基于SVM-BP神经网络的风暴潮灾害损失预评估

发布时间：2018-02-16 04:00

本文关键词： 风暴潮损失预评估支持向量机 BP神经网络组合预测　出处：《海洋环境科学》2017年04期 　论文类型：期刊论文

【摘要】：风暴潮灾害是影响我国最严重的海洋灾害,风暴潮灾害损失的预评估对防灾减灾有重要作用。本文选用2002~2014年的40组风暴潮历史灾情资料进行试验,首先建立风暴潮灾害损失评估指标体系并用灰色关联分析法对指标进行筛选,然后采用最优权重组合将支持向量机和BP神经网络进行组合预测分别对风暴潮直接经济损失和受灾人口数进行预测,并与单一预测方法进行对比,发现组合预测方法可以降低误差,提高损失预测的准确性,建立风暴潮灾害损失预评估模型,为决策者进行预警信息的发布提供有效依据。
[Abstract]:Storm surge disaster is the most serious marine disaster in China, and the pre-assessment of storm surge disaster loss plays an important role in disaster prevention and mitigation. 40 groups of historical disaster data of storm surge from 2002 to 2014 are selected in this paper. First of all, the index system of storm surge disaster loss assessment is established, and the grey relational analysis method is used to screen the index. Then the combination of support vector machine and BP neural network is used to forecast the direct economic loss of storm surge and the number of affected population respectively and compared with the single forecasting method. It is found that the combined forecasting method can reduce the error, improve the accuracy of the loss prediction, establish the pre-assessment model of storm surge disaster loss, and provide an effective basis for the decision makers to release the early warning information.
【作者单位】：中国海洋大学工程学院土木工程系;
【基金】：国家自然科学基金(41072176,41371496) 国家科技支撑计划项目(2013BAK05B04)
【分类号】：P731.23

【相似文献】