基于灰色关联分析与SA-PSO-Elman结合的地震直接经济损失评估  

Earthquake-caused Direct Economic Loss Assessment Based on Grey Relational Analysis and SA-PSO-Elman Neural Network

作　　者：宗学军[1] 李强[1] 杨忠君[1] 何戡[1] Dimiter Velev 

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142 [2]国民与世界经济大学信息与通信技术系,保加利亚索菲亚999139

出　　处：《安全与环境工程》2016年第2期19-22,共4页Safety and Environmental Engineering

基　　金：中国-保加利亚科技合作项目(14-4)

摘　　要：对地震灾害造成的损失进行评估是国家采取应急救援和灾后援建工作的重要依据。为快速评估地震灾害引起的直接经济损失,提出一种基于灰色关联分析与模拟退火-粒子群-Elman神经网络(SA-PSO-Elman)结合的地震灾害直接经济损失评估模型。该模型先采用灰色关联分析方法客观地选出地震灾害直接经济损失的主要影响因素,即为Elman神经网络的输入,然后将全局寻优能力强及收敛速度快的粒子群算法与能跳出局部极值的模拟退火算法相结合来优化Elman神经网络的权值和阀值,最后将训练好的Elman神经网络运用到地震灾害直接经济损失评估中。通过仿真试验结果表明:该混合算法优化的Elman神经网络模型比Elman神经网络模型和PSOElman神经网络模型具有更高的预测精度和收敛速度。Assessment of earthquake-caused loss is an important basis for government to carry out emergency relief and post-disaster reconstruction work.For the rapid assessment of earthquake-caused direct economic loss,this paper puts forward a model based on the grey relational analysis method and Elman neural network combined with particle swarm optimization and simulated annealing（SA-PSO-Elman）.The model first objectively selects the evaluation indexes of earthquake-caused direct economic loss by the grey relational analysis method,which is the input of Elman neural network.Then,the model optimizes the weights and thresholds of Elman neural network by combining the PSO algorithm with the simulated annealing algorithm,of which the former has strong global optimization capability and high convergence speed and the latter has the ability to jump out of the local extreme.Also,the paper uses the trained Elman neural network to predict the earthquake-caused direct economic loss.The experimental results show that SA-PSOElman neural network has higher precision and convergence speed compared with PSO-Elman neural network and Elman neural network.

关 键 词：地震灾害 直接经济损失评估 灰色关联分析 模拟退火算法 粒子群算法 ELMAN神经网络 

分 类 号：X43[环境科学与工程—灾害防治]