Abstract: The occupational safety situation in China is complex and alarming, with over 38,000 safety accidents occurring nationwide in 2020, resulting in over 27,400 deaths. As one of the most dangerous industries, the construction industry in China sees almost 4,000 deaths at construction sites every year. Therefore, researching and resolving safety issues at construction sites is paramount. Workers' unsafe behavior at the workplace, exhibiting diverse, dynamic, and complex characteristics, is a long-standing weak link in accident prevention and control systems at construction sites and often leads to safety accidents. Traditionally, safety management research and practice focus mainly on the impact of organizational and policy factors on worker behavior. The most direct approach to improving the safety conditions of workers at construction sites typically involves strengthening management and establishing strict rules and regulations to prohibit workers' unsafe behavior. However, research has shown that workers have a higher acceptance of safety reminders from team leaders and colleagues within the same operational group compared to rigid safety regulations. Workers also make autonomous decisions on whether or not to comply with regulatory constraints and whether to adhere fully or partially to the system, which significantly impacts their safety. Ignoring workers' subjective intentions and the complex, dynamic interactions between them can lead to less effective research results.
Taking into account subjective factors such as workers' characteristics and their differing sense of belonging to the organization, this paper introduces the concept of a “safety bond” and constructs an evolutionary game dynamic model for the formation of safety bonds among workers at the construction site. A payoff matrix for safety bond formation among workers is established, depicting the process by which safety interactions among workers influence their decision-making probability to initiate “safety interactions”. By solving the replicator dynamic equations of the evolutionary game dynamic model, we obtained the evolutionary equilibrium points of the game system and analyzed them to get a stable equilibrium solution. We simulated the dynamic game process, described by a system of differential equations, to analyze the stability of the evolutionary game equilibrium points and evolutionary stable points. The Jacobian matrix of the dynamical system is derived from the local stability analysis, which then verified whether the combination of the game equilibrium points and evolutionary stable points is the final evolutionary stable strategy achieved by both parties.
The results elucidate the evolutionary process of safety bond formation among onsite workers and the mechanisms driving rapid evolution. We found that when the organizational reward for a worker initiating a safety interaction is less than the cost of initiation, or when the organizational reward for both parties initiating a safety interaction is less than the benefits of forming a friendship-based safety bond, both parties will eventually evolve towards a stable strategy of either both initiating a safety interaction or not initiating at all. The evolutionary path of workers' strategy choices and the probability of initiating a safety interaction are closely related to the initial state of the game and the parameters of the payoff matrix. The speed of system evolution is positively correlated with the number of safety interactions initiated by workers and is influenced by the initial values of the system environment parameters. Factors such as workers' safety awareness, organizational belonging, job satisfaction, and the cost of safety interactions are the main influencers of workers' safety behavior decisions and the rate of system evolution.Finally, considering the reality of onsite workers, based on the analysis and conclusions, and the characteristics of workers in high-accident industries, we propose four management recommendations to reduce unsafe behavior, promote the transition from “required safety” to “desired safety”, and prevent occupational accidents: 1)Form a core operation team, establish a safety interaction reward mechanism, and provide special rewards for workers who are willing to interact altruistically. 2)Improve the vocational skills training system to encourage workers to initiate safety interactions from an emotional standpoint. 3)Establish a long-term wage payment guarantee mechanism, ensure all workers on the job site are covered by workers' compensation insurance, and make workers feel cared for by the government or company. 4)Pay attention to and mobilize the subjective initiative of workers to initiate safety interactions; During recruitment, select workers who are more proactive in initiating safety interactions to reduce unsafe behavior and accidents; Regularly conduct tests on workers' subjective initiative in initiating safety interactions during operations and reward those with excellent test results to motivate workers to initiate safety interactions.

Key words: safety ties, evolutionary game, driving mechanism, evolutionary speed, field workers

摘要： 我国作业现场工人不安全行为导致安全事故频发,以往安全管理研究与实践多仅关注组织与政策因素对工人行为的影响,忽视工人主观意愿,工人之间的复杂、动态交互过程及其安全行为决策的内在形成机制,导致效果不佳。本文通过引入安全纽带,构建作业现场工人安全纽带缔结的演化博弈动态模型。通过求解复制动态方程得到演化博弈的稳定均衡解,并进行仿真分析,阐释了现场工人间安全纽带缔结的演化以及如何驱动演化快速进行的机制。结果表明,工人安全意识、组织归属感、工作满意度、安全交互成本等是工人安全行为决策以及系统演化速率的主要影响因素。最后,针对现场工人实际,提出管理建议:改进作业环境,关注并调动工人发起安全交互的主观能动性,可有效减少工人不安全行为,防范生产作业事故的发生。

关键词: 安全纽带, 演化博弈, 驱动机制, 演化速率, 作业现场工人