摘要
针对地震火灾的复杂性和多变性的特点,在综合分析了各种震后火灾损失预测方法的基础上,提出了基于支持向量机的城市震后火灾损失预测模型。以地震火灾的历史资料为依据,建立学习样本和测试样本,并与BP神经网络预测方法相比较,验证了其可行性,也为其它自然灾害的损失预测提供了简单、有效的方法。
According to complexity and multiplicity of the post-earthquake fire, the method of support vector machines is proposed to predict the loss of earthquake fire based on the comprehensive analysis of other different methods. On the basis of historical data and documents of earthquake fire, the applicability and validity of the model is manifested through testing and discussion. A simple and available method is provided for the prediction of losses of other natural disaster.
出处
《灾害学》
CSCD
2008年第1期6-9,共4页
Journal of Catastrophology
基金
国家十一五科技支撑计划重点项目(2006BAJ06B01,2006BAJ04A03,2006BAJ13B04)
国家自然科学基金资助项目(50678060)
北京工业大学研究生科技基金资助课题(ykj-2007-1071)
