Journal of Construction Engineering and Management

Abstract

Ethics in the construction industry has emerged as a critical topic of concern due to the industry’s complex and multifaceted nature and susceptibility to corruption and unethical practices. However, there is more to ethics than corruption, and the existing literature lacks a state-of-the-art review of ethical discourse in construction. Due to this limitation, it could be challenging for construction management scholars to understand the advances made in the realm of construction ethics, grasp a holistic understanding, and chart future research. Therefore, our study aims to examine ethical discourse in construction comprehensively. This study uses a mixed-method literature review (MMLR), combining bibliometric analysis and a systematic literature review. We analyzed 328 articles published between 1990 and 2023. First, through a bibliometric analysis conducted on the data sets obtained from Scopus and Web of Science, we identified publication trends, influential authors, prolific journals, impactful institutions, and countries in the field of construction ethics. Second, through a systematic literature review of 125 journal articles, we mapped the qualitative aspects of ethical discourse in construction. The meticulous review process elucidated the critical theoretical paradigms (economics, institutional, strategic stakeholder management, psychology, sociology, and ethics), emerging research themes (culture, corruption, professional development, welfare, leadership, technology, compliance, environment, social and governance (ESG), practices, project procurement, and stakeholder), and prominent analytical approaches used to study ethics-related issues in construction. The present study also mapped the ethical discourse on a multilevel (macro, meso, and micro) ethical governance framework and explored the geographical contexts and spread of ethics and ethical discourse. Finally, we presented potential future research trajectories by employing the theory-characteristics-context-methods (TCCM) framework that could encourage scholarly investigation in this domain.

Practical Applications

This review study provides several implications for research scholars and practitioners in the construction sector. This study provides fertile grounds for construction management scholars to begin by exploring the theoretical paradigms that are the foundation for research. Researchers should carefully select research methodologies to explore the ethical issues. This research highlights the most influential articles, authors, countries, institutions, and journals, helping the researchers explore the opportunities to collaborate with these entities. The paper has significant implications for various stakeholders, including construction companies, regulatory bodies, workers, clients, and broader society. It has highlighted the key thematic areas that help shed light on the multidimensional nature of ethics in the construction sector and encourage construction companies to adopt responsible practices. This can enhance the industry’s reputation and credibility, making it more attractive to investors and clients. We also discussed the ethical governance framework at the macro, meso, and micro levels that will help the stakeholders understand various aspects of ethical compliance across these levels within the construction industry. Further to chart the future research directions, we employed the TCCM (theory, context, characteristics, methods) framework with which the construction management practitioners and researchers may broaden their understanding of construction ethics.

Abstract

E-ticketing was an intriguing technology to many state transportation agencies (STAs) pre-COVID-19 pandemic but gained significant attention upon arrival of the pandemic due to the contactless nature of the technology. Research completed prior to the pandemic into E-ticketing for asphalt paving primarily identified qualitative benefits and concerns. However, minimal academic literature exists in the postpandemic era discussing the additional E-ticketing benefits not previously captured which resulted in increased implementation by STAs. This research seeks to address this gap by gathering information from state members of the American Association of State Highway and Transportation Officials Committee on Construction (AASHTO COC), employees of the Kentucky Transportation Cabinet (KYTC), and members of the Kentucky Association of Highway Contractors (KAHC) and the Plantmix Asphalt Industry of Kentucky (PAIKY) regarding E-ticketing practices for asphalt paving in the postpandemic era. The first portion of the study combines national and state-level survey responses to create a qualitative benefit-cost analysis for E-ticketing for asphalt paving operations based upon experience gained through emergency implementation during the pandemic. To reinforce the results of the qualitative analysis and increase the power of the study, the second portion of the research creates a quantitative benefit-cost analysis (BCA) from data collected from KYTC construction projects to compare the traditional weigh ticket collection process with E-ticketing processes for both project engineers and inspectors. The analysis shows a statistically significant time savings for field employees but not project management personnel. The monetary analysis for Kentucky indicates that whereas E-ticketing does not result in significant time savings for all employees, there is substantial monetary benefit worthy of STAs adopting E-ticketing as policy. The primary contributions to the body of knowledge include an improved BCA methodology applied to a new domain of electronic bulk material tickets and an informed quantitative BCA framework useful for STAs as a proof-of-concept to champion E-ticketing implementation and by transportation-focused researchers for evaluating emerging technological applications.

Practical Applications

Previous research has focused on improving safety and task efficiency for field employees associated with asphalt paving operations. E-ticketing (electronic ticketing) has been proposed as an innovative technology to help augment typical tasks associated with asphalt paving. Before the COVID-19 pandemic, several state transportation agencies (STAs) had introduced pilot projects to study E-ticketing, but these yielded minimal policy adoption results. During the COVID-19 pandemic, many STAs pivoted to emergency usage of E-ticketing as the primary means of collecting information about asphalt material used on construction projects. This research compares previous qualitative analyses regarding the benefits and concerns associated with E-ticketing prior to the COVID-19 pandemic with lessons learned by STAs across the country after utilizing E-ticketing during the pandemic. Additionally, a case study focused on the Kentucky Transportation Cabinet (KYTC) is used to create a quantitative benefit-cost analysis from project-level data collected from various asphalt paving projects during the 2022 construction season. The findings indicate a statistically significant time savings for field employees utilizing E-ticketing and a benefit-cost ratio that promotes STAs adopting E-ticketing as a permanent policy.

Although welding is one of the essential steel fabrication processes, the American Welding Society expects that the labor shortage in the United States will reach a deficit of 360,000 welders by 2027. Developing an automatic robotic welding system could potentially alleviate the labor shortage and provide better welding quality. As a first step, this paper designs a system pipeline that can automatically detect different welding joints and plan and track the joints’ trajectory with the initial point alignment in real time. There are rare studies that could achieve automatic initial point alignment in real time because the laser stripe’s deformation is not obvious at the narrow weld. In this study, the target joint’s endpoints were detected once the joint was detected on live video. Then, the joint trajectory was planned, and the robotic arm automatically aligned with the initial point and tracked the planned trajectory while scanning. The results demonstrate the accuracy and effectiveness of the proposed method.

Abstract

Determining whether to invest in a project is a complex and challenging decision. This project investment decision relies heavily on its promised strategic goals (i.e., target benefits) stated in its business case and which are expected to be realized following project completion. However, in many cases, the proposed project benefits are far greater than those that the project realistically can achieve, leading to the risk of projects being funded erroneously. The construction management literature has offered approaches such as reference class forecasting (RCF) to rectify unrealistic project estimates. This paper identifies weaknesses in the current approaches, exposes their inefficiency, and proposes a step-by-step stratified approach as an effective alternative to detect, measure, and rectify unrealistic estimates of project benefits, and utilizes the proposed approach on a light rail construction project. The contribution of this approach is that each project benefit target can be rectified differently, based on the measurable levels of deviation from more-realistic estimates. This paper advances the construction management literature by enabling managers to set realistic project goals in the project front end, increase the reliability of estimates in the project business case, and improve the quality of project investment decisions.

Practical Applications

The decision to fund a construction project is impacted by information included in business cases and proposals, such as the estimated project cost and expected goals. However, such information often is unrealistic or exaggerated (to make the project more attractive for funding), which negatively impacts investment decisions made by senior executives. As a result, projects can be funded erroneously, causing a waste of resources and potential failure to achieve its unrealistic expectations. Because the existing literature offers little guidance for how to ensure that project goals are realistic, this paper developed and illustrates a step-by-step process for measuring and rectifying the gap between proposed and realistic project goals to alert senior executives when a proposed project is unlikely to meet its proposed goals.

Though road authorities have well-defined inspection protocols for evaluating bridge serviceability, gaps remain in understanding optimal timing and the impact of maintenance strategies on an individual bridge and its subsystems and, consequently, on the holistic performance of a road network. This paper proposes an innovative model based on system reliability principles that develops optimized bridge maintenance strategies aimed at enhancing the overall reliability of a bridge-dominated road network by considering the quantified impact of bridge component deterioration. Using the first-order second-moment method, we initially determine the reliability index for each bridge based on condition ratings derived from routine inspections as outlined in the bridge inspection manual. Advanced optimization algorithms then formulate the optimal maintenance strategies for the bridge structures. Finally, we gauge the system reliability of the network using Ditlevsen bounds approximation, considering the impact of maintenance actions on varied bridge structural components. The developed model was implemented on a typical bridge-dominated road network in Chongqing, China. The results demonstrate the model has the capacity to quantify the impact of diverse bridge maintenance strategies on the total road network, offering valuable insights for decision makers to design effective, efficient maintenance strategies aimed at extending the lifespan of bridge-dominated road networks.

Highway work zones are essential for the preservation and improvement of the national road system. Nevertheless, these areas are reported to be among the most hazardous workplaces. Thus, it is crucial to develop appropriate measures to effectively mitigate the safety risks, which require a good understanding of the critical causes of accidents. While there are many previous studies on critical causes of construction accidents, none of them was specifically focused on highway work zones. This type of construction workplace has its own characteristics (e.g., near-passing traffic), which can lead to a unique set of critical causes of accidents. This study used text mining to extract root causes from a large narrative data set of construction accidents at work zones obtained from the Occupational Safety and Health Administration (OSHA). The study applied latent Dirichlet allocation (LDA) modeling on the text corpus to extract 12 root causes, which were subsequently classified into five groups: management, human, unsafe behavior, environmental, and material factors. In addition, social network analysis (SNA) was conducted to gain further insights into the interrelations between the root causes to determine their criticality degree. As a result, four highly ranked causes were identified: supervision dereliction of duty, weak safety awareness, poor construction environment, and risk-taking behavior. The findings of this study offer a new understanding of critical factors that highway agencies and contractors should focus on when developing construction accident prevention strategies at work zones.

Safety academics and practitioners in construction typically use safety prediction models that employ information associated with past incidents to predict the likelihood of future injury or fatality on site. However, most prevailing models utilize only information related to failure (i.e., incident), so they cannot distinguish effectively between success and failure without well-informed comparison. Furthermore, recordable incidents on construction sites are extremely rare, which results in data that are too sparse to make predictions with high statistical power. This paper empirically reviews different approaches to safety to increase the understanding of conditions associated with safety success and failure. Empirical data about business-, project-, and crew-related factors were collected to predict serious injury and fatality (SIF) exposure conditions. A variety of modeling techniques were tested in a machine learning pipeline to identify the most accurate and stable predictive models. Results showed that the multilayer perceptron (MLP) approach best distinguished SIF exposure conditions from safety success conditions using nonlinear decision boundaries. The most influential factors in the models included the crew experience working together, supervisor experience with the crew, total number of workers under the supervisor’s purview, and the maturity of leadership development programs for frontline supervisors. This study showed that data sets with both success and failure information yield more reliable and meaningful predictions than data sets with failure alone. Such an approach to safety data collection, analysis, and prediction could be used by future researchers to generate new insights into the causes of serious incidents and the relationships among causal factors.

Abstract

Photogrammetry has gained the interest of professionals and researchers for activities related to construction projects’ progress monitoring via attaining precise 3D point models. However, the precision of the generated models is directly linked with the precise scaling of the point cloud to ground truth dimensions (GTDs). Available scaling-up procedures for the close-range photogrammetry technique are complex, time consuming, and require human intervention, which adds the risk of error in the scaled-up model dimensions. Such a scenario creates hesitation among industry professionals toward implementing point cloud technologies. This paper devises an automated scaling-up methodology to overcome the said concerns by considering the construction progress monitoring theme. The intact process of automated scaling up of point cloud model to GTDs is controlled by two main parameters, that is, Python-based modules and designed ArUco-supported controlled markers. Remarkable outcomes are achieved with less than 1% scaled-up error compared with GTDs, which will improve the confidence of industry professionals toward point cloud technologies.

Practical Applications

Photogrammetry applications have been adopted in several domains and the optimum usage of attained models can be executed with 3D replicas having precise details of surface features and geometry. Therefore, to attain 3D point cloud models with ground truth dimensions (GTDs), or actual dimensions of the targeted object the practitioners mostly follow the markers/ground control points (GCPs) technique (minimum three GCPs/markers), manual scaling, or georeferencing data. However, the accuracy of traditional GCPs/markers’ technique and manual rescaling is dependent on the experience of the site staff/operator, and error chances may increase with the increasing number of GCPs/markers, whereas the georeferencing data-based technique is more technical and complex. Therefore, this paper developed an automated system for scaling up 3D point cloud models to GTDs with minimal human involvement. The system works with the help of specialized designed markers known as ArUco-supported controlled markers (ASCM). Only one ASCM marker is placed beside the targeted object for imaging; the devised system detects the marker in the images and rescales the developed point cloud model following the designed strategy. The system has high accuracy and can easily be implemented for scaling up close-range photogrammetry models in any domain.

Project requirements indicate specific works, events, or conditions that should be fulfilled to ensure the construction project success within planned budgets and times. To effectively manage project requirements, requirement prioritization allows for the proper allocation of limited project resources by determining the relative importance and urgency of different requirements. However, because project requirements are typically communicated through textual data in documents, the current approach to prioritizing requirements heavily relies on individuals’ expertise, practical knowledge, and experiences. This subjective judgment-based process poses a challenge in ensuring consistent and reliable prioritization, because there may be variations in practitioners’ prioritization results. Moreover, a large amount of text in documents can complicate capturing significant requirements within limited bidding times. To address these issues, this study proposes a novel method using historical data analysis and computational techniques. This study adopts historical change orders in order to evaluate impact levels of adjusted work items during construction and natural language processing (NLP) techniques, which enable the automated classification of requirements by the most-related work items. This study conducts a case study by examining documents from resurfacing projects and validating the feasibility and effectiveness of the proposed method. It will also provide a cornerstone for a smarter review and understanding of project documentation and improved decision-making for project planning.

To adequately facilitate project digital responsibility implementation (PDRI), a solid comprehension of the influencing factors and effective strategies that influence project digital responsibility implementation are necessary. Surprisingly, despite the increased interest in digital responsibility over the last few years, there still is a significant dearth of digital responsibility implementation. In practice, the factors that influence project digital responsibility implementation are complex and interconnected, not simple and isolated. Due to the omission of the causal relationship between these influencing factors, the project digital responsibility is not implemented effectively. This paper describes the process of identifying and classifying the factors influencing the project digital responsibility implementation, taking into account the network of interconnected factors. Four primary data sources and methodologies yielded new insights into the nature of these influential factors: a systematic literature review, semistructured interviews, fuzzy interpretative structural modeling, and cross-impact matrix multiplication applied to classification. Our analysis demonstrates that the 26 factors influencing project digital responsibility implementation interact and form a 9-level hierarchy. The 26 factors can be subdivided further into four dimensions: project dimension, stakeholder dimension, market environment dimension, and government dimension. Notably, the influencing factors in the government dimension have the most driving force, and adjusting the influencing factor in this dimension can enhance the influencing factors in the other dimensions. Finally, 11 policy proposals are proposed along 4 dimensions. This study bridges the knowledge gap of PDRI by providing new insights targeting the factors influencing PDRI. The findings provide valuable information for policy makers and practitioners to adopt effective measures to promote the implementation of PDRI.

To improve the current manual and iterative nature of clash resolution on construction projects, current research efforts continue to explore and test the utilization of machine-learning algorithms to automate the process. Though current research shows significant accuracy in automating clash resolution, many have failed to provide clear explanation and justification for the selection of their feature and label space. Since this is critical in developing an effective and explainable solution in machine learning, it is crucial to address this research gap. In this paper, the authors utilize an in-depth literature review and industry interviews to capture domain knowledge on how design clashes are resolved by industry experts. From analysis of the knowledge captured, we identified 23 factors considered by experts when resolving clashes and five alternative solutions/options to resolve a clash. Using a pool of industry experts, a modified Delphi approach was conducted to validate the factors and options and to determine a priority ranking. The authors identified 94 industry experts based on a predetermined qualification matrix to take part in the modified Delphi. Twelve participants responded and took part in the first round, and 11 completed the second round. A consensus was reached on all clash factors and resolution options. Factors including “clashing elements type,” “constrained slope,” “critical element in the clash,” “location of the clash,” “code compliance,” and “project stage clashing element is in” were ranked as the most important factors, while “clashing element material” and “insulation type” were considered the least important. Participants also showed more preference to the “moving the clashing element with low priority in/along $x\text{-}y\text{-}z$ directions” option to resolve clashes. These identified factors and options will be utilized to collect specific clash data to train and test effective and explainable machine-learning algorithms toward automating clash resolution.

Construction is a hazardous industry due to the dynamic nature of the work environment. However, limited studies have investigated how the introduction of additional hazards affects construction fieldworkers’ perception of safety risks. Therefore, this study aimed to investigate the influence of varying levels of hazard information on fieldworkers’ safety risk perception (SRP) for the OSHA “fatal four” hazards. A sample of 181 fieldworkers rated the severity and frequency of safety incidents for five construction work scenarios. The study found that as new hazards were introduced in the work scenarios, the fieldworkers had a significant change in their perception of risks for about 71% of the accident causes. They were able to assess safety risks related to slip, trip, fall; struck-by; caught-in or -between; electrocution; stepping on an object; fall to a lower level; and trapped-in by collapsing. However, they failed to perceive the risk of increasing the number of workers in a hazardous work environment, indicating that workers focus on the severity of the occurrence of accidents and do not pay attention to the workers present on the scene. This study provides new insight into the role of varying degrees of hazard information in situational awareness and highlights the need for additional research at the intersection of worker exposure and situational awareness. Additionally, results from the study could inform the development of safety training to enhance workplace safety.

Although prior research has examined the influence of site supervision on construction safety, a comprehensive systems-level perspective incorporating interdependent relationships among key variables has yet to be fully investigated. In particular, the intervariable mechanisms by which owners impact contractor safety behaviors require further elucidation through a system dynamics (SD) approach. A primary objective of this study is to identify the feedback mechanisms that affect the safety performance of a project as a result of the site supervision systems using SD. The identification of relevant safety variables and the creation of a conceptual causal loop diagram incorporating key variables facilitate the development of corrective actions. After creating a qualitative SD model based on the identified safety variables, the Delphi technique is employed for validation. Lastly, a portion of the validated conceptual model is selected for quantitative analysis. The simulation model is validated through behavior reproduction and sensitivity analysis tests. The results indicate the accurate performance of the quantitative model in capturing the system’s behavior. For instance, it suggests that the owner’s safety role is a root cause of accidents in construction projects, and owners can influence the safety performance of projects through contractors. The findings revealed an insignificant relationship between changes in the number of accidents and the frequency of owner’s representative site visits. Moreover, the quantitative SD model shows that schedule delays should be considered in planning for safety performance improvement in addition to effective safety actions. Practitioners may use the SD model in three ways: detecting the safety outcome during the project, assessing impacts of various model variables on safety outcome, and creating new policies or corrective actions in response to project changes. This paper contributes to the understanding of how site supervision systems impact safety performance in construction projects, providing practitioners with a valuable tool to enhance safety outcomes.

Accurate estimation of activity durations is a critical project scheduling step. If abundant data are available for activity durations, then distribution fitting is straightforward. However, if sample sizes are small, then choosing an appropriate theoretical distribution for activity duration is challenging. In this study, we propose a methodology for experimenting with the impact of sample size on the accuracy of the estimated duration of a stochastic project network. We synthesized a total of 3,264 stochastic project networks and estimated the true project duration distribution for each network using large-sample Monte Carlo simulation. We varied the sample size per activity and investigated how the choice of theoretical distribution (normal, beta, triangular, and lognormal) for estimated activity durations impacted the accuracy of the estimated project duration distribution. We find that the best choice for a theoretical activity duration distribution highly depends on the sample size. For sample sizes below 50, we find that not much is gained from the common practice of putting extensive effort into selecting best-fit theoretical distributions compared to simply using normal distributions.

Various types of government financial support are used to facilitate the financing of public–private partnership (PPP) projects in developing countries, wherein direct equity investment can have a unique impact on capital structure in project finance because of the changing role of the government as a shareholder. However, the distinct impacts of government equity investments have not yet been clearly revealed. This study clarifies the effect of government equity investment on leverage in project finance based on a two-step treatment effect model using a large data set containing 1,326 PPP projects in developing countries. Six hypotheses are proposed to test the direct impacts of government equity investment, participation of development-financing institutions, project characteristics, and moderating impacts from the governance environment on leverage. Empirical evidence reveals that government equity investment can significantly increase leverage in project finance when the ultimate ownership of the project will be transferred back to the government, and that the governance environment can affect the government’s decision to invest. The findings suggest that host governments in developing countries can boost lenders’ confidence in PPP projects through direct equity investment, but the investment decision should be prudent after considering the domestic governance environment and specific project conditions.

The occurrence of defects in recently completed construction projects in many countries has aroused public attention worldwide. Defects in construction projects are caused by a multitude of factors. Numerous studies have identified and evaluated the critical factors affecting quality in construction projects; nonetheless, no consensus has been reached regarding the factors and their level of influence. Thus, to fill this gap, this research used Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines and prioritized the factors quantitatively. Following the systematic review and meta-analysis, 45 critical factors were identified from 22 relevant studies. Further, heterogeneity, subgroup, and sensitivity analyses were conducted. The findings show that the top seven factors leading to poor quality in construction projects include lack of communication, teamwork, and skilled workers; poor supervision; design errors; design complexity and low bid contracting. The study found very low heterogeneity confirming that ranking using the meta-analysis is reliable. This analysis revealed that project location and type, and publication types, contribute to the heterogeneity of the results. This research work contributes to the existing body of knowledge by identifying and ranking the most critical factors influencing the quality of construction projects across the globe. The stakeholders involved in the delivery of construction projects can prepare strategies for enhancing the quality of their projects based on the findings of this research.

The performance of expatriate construction professionals (EXCPs) has become increasingly critical for the success of international construction projects. This study investigated the role of job burnout and cross-cultural adjustment in predicting EXCPs’ performance. Drawing on the job demands-resources model, a comprehensive framework was developed to illustrate the relationships among cultural intelligence, stressors, expatriate management practices, cross-cultural adjustment, job burnout, and performance. Data were collected through a survey of Chinese EXCPs working in various host countries, and the proposed model was tested using structural equation modeling (SEM). The findings reveal that expatriate management practices and cross-cultural adjustment have a more substantial motivational impact on performance than do the negative impacts of stressors and job burnout. Additionally, cultural intelligence is significantly associated with stressors, expatriate management practices, cross-cultural adjustment, and job burnout. This study makes a valuable contribution to the expatriate literature and offers strategies for dual-process management to enhance EXCPs’ performance in international construction projects.

Detecting deformation and segment assembly quality in the construction or as-built phase of the shield tunnel is crucial and significant to ensure structural safety. The traditional detection methods consume much cost and are prone to errors. This study applies point cloud to develop robust algorithms for the deformation detection and reconstruction of shield tunnels. The methodology initially extracts the tunnel axis, serving as the base for deformation detection and reconstruction. A segmentation algorithm for continuous slice point clouds along the tunnel axis is proposed, and the deformation of the section is evaluated by ellipse fitting. In addition, a novel method of creating a binary image using the unrolled point cloud is adopted based on the extracted tunnel axis, and the segmentation of the segment point cloud is realized via image processing. This process is based on the geometric features of the unrolled point cloud, avoiding tedious parameter adjustment. Finally, a novel segment point cloud fitting method is used to create the as-built model of the tunnel in the BIM platform. To evaluate the performance of the proposed method, we select the shield tunnel case for experimental verification. The results show that (1) using point cloud information can realize an automated solution to complete the tunnel deformation detection task and meet accuracy requirements; and (2) the reconstruction method adopted in this study can realize the visualization of segment dislocation and has better efficiency and accuracy than previous algorithms. The work of this study has a certain guiding significance for the automated detection of the shield tunnel.

The purpose of this study is to develop a model for predicting the financial performance of construction companies based on their financial statement data. Several models for predicting financial performance have been developed in the general finance field over the past few decades. However, these conventional models are not always suitable for the construction industry, which operates on a project-based system. While there have been attempts to develop financial models specific to the construction industry, the proposed model in this study stands apart, as it is designed based on the differences between the construction and manufacturing industries. For this research objective, financial variables presumably affecting a construction company’s financial performance are identified through literature review, industry expert interviews, and statistical tests, which explore differences between construction and manufacturing companies’ financial characteristics. Taking the identified variables from these approaches, this study proposed a prediction model for the return on asset and enterprise value per share of construction companies. The prediction model was applied to construction and manufacturing companies’ financial data, and it was verified that it showed significantly higher explanatory power in the construction data. In addition, a panel regression analysis was applied to examine how each variable is correlated with the financial performance of construction companies. Based on the identification of difference between the construction and manufacturing sectors, this study developed a more appropriate explanation model for the financial performance of construction companies. In this regard, this study adds empirical evidence that the factors influencing financial performance vary by industry. Further, the identification of financial variables that affect the performance of construction companies can assist practitioners in establishing investment and financial strategies.

Abstract

The production quality assessment of prefabricated laminated slabs (PLS) is required to ensure the installation of components on the construction site. However, existing inspection methods are difficult to realize PLS production quality inspection during the production process, and the inspection speed cannot match the PLS production speed. To address these issues, a deep learning (DL) method based on You Only Look Once version 4 (YOLOv4) is proposed for the quality detection of PLS in production in this research. Five new measures are introduced to improve the performance of YOLOv4. The ResNet50vd backbone and deformable convolution (DCN) are used to enhance the ability of the model to capture the feature information of small targets with drastically changing morphologies in complex contexts. The $k\text{-}\mathrm{means}++$ clustering algorithm is used to recalculate anchor box specifications to overcome the challenge of large differences in scales across targets as well as within the same target. Intersection over Union (IoU)-aware loss is used to optimize the loss function to improve the localization accuracy of small targets. Geometric median filtering pruning (FPGM) is used to simplify the network, and the model parameters are reduced to 44.1%, which significantly improves the detection speed of the components. The results show that the improved YOLOv4 can quickly and accurately detect laminated panels, PVC boxes, and steel bars in PLS when $\mathrm{IoU}=0.75$, which is more suitable for industrial applications, and the model achieves a mean average precision (mAP), $F1$ score, and FPS of 91.38%, 94.54%, and $22.91\mathrm{f}\mathrm{r}\mathrm{a}\mathrm{m}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}(\mathrm{f}/\mathrm{s})$, respectively, which is superior to the comparison method. The proposed method can be directly applied to PLS in production to preemptively detect errors and discrepancies in the components and improve the qualification rate and quality management efficiency of the components, which has a positive effect on the rapid development of industrialized construction.

Practical Applications

Combined with the actual situation of prefabricated components production, this paper proposes a PLS production quality inspection method that is more in line with the factory production process. The new method was obtained by introducing five improvements to YOLOv4. The improved YOLOv4 can quickly and accurately detect laminated panels, PVC boxes, and steel bars in PLS when $\mathrm{IoU}=0.75$ and is more suitable for industrial applications, and the model achieves a mAP, $F1$ score, and FPS of 91.38%, 94.54%, and $22.91\mathrm{f}/\mathrm{s}$, respectively. The proposed method does not require an additional inspection site and can be directly applied to the PLS production line for quality inspection of the PLS in production without interfering with the normal production process of the components and the outgoing process. By changing the posttest inspection to in-test inspection, the method can preemptively detect errors and discrepancies in PLS production and implement remedial measures in a timely manner, which effectively improves the qualification rate of components and the level of production quality management, and reduces the production cost of components.