ABSTRACT
Multi-agent is an important branch of distributed artificial intelligence. It is developed to solve the large-scale complex problems intelligently. Its basic idea is to separate large complex systems into many small autonomy systems (agents), which can communicate with each other and operate coordinately. Through cooperation with the agent’s intelligent behaviors, such as interaction and collaboration, complex task can be solved [2-3]. the application of multi-agent technology in forest firing is still in the exploratory stage. Most researches are still in the concept proposed and model designed stages. Few of the them can be applied to the actual structure. scholars have studied and developed a series of fire spreading simulation models, which can be divided into several groups, including the Equation-Based Models, System Models, Statistical Models, Evolution Models, Cellular Models, Agent-Based Models and so on (Liu, et al., 2008; Chen, 2010). These simulation methods have their own advantages and disadvantages. The equation-based model regarded as a kind of static model is easy for quantitative analysis, but does not consider spatial complexity simulation of fire spreading (Parker, et al., 2003). The model has strong ability of systematic analysis, but is difficult to achieve spatial analysis. The statistical analysis method is widely used, but in the process of fire spatial decision, it is difficult to be adopted due to the difficulty of data acquisition. The evolutionary
model is a kind of effective method to solve quantitatively decision problems because of its simplicity, common use, strong robustness and the ability of parallel computing. However, it is inclined to fall into locally optimal solution when the problem is more complex. Due to the agent itself has characteristics of initiative, interactive, collaborative, reactivity, autonomy, mobility, it is more suitable for expressing complicated group behavior, in particular, have great advantages over geographical spatio-temporal related events.
