主办单位:中国气象局沈阳大气环境研究所
国际刊号:ISSN 1673-503X
国内刊号:CN 21-1531/P

气象与环境学报 ›› 2021, Vol. 37 ›› Issue (3): 139-144.doi: 10.3969/j.issn.1673-503X.2021.03.019

• 快报 • 上一篇    

基于遗传—神经网络方法的广西台风灾害评估模型研究

李艳兰1(),金龙1,史旭明2,陈丹3   

  1. 1. 广西壮族自治区气候中心, 广西 南宁 530022
    2. 桂林航天工业学院, 广西 桂林 541004
    3. 广西壮族自治区气象科学研究所, 广西 南宁 530022
  • 收稿日期:2020-07-15 出版日期:2021-06-30 发布日期:1900-01-01
  • 作者简介:李艳兰, 女, 1972年生, 高级工程师, 主要从事气候监测评价、气象灾害影响评估和气候变化研究, E-mail: gxnnyanlan@163.com
  • 基金资助:
    国家自然科学基金(41565005);广西自然科学基金(2020GXNSFAA297122);广西自然科学基金(2018GXNSFAA281229)

Study on assessment model of typhoon disaster in Guangxi based on genetic-neural network method

Yan-lan LI1(),Long JIN1,Xu-ming SHI2,Dan CHEN3   

  1. 1. Guangxi Climate Center, Nanning 530022, China
    2. Faculty of Science, Guilin University of Aerospace Technology, Guilin 541004, China
    3. Guangxi Institute of Meteorological Science, Nanning 530022, China
  • Received:2020-07-15 Online:2021-06-30 Published:1900-01-01

摘要:

选取1981—2018年影响广西且灾情记录比较完整的86个台风样本,基于台风灾害伤亡人数、直接经济损失划分灾情等级,选取致灾因子,利用遗传算法与神经网络相结合的方法建立广西台风灾害评估模型。结果表明:选取的台风灾害致灾因子与台风灾情等级之间具有显著的相关性,构建的遗传—神经网络集合预报模型对台风灾情预估效果较好,训练样本拟合一致率为86.1%,测试样本预报准确率为71.4%,其中严重和较重的台风灾情等级预报结果与实况基本一致,较轻等级的预报准确率达83.3%。

关键词: 台风灾害, 预评估, 遗传—神经网络, 人工智能

Abstract:

Using 86 typhoon cases that affected Guangxi with relatively complete disaster records from 1981 to 2018, the typhoon disaster was classified and the disaster causing factors were selected based on the number of casualties and direct economic losses.The assessment model of the typhoon disaster in Guangxi was established by combining genetic algorithm and neural network.The results show that there is a significant correlation between the selected disaster factors and the typhoon disaster grades.The genetic-neural network ensemble prediction model which is constructed has a good effect on the typhoon disaster prediction.The fitting consistency rate of training samples is 86.1%, and the prediction accuracy of test samples is 71.4%.Among them, the prediction results of severe and heavy typhoon disaster grades are generally consistent with the actual situation, and the prediction accuracy of lighter grades is 83.3%.

Key words: Typhoon disaster, Pre-assessment, Genetic-neural network method, Artificial intelligence

中图分类号: