Tropical cyclone intensity prediction based on Kolmogorov-Arnold networks with predictor pruning optimizationOA
Tropical cyclone intensity prediction based on Kolmogorov-Arnold networks with predictor pruning optimization
准确的热带气旋(TC)强度预报对减灾和公共安全至关重要.然而,目前TC强度预报存在因子筛选复杂,预报精度不足等问题.为此,本文构建了具备因子剪枝优选能力的Kolmogorov-Arnold网络(KANs)全球TC强度智能预报模型(TCI-KAN).该模型设计了数据驱动的预报因子筛选方法,通过权重排序分析实现了低影响预报因子的迭代剪枝.结果表明,TCI-KAN在6小时TC强度预报中表现优异,独立测试集平均绝对误差(MAE)为2.85 kt,较美国国家飓风中心官方预报,最优单深度学习模型及最优混合模型分别降低31％,13％和6％.进一步分析表明,TCI-KAN适用于不同海域和TC强度类别.
Keyun Li;Wei Zhong;Yao Yao;Fangzhao Li;Yuan Sun;Hongrang He
College of Meteorology and Oceanography,National University of Defense Technology,Changsha,ChinaCollege of Advanced Interdisciplinary Studies,National University of Defense Technology,Changsha,ChinaCollege of Advanced Interdisciplinary Studies,National University of Defense Technology,Changsha,China||School of Atmospheric Sciences,Nanjing University,Nanjing,ChinaCollege of Advanced Interdisciplinary Studies,National University of Defense Technology,Changsha,ChinaCollege of Advanced Interdisciplinary Studies,National University of Defense Technology,Changsha,ChinaCollege of Advanced Interdisciplinary Studies,National University of Defense Technology,Changsha,China
热带气旋Kolmogorov-Arnold网络预报因子剪枝优选强度预报
Tropical cycloneKolmogorov-Arnold networksPredictor pruning optimizationIntensity prediction
《大气和海洋科学快报（英文版）》 2026 (4)
47-52,6
This research was supported by the National Natural Science Foun-dation of China[grant numbers 42075011 and 42192552].
评论