基于认知能力开发施工安全隐患的动态预警机制研究

Although safety inspection is the key to promote construction safety performance, many auxiliary tools do not sufficiently support onsite inspection. The main reason of insufficient support rests in lacking consideration for cognitive ability to hazards and the dynamic coupling relationships among hazards. This study will develop a new hazard warning mechanism for construction hazards calibrated by human cognitive ability. This research will be conducted with the following steps. First, we will develop a new structural-learning algorithm to derive hazard-coupling relationships from real data based on theoretical principles and expert opinion will be adopted to screen out empirical hazard coupling relationships, forming a hazard-coupling model. Second, data will be collected from an eye-tracking experiment. Statistical approaches will be employed to develop cognitive ability metrics based on these data. Lastly, we will develop a dynamic hazard warning mechanism based on the hazard-coupling model and hazard coupling ability metrics. A second eye-tracking experiment will be developed to collect eye movements and inspection moving trajectories data with the new warning mechanism. The new dynamic warning mechanism for construction hazards will be validated with comparisons between the results of first and the second eye-tracking experiments. This study is featured with the following contributions to the body of knowledge: a new perspective that perceives the construction system as a whole for calculation for warning mechanics, a new developed structural learning approach for hazard coupling relationships, and a novel dynamic warning mechanism for construction hazards that promotes inspection performance of safety personnel with different hazard cognitive ability.

安全检查虽然是提升建设施工安全管理绩效的关键，但是多数研究所开发的安全检查辅助系统仍难以完整地支持现场检查工作，原因是此类系统在很大程度上未充分考虑人的认知差异和各类隐患的动态耦合关系。因此，本项目将基于人的认知能力差异开发新的隐患动态预警机制，以自动化平台实现此机制来辅助施工安全检查。首先，用基于理论规则新开发的算法萃取数据的规律，辅以经验筛选来构建工程施工隐患耦合模型；其次，设计眼动实验获取数据，综合考虑隐患信息量及眼动规律，以数理统计方法开发人的隐患认知能力指标；最后，基于隐患耦合模型与隐患认知能力指标开发隐患的动态预警机制，二次开发眼动实验来检验预警机制的效果。本项目具有三个特色：视角上，将所有施工隐患发生的不确定性视为一个整体，以不确定性的变化方向作为预警机制；方法上，合理安排定性与新开发的定量方法揭示隐患耦合机理；结果上，预警机制可让不同能力的安全检查人员发挥最佳安全检查效能。

本课题基于人的认知能力差异开发了新的隐患动态预警机制，以自动化平台实现了施工隐患的动态预警。首先，研究团队构建了隐患-事故耦合模型，基于事故案例、设计建议和理论推导提炼施工隐患的关联规则，以此为基础收集数据进行结构化学习，得到了建设工程的隐患-事故耦合机制。第二，研究团队量化了人类的隐患认知能力，设计了脑电与眼动结合的实验获取数据，以数理统计方法开发了人的施工安全隐患认知能力指标。第三，研究团队开发了隐患的动态预警机制，基于隐患-事故耦合模型推导得到了隐患的动态预警参数，将新开发的指标应用于工程实际，不断优化了动态预警机制。本课题从理论上完善了施工安全管理理论和隐患认知理论，并在实践上为改善施工安全绩效提供了重要指导。