海洋科学2023，Vol.47Issue(12)：10-20,11.DOI:10.11759/hykx20220525002
基于改进的神经网络方法的风暴潮灾害经济损失预测
Improved neural network-based economic loss prediction of storm surge disaster
摘要
Abstract
Storm surge disasters have serious negative impact on the social and economic development of China's southeast coastal areas and are one of the most serious marine disasters in China.Therefore,it is highly important to establish an accurate and effective loss assessment model for storm surge disaster loss prediction,which is crucial for the prevention and management of storm surge disasters.Based on existing research,this study collects rela-tively complete storm surge disaster-related data of Hainan,Guangdong,Fujian,and Zhejiang provinces on the southeast coast of China from 2000 to 2018.It establishes a complete indicator system for storm surge disaster losses based on comprehensive consideration of risk,the vulnerability of disaster bearers,pregnant environment,and disaster prevention and mitigation capabilities.Compared with a single back propagation(BP)neural network,this study constructs a BP neural network optimized by the differential evolutionary gray wolf algorithm(DEGWO)based on machine learning-related theories and trains and simulates the samples.The results show that the proposed network model demonstrates a smaller error and a higher fit of the data than single BP neural network model,thus improving the accuracy of storm surge disaster loss prediction.These results can provide new insights for the study of storm surge disaster loss prediction and guidance for storm surge disaster prevention and mitigation management.关键词
风暴潮/损失预测/差分进化灰狼算法(DEGWO)/BP神经网络Key words
storm surge disaster/economic loss assessment/differential evolution grey wolf optimization/BP neural network分类
资源环境引用本文复制引用
赵领娣,綦艳玲,王小华..基于改进的神经网络方法的风暴潮灾害经济损失预测[J].海洋科学,2023,47(12):10-20,11.基金项目
国家自然科学基金资助项目(71974176,71473233)[National Natural Science Foundation of China,Nos.71974176,71473233] （71974176,71473233）
