Developing safety checklists for predicting accidents

Industries realize seriousness of industrial accidents and make efforts to minimize industrial accidents. Nonetheless, rates of industrial accidents are not decreasing. Because current safety management system depending on safety supervisors has limitations to manage and monitor every workplaces during working time. Therefore, safety management system that allows to manage safety by considering continuously changed workplace circumstances without depending on safety supervisors is needed. To establish the system, we should develop a way to reflect real-time status of workplaces and know what specific actions that are needed to prevent accident. Generally, these information can be indirectly applied to safety checklists. In this respect, we identified major risk features and accident types from the past shipbuilding fatal accident cases and then derived safety checklist by analyzing characteristics of accidents. Finally, we evaluated results of accident prediction using learning model based on the derived safety checklist and the prediction accuracy was 93.75 percent. It was found from the result that these safety checklist can be primary information that are used by safety management system.