Unsafe behavior is a critical factor leading to construction accidents. Despite numerous studies supporting this viewpoint, the process by which accidents are influenced by construction workers’ unsafe behaviors and the extent to which unsafe behaviors are involved in this process remain poorly discussed. Therefore, this paper selects cases from Chinese building construction accidents to explore the probabilistic transmission paths from unsafe behaviors to accidents using a Bayesian network. First, a list of unsafe behaviors is constructed based on safety standards and operating procedures. Second, several chains of unsafe behaviors are extracted from 287 accident cases within four types (fall, collapse, struck-by and lifting) to form a Bayesian network model. Finally, two accidents are specifically analyzed to verify the rationality of the proposed model through forward reasoning. Additionally, critical groups of unsafe behaviors leading to the four types of accidents are identified through backward reasoning. The results show the following: (i) The time sequence of unsafe behaviors in a chain does not affect the final posterior probability of an accident, but the accident attribute strength of an unsafe behavior, affects the growth rate of the posterior probability of an accident. (ii) The four critical groups of unsafe behaviors leading to fall, collapse, struck-by, and lifting are identified. This study is of theoretical and practical significance for on-site behavioral management and accident prevention.

Various construction accidents are proven to be caused by multiple unsafe behaviors (e.g., wrong use of PPE), but the risk transmission among different behaviors remains unclear. This paper provides insight into risk transmission through behavioral risk chain that leads to accidents from a system safety perspective. To better understand the coupling mechanism of various unsafe behaviors, integrate different behavioral risk chains and present the risk transmission process, a directed-weighted complex network (DWCN) method was adopted. Historical urban railway construction accidents in China are investigated to extract behavioral risk chain. A DW-BRCNA is applied to integrated behavioral risk chain and the behavioral risk transmission characteristics are explored and clarified by the five network properties, including degree and degree distribution, node strength and node strength distribution, average path length and diameter, weighted clustering coefficient and betweenness centrality. The results show that DW-BRCNA has the characteristics of a small-world, scale-free and hierarchical network, indicating that some unsafe behaviors are of greater importance in the process of risk transmission through behavioral risk chains. In addition, risk transmission in critical behavioral risk chains is more potentially to lead to accidents. This study proposed a new perspective of accident causation analysis from risk transmission among unsafe behaviors. It explains the risk transmission characteristics by a DWCN method based on behavioral risk chains. The findings also provide a practical guidance for developing control strategies on sites to prevent risk transmission and reduce accidents.

Unsafe behavior is a critical risk factor in the construction industry, but it cannot be quantitatively evaluated by the traditional risk matrix (RM) method. Based on accident cases in the building construction industry in China, probabilistic reasoning from unsafe behaviors to accidents is explored using a Bayesian network (BN) to modify the RM method. The unsafe behaviors involved in 287 accident cases of four types (fall, collapse, struck-by and lifting) are used to construct a BN model. The parameters of the model are revised via Beta and Dirichlet distributions, and the conditional probabilities of the occurrence of accidents caused by unsafe behaviors are then obtained to determine the impact of unsafe behaviors on different types of accidents. A list of behavioral risk levels for the four accident types is generated by considering the initial probabilities and the conditional probabilities. The results show that the MRM method is feasible and effective, and unsafe behaviors that have a great impact on the four accident types are identified. The MRM method provides a method for quantitatively evaluating behavioral risk and enhances behavioral risk management on construction sites; thus, it is of theoretical and practical significance.

The construction industry is extremely high-risk, and safety checklist analysis (SCA) is a widely used approach for safety assessment. To overcome its limitations, this paper proposes an improved safety checklist analysis (ISCA) approach using intelligent video surveillance to replace on-site inspection. Then, a case study on metro tunnel construction is adopted to illustrate the process. First, the checklist is prepared. Second, the inspection items are correlated with construction areas, and intelligent cameras are positioned to cover the major areas of the construction site to guarantee that all the items can be checked. Thus, problems with inspection items are automatically identified and recorded. Third, the inspection items are marked by a remote scoring mechanism for safety assessment. Finally, the efficiency of the improved approach is tested by a comparative analysis among three groups. The application results indicate the feasibility of the improved approach for evaluating the safety management performance at construction sites.

To prevent fire accidents on construction site and improve the accuracy of fire detection, an improved YOLOv3-tiny method (I-YOLOv3-tiny) is proposed in this paper. Although the YOLOv3-tiny has a fast detection speed and low equipment requirement, the accuracy is relatively low on fire detection. The improvement of the I-YOLOv3-tiny method is followed by three steps. Firstly, the feature extraction of fire images is enhanced by optimizing the network structure. Secondly, a multi-scale fusion is used to improve the detection effect of fire targets. Finally, the anchor boxes that are suitable for fire data sets are selected by k-means clustering. The results show that I-YOLOv3-tiny has an increased percentage of 4 on the mAP, the Recall rate has an increased percentage of 4, and AVG IOU has an increased percentage of 6. The proposed model meets the real-time performance of fire detection. This study is of theoretical and practical significance on fire safety management and accident prevention in the construction industry.