 Connector Ramp –Oakland, California, collapsed on April 2007. (g) Prosperity Pedestrian Bridge—FIU/University City, Florida, collapsed on March 2018. (h) I 40 Hernando deSoto Bridge Memphis, TN, crack found on May 2021.

4. Structural health monitoring (SHM)

Engineering assets such as high-rise buildings, long-span bridges, dams, oil platforms, hydraulic structures, wind turbines, transmission towers, ships, offshore structures, aircraft, and rail tracks may experience damage during construction or while in service induced by different reasons such as common weakening of material properties, fatigue, aging, delamination, wear, corrosion, creep, environmental effects (e.g., microstructural defects, cracking, thermal stress, residual stress, instability, and fastening or adhesive faults), overloading, changes in loading patterns or various unexpected causes such as wind excitations, earthquake, vehicle impact or blast loads during their service life, which can critically disturb their integrity, serviceability, and safety [32, 33]. Therefore, the unpredicted structural failure in such assets can produce catastrophic collapse, economic costs, human injuries, and death.

SHM is a transdisciplinary area of engineering applied to guarantee the operational safety and structural integrity of the materials, different components, or whole structure [34]. Engineers define health monitoring as the measurement of the operating and loading environment and the critical responses of a structure in order to track and evaluate the symptoms of operational anomalies and deterioration or damage that may impact service or safety reliability. The functionality of SHM systems is extremely similar to the human nervous system, as shown in Figure 8. The nervous system of humans consists of a complex collection of nerves and specialized cells and the main processing unit (brain). The nerve cells transmit signals between different parts of the body and the brain. The brain is the main control unit for receiving and processing information as well as issuing instructions. In the same manner, SHM consists of a sensory network to gather information and a control unit for data processing and decision making [35]. Broadly speaking, the aims of conducting an SHM system are as follows: (1) To determine the current condition, (2) To predict future behavior, and (3) To early detect the structural damages.

Figure 8.

Similarity of SHM and human nervous system.

The first bridge health monitoring was conducted in 1937 on the Golden Gate Bridge in San Francisco. Likewise, the direct application of data mining in structural damage identification was started in 2014 [36]. An overview of SHM and soft computing evolution over the years is shown in Figure 9. Eventually, the monitoring process in SHM generates massive data. Hence, precious information has to be acquired from unprocessed datasets [37]. However, data analysis of the generated big data obtained from the sensor network is a challenging task [36, 38]. To overcome this, data mining can therefore be employed to develop damage detection schemes [39, 40]. In computer science, data mining has become a research hotspot in recent years [41]. Consequently, data mining is offering a bright future ahead for other inquiries in various areas, such as aerospace, civil, industrial, and mechanical engineering. It is because data mining has a key role in the extraction of valuable information from different databases [6, 42].

Figure 9.

An overview of SHM and soft computing evolution over the years.

With demanding needs to generate a massive volume of datasets, there has been a revolution in measuring and monitoring systems in the 1990s such as the development of sophisticated signal technologies, wireless networks, optical sensors, and global positioning systems. In this direction, the growth in the number of sensors installed on several important bridges worldwide during the past 20 years is shown in Figure 10 [43].

Figure 10.

An example of the evolution of the number of sensors for bridge monitoring during the past 20 years, adopted from [43].

5. Soft computing

Soft computing includes a series of strategies, that aim to exploit tolerance for imprecision, uncertainty, and partial truth to establish robustness and flexibility along with low solution cost. Major soft computing techniques and topics are summarized in Figure 11. In bridge monitoring, different soft computing techniques have been utilized for damage detection and system identification. The followings present the most commonly used soft computing applications for bridge monitoring.

Figure 11.

Soft computing techniques.

Applications of data mining in SHM have recently been reported [44, 45], though due to the novelty of data mining, the application of data mining in SHM is still controversial and is not as much as expected. Therefore, it seems necessary studies are required to advance the data mining application in SHM. One of the most widespread systematic data mining tools is Cross Industry Standard Process for Data Mining (CRISP-DM), which was introduced by a consortium of several companies such as National Cash Register (NCR) System Engineering Copenhagen from the USA and Denmark, Integral Solutions Ltd. (ISL)/SPSS from the USA, Daimler Chrysler AG from Germany and an insurance corporation in the Netherlands, called OHRA [46, 47]. A generalized form of CRISP-DM based on the SHM system has been developed by the authors of [48]. The proposed data mining-based damage identification approach consisted of six new defined stages: target identification, data exploration, database construction, pattern identification, pattern evaluation, and knowledge extraction. At the first stage, specimen description and experimental setup have been presented using a slab-on-girder bridge structure (see Figure 12). In the second stage, vibration data were collected from experimental modal analysis of healthy (baseline) and damaged structures. Analysis of collected data was done in the third stage to generate datasets using the first four flexural modes and all corresponding mode shape values of double-point damage cases as inputs for next stage, which was pattern identification. In this stage, Artificial Neural Network (ANN)-based Imperial Competitive Algorithm (ICA) was employed to train datasets and build a model for damage identification of the structure. Then, in the fifth stage, model performance was assessed using evaluation methods and ANN approach. Finally, the last stage extracted valuable knowledge and damage identification.

Figure 12.

Experimental test of the lab-scale bridge [37].

A multi-layer perceptron ANN-based damage detection of truss bridge joints was proposed by Mehrjoo et al. [49]. The proposed network was conducted using a single hidden layer and 729 training patterns obtained from the first five modes of the structure. It should be noted that the stiffness reduction of members was considered as damage in this study. Figure 13 illustrates the fatigue damage in truss bridge joints. Around 40% of these types of structures usually experience fatigue damage in their joints during their service life. Louisville bridge was used as a test specimen for this research. The standard back-propagation (BP) algorithm with 75,000 epochs and the root mean square (RMS) was employed to train the networks and evaluation of patterns, respectively. The average error of the predicted damage percentage value for all joints was 1.28%. The authors summarized that their presented methodology was effective in system identification of truss bridges.

Figure 13.

Fatigue damage in truss bridge joints in Louisville bridge truss [49].

A bridge monitoring scheme using operational modal analysis was developed by [50] through a hybrid Fuzzy Krill Herd approach. The proposed fuzzy logic-based SHM diagram is displayed in Figure 14a. Two types of bridges, i.e., Banafjäl bridge in Sweden and the Tirehrood bridge in Iran, were considered as test specimens for this paper. The damage scenarios were presented by the output of the fuzzy logic-based SHM approach, as shown in Figure 14b. The outcomes revealed the proficiency of the proposed approach in achieving precise knowledge in the existence of noisy data.

Figure 14.

(a) The proposed fuzzy logic-based SHM diagram, and (b) damage scenarios by [50].

Padil et al. [51] proposed a PCA-based non-probabilistic technique. The proposed method was verified using a big Frequency Response Function (FRF) matrix comprising 1200 FRFs with 512 frequency points obtained from intact and damaged simply-supported steel truss bridge model (see Figure 15a). To aid the aim, a number of damage scenarios were considered by cutting the structural members, i.e., M1 and M2 in the main girder bar, and W1 in the web bar, as shown in Figure 15b. In this study, it was shown that the results of hybrid PCA were better than traditional PCA.

Figure 15.

(a) Experimental model of a steel truss bridge and (b) imposed damages [51].

The application of deep learning in SHM was implemented in [52] by different PCA-based methods, i.e., the deep principal component analysis (DPCA), nonlinear principal component analysis (NLPCA), and kernel principal component analysis (KPCA). The damage-sensitive features of Z24 and Tamar Bridges were used to evaluate the applicability of the aforementioned algorithms (see Figure 16). The DPCA showed the best performance.

Figure 16.

(a) Z24 bridge, and (d) Tamar suspension bridge [52].

In [36], an SVM model was carried out using four different kernel functions including the Gaussian radial basis function (RBF), Polynomial, Sigmoid, and Linear kernel functions. Experimental modal analysis of a bridge structure was performed to generate the modal parameters as the input database for the model creation. A number of damage cases were conducted to predict the damage severity. As shown in Figure 17, amongst all patterns, SVM-Polynomial achieved the most accurate predicted outputs. To offer an explanation, kernel functions were used in order to bring the data from a lower dimension to a higher dimension. To this end, SVM classifier divided the data with a new plane, i.e., hyperplane. Therefore, despite the better learning power in RBF kernel amongst others, this local function could not provide a satisfying dissemination efficiency. Instead, the polynomial kernel, which is a global function, performed a superior data dissemination strategy. Nonetheless, the learning process of polynomial function experienced a lower level of learning capacity.

Figure 17.

SVM results for composite bridge structure using different kernel functions [36].

A finite element model reduction methodology using a Bayesian inference was developed in [53] for structural bolted-connection damage detection. A novel likelihood-free Bayesian inference method for structural parameter identification has also been proposed in [54]. To aid the aim, the numerical simulations of a three-dimensional bridge structure were carried out to validate the performance and accuracy of the proposed approach (see Figure 18). It was concluded that the reported Bayesian model updating technique was capable of predicting the posterior probabilities of unknown structural parameters.

Figure 18.

Finite element model of a bridge structure [54].

The Nam O Railway Bridge is a large-scale steel truss bridge located on the unique main rail track from the north to the south of Vietnam (see Figure 19). In [55], the experimental measurements obtained from this bridge were carried out under ambient vibrations using piezoelectric sensors, and a finite element model was also created in MATLAB to represent the physical behavior of the structure. By model updating, the discrepancies between the experimental and the numerical results were minimized. For the success of the model updating, the efficiency of the optimization algorithm was essential. Particle swarm optimization (PSO) algorithm and genetic algorithm (GA) were employed to update the unknown model parameters. The authors claimed that not only the PSO result showed better accuracy but also reduced the computational cost compared to GA. This study focused on the stiffness conditions of typical joints of truss structures. According to their results, the assumption of semi-rigid joints (using rotational springs) could most accurately represent the dynamic characteristics of the truss bridge.

Figure 19.

Nam O railway bridge [55].

A damage assessment based on ACO was proposed in [56] from changes in natural frequencies. A plane truss structure was considered in this study to validate the efficiency and robustness of the presented methodology. The authors reported that their method was capable of correct damage assessment even with a noisy dataset. An inverse vibration-based approach using ACO and natural frequencies changes has also been carried out in [57]. The authors of this research made a comparison between ACO and PSO. Furthermore, the performance of a continuous ACO and PSO was also evaluated in [58] for damage detection of plane and space truss structures based on frequency and mode shapes-based objective function. The details of the truss structure are displayed in Figure 20.

Figure 20.

25-member plane truss [58].

A regression-based damage detection approach was developed by [59] using the natural frequencies of the Z24 Bridge. In this research, the traditional regression, as well as the developed regression, was applied to the illustrative structures for identifying the existence of damage. It was established that both methods could detect the presence of damage. However, better outcomes were acquired by the developed regression-based damage detection approach. Another regression model has been developed by [60] as an up-to-date damage detection scheme in a field-monitored bridge. To aid the aim, the Infante D. Henrique Bridge in Portugal has been continuously monitored since 2007 using two synchronized Global Positioning System (GPS)-based data analyzers, temperature and vibration accelerometers (see Figure 21). It is worth noting that the recorded data were less contaminated by noise through GPS connection. The measured natural frequencies change corresponding to the damage scenarios by reduction in the vertical bending inertia of the arch were employed in this work to obtain damage-sensitive features.

Figure 21.

SHM of the infante D. Henrique bridge (a) bridge image, and (b) sensors locations (a and T represent acceleration and temperature, respectively) [60].

Table 1 summarizes the latest applications of artificial intelligence methods in bridge health monitoring.

Application	Algorithm	References
Faulty data detection	Convolutional Neural Network (CNN)	[61]
Shear loading detection	Memory-Augmented CNN	[62]
Concrete crack detection	Faster region-based CNN	[63]
Data anomaly detection	One-dimensional CNN	[64]
Rust grade recognition	Ensemble CNN with voting strategy	[65]
Data anomaly detection	Deep Neural Network (DNN)	[66]
Automated crack evaluation	Deep learning-based segmentation	[67]
Feature selection	ANN	[68]
Fatigue damage detection	ANN	[69]
Probabilistic Damage detection	ANN with Bayesian	[70]
Pattern recognition	ANN	[71]
On-line early-damage detection	ANN	[72]
Damage detection	Extended Kalman filter-based ANN	[73]
Damage quantification	ANN	[74]
Temperature effect removal	Auto-associative ANN	[75]
Damage detection	Auto-associative ANN	[76]
Damage classification	ANN	[77]
Continuous online damage detection	PCA	[78]
Bridge health monitoring	Robust PCA	[79]
Damage identification	Double-window PCA (DWPCA)	[80]
Data-driven damage detection	Fixed moving PCA	[81]
Nonparametric damage detection	PCA	[82]
Missing data estimation in SHM	Probabilistic PCA	[83]

Table 1.

Summary of bridge monitoring using advanced computational techniques.

6. Conclusion

A bridge is one of the most symbolic, important, as well as expressive infrastructures worldwide for social and economic activities of mankind where it serves as the crucial link in the transport network. Therefore, condition assessment and damage detection of this asset is frequently required to guarantee the safe functioning of the infrastructure. To do so, SHM systems have been applied to make satisfactory decisions on structural maintenance, repair, and rehabilitation. However, conventional SHM cannot be used for structural continuous monitoring, real-time and online assessment to solve real-world problems. Therefore, integration of SHM with soft computing techniques has been successfully applied for optimized monitoring of bridges in recent years. This is due to the fact that soft computing is an umbrella of computational techniques that tolerates uncertainty imprecision, partial truth, and ambiguity. Hence, this chapter introduced the optimized SHM-based soft computing techniques of bridge structures through artificial intelligence and machine learning algorithms in order to illustrate the performance of advanced bridge monitoring approaches, which were required to maintain the health condition of infrastructures and for smooth functioning of cities.

Acknowledgments

The authors wish to acknowledge the University of Malaya and K.N.TOOSI University of Technology for providing the resources and supporting this research. The authors would also like to express their sincere thanks to the Structural Health Monitoring Research Group (StrucHMRSGroup), which was led by Professor Emeritus Hashim Abdul Razak. (Program Number: IIRG007A-2019).