Adaptive dynamic fire danger evaluation of logistics warehouses with fusion of evidential reasoning and smart optimization

Abstract

Considering the background that the current fire risk assessment methods often rely on subjective assessments based on professional knowledge and experience, they are usually limited to the special fire scene. Given the lack of effective tools for assessing fire risks in logistics warehouses across varying scenes, this paper aims to propose a multi-source data-driven, adaptive dynamic fire danger evaluation method. The method is established based on the fusion strategy of evidential reasoning and artificial fish swarm algorithm. According to the method, rules of danger evaluation levels related to each attribute of fire factors can be generated automatically without necessary of professional knowledges and experiences. The lightweight multi-source data at this moment can be fused to evaluate current fire danger level directly, which does not require training on extensive datasets in advance. Finally, multi-source data including temperature and CO concentration in two fire scenes of one logistics warehouse at different times are utilized to support the ability of the method. The results show that the estimated fire source location according to the center of domain with the evaluated highest danger level matches well with the real fire source location. In conclusion, the main contribution of the research is that the reasonable dynamic danger level of the logistics warehouse fire with time variation can be evaluated automatically. The method has an excellent generality, which is not limited to the specific fire scene. The danger zone of the logistics warehouse under sudden fires can be determined adaptively, which can support the fire prevention and fire safety countermeasures of logistics warehouses.