人工智能技术的热带气旋预报综述(之一)——BP神经网络和集成方法的热带气旋预报研究和业务应用  

A review of tropical cyclone forecast based on artificial intelligence(part 1)--BP neural network and ensemble method for tropical cyclone forecast research and operational application

作　　者：金龙[1] 黄颖 姚才[3] 黄小燕 赵华生 Jin Long;Huang Ying;Yao Cai;Huang Xiaoyan;Zhao Huasheng(Guangxi Climate Center,Nanning 530022;Guangxi Institute of Meteorological Sciences,Nanning 530022;Guangxi Meteorological Service,Nanning 530022)

机构地区：[1]广西壮族自治区气候中心,南宁530022 [2]广西壮族自治区气象科学研究所,南宁530022 [3]广西壮族自治区气象局,南宁530022

出　　处：《气象研究与应用》2020年第2期1-6,共6页Journal of Meteorological Research and Application

基　　金：国家自然科学基金(41765002);广西自然科学基金(2018GXNSFAA281229)。

摘　　要：热带气旋路径和强度的客观定量预报是当前热带气旋预报研究和业务预报工作中的重点和难点。简要介绍了有关热带气旋业务预报技术的研究现状,综述了人工智能神经网络技术方法中最广泛使用的BP神经网络模型、进化计算遗传算法-神经网络的热带气旋路径、强度集成预报方法,以及近十多年来,这些方法在实际业务预报中与国内外主要数值预报模式及其他客观预报方法预报性能的对比分析,指出了人工智能神经网络和遗传神经网络集成预报方法等在热带气旋强度、路径预报中的优势和不足,并提出了未来更深入研究需要关注的重点关键问题。The objective and quantitative prediction of tropical cyclone track and intensity is the key and difficult point in the current tropical cyclone prediction research and operational prediction work.This paper briefly introduced the research status of tropical cyclone business forecasting technology,and summarized the most widely used BP neural network model and evolutionary calculation genetic algorithm-ensemble forecasting method of tropical cyclone track and intensity based on neural network.Besides,these methods were compared with the main numerical prediction models and other objective prediction methods in the past decade.The advantages and disadvantages of artificial intelligence neural network and genetic neural network ensemble prediction method in tropical cyclone intensity and track prediction were pointed out,and the key issues that need to be paid attention to in further research were put forward.

关 键 词：人工智能 热带气旋 神经网络 预报建模 

分 类 号：P456[天文地球—大气科学及气象学]