基于VLBP神经网络的林火风险预测模型

以北京房山区林火为研究对象,用神经网络方法研究了林火发生及蔓延与环境因子间的关系。研究流程为用GPS采集了林火火场中心数据,测量了火场面积。用GIS做空间分析,提取了火场空间环境因子。将数据在Matlab中用VLBP神经网络进行训练,建立了林火风险的预测方程。该风险模型由林火发生的风险模型和林火的蔓延风险模型组成。由于该模型采用的方法不依赖于先验统计模型,所采用的神经网络方法对非线性的数学关系有很强的预测能力,因而该模型具有很强的通用性。

Forest Fire Prediction Research Based on VLBP Neural Network

In this research,the Fangshan district was used as the research district and the relation between the fire risk and its environment factors were studied using the the artificial neural network methods.The procedure framework is as follows:the fire centeral spot and its area was 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the data was put into the GIS for spatial analysis.The data of the environment factors affecting the forest fire were withdrawn using the GIS,training of the data was done in Matlab with the VLBP Artificial Neural Network,and the forest fire risk forecast equation was established.This risk model is composed of 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 very strong forecast ability to the non-linear mathematical relation,this model has very strong versatility.