人工神经网络在林火发生预报中的应用

The Application of Artificial Nervous Net to Forecasting Forest Fire

对于林火发生预报，就是要将着火样本和不着火样本进地划分，得到一个判断是否发生林火的阈值．以往我们采用过数量化理论和回归分析判断，但那些都是线性划分，而对于不能线性划分的样本集，划分的效果就不是很好．本文利用人工神经网络的二层感知器模型对着火样本和不着火样本进行了重新划分，二层感知器是在二维输入空间上用三条相交的直线，得到一个封闭区域以实现样本集的正确划分．

To forcast forest fire, it is to divide specimen into fire and firelessness and to get a dgment for firing and unfiring value, in general. We adopt the numerical theory of regression analysis, but these are linear. For those of non linear, the dividing method is Perfect. This paper gives the way to use two Iaper feeling eqiupment of artificial nervous net to divide speciment of fire and firelessness. Two layer feeling eqiupment is three line in two dimension space and get a closed region to divide for specimen.