Abstract: In view that forestry fire has the nature of multi-phase and multi-component, a forestry fire spatial diffusion model based on multi-agent algorithm with cellular automata is proposed. A method based on Wang Zhengfei-CA model is employed to approximately simulate the spread of forestry fire. On the basis of the status transition rule determined by CA and using the self-learning and variation methods of multi-agent algorithm, the measured forestry fire pattern is most similar to CA rule-driven evolution model and the CA rule can be automatically optimized. By comparatively analyzing the simulation results of the traditional Rothermel model and the method proposed in this paper, it is found that in the respects of fire area and fire perimeter, the actual forestry fire spread process has higher similarity to the simulation results of the proposed forestry fire spatial diffusion model.

Key words: evolutionary agent, multi-agent algorithm cellular, automata, rules function, forestry fire