To investigate the evolutionary mechanisms of coal mine safety systems, and to improve the level of system safety, this paper introduces a new method of assessing and managing the system safety of coal mines. A multi-agent modeling and simulation method is used to construct a dynamic evolutionary model of a coal mine safety system based on complex adaptive system theory. The NetLogo platformis used to design experiments that dynamically simulate the interactions between leaders, managers, and operators under different conditions, and to assess their impact on system safety. The results reveal that the influence of each agent on system safety is different, and safety leadership, safety management, behavioral safety, and environmental suitability all directly or indirectly affect system safety. Among these factors, the safety management level and operators' behavioral safety level exert the strongest influence on system safety. Therefore, in the specific implementation of safety management, coal mine enterprises should adjust and control these attributes to reduce accidents and improve system safety. (C) 2021 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.