Abstract: In this research,the Fangshan district as the research district,the relation between the fire risk and its environment factors were studied using the artificial neural network methods. The procedure framework is as the fellow: the fire regimes' center and its' area is surveyed with GPS,and the data was put into the GIS for spatial analysis. The data of the environment factors affecting the forest fire were withdrawn using the GIS,training the data in Matlab with the VLBP artificial neural network,and establishing the forest fire risk forecast equation,this risk model is composed with the fire ignition model and the forest fire spread model. Because this model uses the artificial neural network which does not rely on fire data's probability distribution,and the nerve network method has the very strong forecast ability to the non-linear mathematical relation,thus this model has the very strong versatility.

Key words: forest fire, artificial neural network, risk model, ArcView GIS