学习事故是避免安全事故重复发生的重要手段。然而，目前鲜有研究深入探讨了组织学习事故成效与组织安全事故重复发生的关系。研究表明学习事故并非仅限于对某一事故的信息分享，更应着重提升组织安全文化。因此本项目基于组织学习、韧性安全文化与安全信息认知理论，采用定量和定性的研究方法，目的是揭示工程项目安全事故重复发生的根本原因，并制定提升组织学习成效的关键策略，从而避免同类事故再次发生。本项目首先对已初步建立了的建筑业安全事故数据库（1999-2019年间2894次事故）进行数据挖掘，查找出同类安全事故重复发生的规律特征。其次本项目采用访谈与问卷调查等方式，运用主成分分析法、结构方程模型、人工神经网络和贝叶斯网络等分析方法，以确定组织学习事故成效、韧性安全文化与安全信息认知之间的关系。本项目不仅能够填补目前工程项目安全研究领域的空白，推动理论的发展，更有助于增强项目组织与个人学习事故的能力，从而有效解决“同类安全事故重复发生”的现实问题。

Learning from accidents has long been an essential means to prevent the recurrence of similar events. However, learning is far from becoming a successful endeavor given that similar accidents continue to plague the construction industry. Research has shown that the approaches of learning from accidents are not only by sharing safety information about a specific accident but also by fostering a safety culture where learning is a process of continuous information flow. So far, research efforts have been limited to determine the relationship between the quality of organizational learning and the recurrence of similar construction accidents. Through adapting from the theories of organizational learning, resilient safety culture and safety information cognition, the current project aims to determine the root causes of recurrence of similar accidents and to formulate critical re-action plans, in order to enhance the ability of organizations and individuals to learn from accidents and thereby achieve “never again”. Construction accidents database (2894 accidents recorded between 1999-2019) and data mining technique will be used to identify the key features of similar accidents. By a comprehensive literature review, questionnaire survey and semi-structured interview, the study will develop the instruments to measure the quality of organizational learning, resilient safety culture, and safety information cognition. Built on this, interviews and questionnaire surveys will be administered among management staff and frontline workers to examine the significant relationship among the quality of organizational learning, resilient safety culture, and safety information cognition. Principal component analysis (PCA), structural equation modeling (SEM), artificial neural network (ANN) and Bayesian network (BN) modeling will be adopted to analyze the aforementioned relationships. The research findings will not only enrich and improve the current body of knowledge in safety science but also improve the learning ability of organizations and individuals to avoid the recurrence of similar accidents in the future.