基于遗传算法和共轭梯度法结合四维同化方法.PDF

29 6 Vo.l 29 N o. 6 2006 12 Journal ofN anjing Institute ofM eteorology Dec. 2006 : 1000-2022( 2006) 06-0 50-05 1 1 2 白晨, 吴诚鸥, 吴令云 ( 1. , 210044; 2. , 210029) : 对气象资料的四维同化, 采用共轭梯度算法和遗传算法, 可以使两种方法相互补充, 验证明 两者的结合能够增强搜索能力和收敛速度在编写此程序时, 采用标准化编程方法, 使程序能够接 近全自动化, 从而缓解了编程的繁重劳动, 减少程序中出现的错误 : 遗传算法; 共轭梯度; 四维同化 : O1745 : A A Four-D im ension A ssim ilation Schem e Based on Genetic A lgorithm and Conjugate GradientM ethod 1 1 2 BA I Chen , WU Cheng-ou , WU Ling-yun ( 1. Departm ent ofM athem atic, NU IST, Nanjing 210044, China; 2. Institute of Inform ation Science, NAYU, Nanjing 210029, China) Abstract In four-dmi ension assmi ilation, comb in ing conjugate gradient metnod and genetic algorithms ( DA ) cou ld remedy the shortcom ing of each other. It is proved in the expermi ents of th is papar that the scheme not only could strengthen search ing ability but also cou ld accelerate the convergent speed of opt-i m ization. W e adopt standard program in programm ing, therefore the assmi ilation program is almost auto- matic, so w e cou ld be eased from the hard programm ing work and avoid most of the errors. K ey w ord s genetic algorithm; conjugate gradientm ethod; four-dmi ension assmi ilation , LeD-i 0 [ 5] ment , [ 6] [ 7]