摘要
采用ISOMAP-Copula集成方法,分别构造台风灾害的致灾源指数、承灾体指数、防灾减灾指数及灾情指数。借助Copula熵研究台风灾情与影响因子的相关性,构造T检验的阈值,筛选与灾情显著相关的影响因子。构建的ELM和PSO-ELM模型,分别应用于广西台风灾情的预测。结果表明,基于Copula熵影响因子选取的PSO-ELM灾情预测模型的拟合度(0.9121)最高,均方误差(0.0044)最小,说明该模型的预测精度最高,可为广西台风灾情的预测提供一种新的途径。
Using ISOMAP-Copula integration method, the source index, body index, disaster prevention and mitigation index and disaster situation index of typhoon disasters are constructed respectively. Copula Entropy is used to study the correlation between typhoon disaster and its influencing factors, and T test threshold is constructed to screen the influencing factors which are significantly related to the disaster. The ELM and PSO-ELM models are applied to predict the typhoon disaster in Guangxi. The results show that PSO-ELM disaster prediction model based on Copula entropy influence factor has the highest fitting degree (0.9121) and the smallest mean square error (0.0044), indicating that the prediction accuracy of the model is the highest and the model can provide a new way for typhoon disaster prediction in Guangxi.
作者
陈燕璇
刘合香
倪增华
Chen Yanxuan;Liu Hexiang;Ni Zenghua(Schod of Mathematics and Statistics Sciences,Nanningg normal University,Nanning 530023;Key Laboratory of Marine Disaster Research in Beibu Gulf,Qinzhou Guangxi 535011;Nanning University,Nanning 530023)
出处
《气象研究与应用》
2019年第2期7-11,55,共6页
Journal of Meteorological Research and Application
基金
国家自然科学基金项目（41665006,41465003）
南宁学院科研项目（2018XJ43)
作者简介
陈燕璇(1988-),女,硕士研究生,主要从事统计建模研究。E-mail:hx_post@163.com;通讯作者:刘合香(1962-),女,教授,硕士生导师,主要从事概率统计、数学模型、自然灾害风险分析研究。E-mail:hx_post@163.com.
