欢迎访问林业科学,今天是

林业科学 ›› 2005, Vol. 41 ›› Issue (4): 127-132.doi: 10.11707/j.1001-7488.20050422

• 论文及研究报告 • 上一篇    下一篇

基于人工神经网络预测广东省森林火灾的发生

杨景标 马晓茜   

  1. 华南理工大学电力学院,广州510640
  • 收稿日期:2003-06-02 修回日期:1900-01-01 出版日期:2005-07-25 发布日期:2005-07-25

On the Basis of Artificial Neural Network to Forecast the Forest Fire in Guangdong Province

Yang Jingbiao,Ma Xiaoqian   

  1. College of Electrical Engineering, South China University of Technology Guangzhou510640
  • Received:2003-06-02 Revised:1900-01-01 Online:2005-07-25 Published:2005-07-25

摘要:

应用人工神经网络建立热带森林火灾发生情况预测的多层神经网络模型,并将林火发生影响因子的历史数据作为样本值,输入模型进行训练。结果表明:利用所选取的输入因子作为样本的人工神经网络,可以对林火的发生发展作出准确有效的预测。文中还对模型的准确性和训练精度进行讨论,进而分析人工神经网络在林火预测中的可行性,证明人工神经网络在林火预测中的应用价值。

关键词: 森林火灾, 人工神经网络, 林火预测

Abstract:

The multi_layer neural network model of the forest fire forecast was built based on the artificial neural network, and the historical data of the factors that were related to the forest fire closely were treated as samples to be trained for the network as well. The results indicated that the forest fire could be forecasted exactly and effectively by the network with the samples consisted of the factors chosen. In_depth discussions on the performance and the training precision of the model were presented, and the feasibility of forest fire forecast was analyzed. The artificial neural network had comprehensive application value in the forecast of the forest fire.

Key words: forest fire, artificial neural network, forest fire forecast