摘要
通过对地震伤亡资料的综合考察研究,本文构建了影响地震伤亡人数的指标集,利用粗糙集理论约简此指标结构,建立最小二乘支持向量回归机预测模型预测伤亡人数,并用粒子群优化算法对模型参数进行优化。最后,将模型运用于云南地震伤亡人数预测,结果和RS-BP神经网络预测模型对比分析,验证了该模型预测的有效性。又将模型应用于芦山和玉树地震死亡人数的预测,验证了模型的适用性。实验结果表明该模型能在地震发生后,给决策者提供人员救护、安置以及应急物资供应、统筹调度的有效依据。
Through a comprehensive study of datas on casualties in the earthquake,in this paper,indicator set that affect earthquake casualties is built up and rough set theory is used to simplify the indicators. Then the least squares support vector regression( LSSVR) model is established to predict the casulties of the earthquake,before which particle swarm optimization( PSO) theory is advanced to optimize the parameters of the model. Finally the model is adopted to predict past earthquake casualties of Yunnan province. Besides,the results is compared with rough set and BP network theory,which can verify the validity of the model. And the model is used in the death toll prediction of Lushan and Yushu County earthquake,which verified the applicability of the model. The experimental results show that,this model can provide effective basis to the decision makers on personnel rescue,resettlement and emergency material supplying and dispatching.
出处
《地震工程与工程振动》
CSCD
北大核心
2015年第6期226-231,共6页
Earthquake Engineering and Engineering Dynamics
基金
国家科技支撑计划项目(2012BAK15B06-04)
青岛市公共领域科技支撑计划项目(12-1-3-59-nsh)~~
关键词
地震伤亡
粗糙集
支持向量回归机
粒子群优化
BP神经网络
earthquake casualties
rough set
support vector regression
particle swarm optimization
BP network
