全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...
大气科学  2013 

基于主成分分析的人工智能台风路径预报模型

DOI: 10.3878/j.issn.1006-9895.2012.12059

Keywords: 主成分分析,遗传算法,集合预报,气候持续法,台风路径

Full-Text   Cite this paper   Add to My Lib

Abstract:

利用主成分分析可以从具有随机噪声干扰的气象场提取主要信号特征,排除随机干扰的能力,论文以1980~2010年共31年6~9月西行进入南海海域的台风样本为基础,综合考虑台风移动路径的气候持续因子和数值预报产品动力预报因子,采用主成分分析的特征提取与逐步回归计算相结合的预报因子信息数据挖掘技术,以进化计算的遗传算法,生成期望输出相同的多个神经网络个体,建立了一种新的非线性人工智能集合预报模型,进行了分月台风路径预报模型的预报建模研究。在预报建模样本、独立预报样本相同的情况下,分别采用人工智能集合预报方法和气候持续法进行了预报试验,试验对比结果表明,前者较后者在6、7、8和9月份台风路径预报中,平均绝对误差分别下降了7.4%、4.8%、12.4%、17.0%。另外,论文进一步在初选预报因子和样本个例相同的情况下,通过比较新模型与直接采用主成分分析方法选因子并分别运用逐步回归和遗传—神经网络集合预报模型进行计算的预报精度差异表明,前者具有更高的预报精度,其原因是该方法挖掘利用了全部备选预报因子的有用预报信息,而且遗传—神经网络集合预报模型的是由多个神经网络个体预报结果合成,集合模型的各个神经网络个体的网络结构,是通过遗传算法的优化计算确定的,因此,该集合预报模型的泛化能力显著提高,在实际天气预报中具有较好的实用性和推广价值。

References

[1]  Aberson S D, Sampson C R. 2003. On the predictability of tropical cyclone tracks in the Northwest Pacific basin [J]. Mon Wea Rev, 131 (7): 1491-1497.
[2]  Bessafi M A, Lasserre-Bigorry C J, Neumann F, et al. 2002. Statistical prediction of tropical cyclone motion: An analog-CLIPER approach [J]. Wea. Forecasting, 17: 821-831.
[3]  陈刚毅,丁旭羲,赵丽妍. 2005.用模糊神经网络自动识别云的技术研究 [J].大气科学,29(5): 837-844. Chen Gangyi,Ding Xuxi,Zhao Liyan. 2005. An automatic pattern recognition techniques of cloud based on fuzzy neural network [J]. Chinese Journal of Atmospheric Sciences (in Chinese),29(5): 837-844.
[4]  陈国良, 王熙法, 庄镇泉, 等. 1995. 遗传算法及其应用 [M]. 北京: 人民邮电出版社, 28-127. Chen Guoliang, Wang Xifa, Zhuang Zhenquan, et al. 1995. Genetic Algorithms and Its Application (in Chinese) [M]. Beijng: People\'s Posts and Telecommunications Press, 28-127.
[5]  Demarid M, Kaplan J. 1991. A statistical model for predicting tropical cyclone intensity change [C] //19th Conference on Hurricanes and Tropical Meteorology. Miami, Florida, AMS, 521-526.
[6]  丁裕国, 吴息. 1998. 经验正交函数展开气象场收敛性的研究 [J]. 热带气象学报, 4 (4): 316-326. Ding Yuguo, Wu Xi. 1998. Study the convergence for the expansion of meteorological fields with empirical orthogonal functions [J]. Journal of Tropical Meteorology (in Chinese), 4 (4): 316-326.
[7]  丁裕国, 程正泉, 程炳岩. 2002. MSSA-SVD典型回归模型及应用于ENSO预报的试验 [J]. 气象学报, 60 (3): 361-369. Ding Yuguo, Cheng Zhengquan, Cheng Bingyan. 2002. A prediction experiment by using the generalized canonical mixed regression on model based on MSSA-SVD for ENSO [J]. Acta Meteorologica Sinica (in Chinese), 60 (3): 361-369.
[8]  邓爱军, 陶诗言, 陈烈庭. 1989. 我国汛期降水的EOF分析 [J]. 大气科学, 13 (3): 289-295. Deng Aijun, Tao Shiyan, Chen Lieting. 1989. The EOF analysis of rainfall in China during monsoon season [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 13 (3): 289-295.
[9]  郭品文, 居丽丽, 徐同. 2008. 非线性主成分分析在中国四季降水异常分布中的应用 [J]. 南京气象学院学报, 31 (4): 460-467. Guo Pinwen, Ju Lili, Xu Tong. 2008. Application of nonlinear principal component analysis to seasonal precipitation anomaly over China [J]. Journal of Nanjing Institute of Meteorology (in Chinese), 31 (4): 460-467.
[10]  郭章林, 刘明广, 解德才. 2004. 震灾经济损失评估的遗传神经网络模型 [J]. 自然灾害学报, 13 (6): 92-96. Guo Zhanglin, Liu Mingguang, Xie Decai. 2004. Genetic algorithm-neural network-based economic losses assessment of seismic disaster [J]. Journal of Natural Disasters (in Chinese), 13 (6): 92-96.
[11]  Haupt S E, Young G S, Allen C T. 2006. Validation of a receptor-dispersion model coupled with a genetic algorithm using synthetic data [J]. Journal of Applied Meteorology and Climatology, 45 (3): 476-490.
[12]  哈斯巴干,马建文,李启青,等. 2004.容差粗糙集与神经网络结合的遥感数据分类方法 [J].中国科学D辑,34(10): 967-974. Hasi Bagan,Ma Jianwen,Li Qiqing,et al. 2004. Classification the remote sensing data with the method of combined the tolerance rough set and neural network [J]. Science in China Series D (in Chinese),34(10): 967-974.
[13]  黄小燕,金龙. 2007.条件数在台风移动路径预报中的应用 [J].自然灾害学报,16(3): 35-40. Huang Xiaoyan,Jin Long. 2007. Application of condition number in forecasting typhoon motion [J]. Journal of Natural Disasters (in Chinese),16(3): 35-40.
[14]  胡娅敏, 丁一汇, 沈桐立. 2006. 基于遗传算法的四维变分资料同化技术的研究 [J]. 大气科学, 30 (2): 248-256. Hu Yamin, Ding Yihui, Shen Tongli. 2006. A research of four-dimensional variational data assimilation based on genetic algorithm [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 30 (2): 248-256.
[15]  洪梅, 张韧, 吴国雄, 等. 2007. 用遗传算法重构副热带高压特征指数的非线性动力模型 [J]. 大气科学, 31 (2): 346-352. Hong Mei, Zhang Ren, Wu Gguxiong, et al. 2007. A non-linear dynamic system reconstruction of the subtropical high characteristic index based on genetic algorithm [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 31 (2): 346-352.
[16]  金龙, 秦伟良, 姚华栋. 2000. 多步预测的小波神经网络预报模型 [J]. 大气科学, 24 (1): 79-86. Jin Long, Qin Weiliang, Yao Huadong. 2000. A multi-step prediction model of wavelet neural network [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 24 (1): 79-86.
[17]  金龙, 罗莹, 李永华. 2003. 长期天气的人工神经网络混合预报模型研究 [J]. 系统工程学报, 118 (4): 331-336. Jin Long, Luo Ying, Li Yonghua. 2003. Study on mixed prediction model of artificial neural network for long-range weather [J]. Journal of Systems Engineering (in Chinese), 118 (4): 331-336.
[18]  Jin L, Yao C, Huang X Y. 2008. A nonlinear artificial intelligence ensemble prediction model for typhoon intensity [J]. Mon Wea Rev, 136: 4541- 4554.
[19]  金一鸣. 1983. 多预报量双重筛选逐步回归在台风路径预报中的应用[J]. 大气科学, 7 (2): 235-238. Jin Yiming. 1983. Double filter the many predictors by stepwise regression for typhoon tracks forecasting [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 7 (2): 235-238.
[20]  李爱军,李贺军,李克智,等. 2003. C/C复合材料CVI工艺人工神经网络建模 [J].中国科学E辑,33(3): 209-216. Li Aijun,Li Hejun,Li Kezhi,et al.,2003. Modeling of isothermal CVI process of C/C composites by artificial neural network [J]. Science in China (Series E) (in Chinese),33(3): 209-216.
[21]  刘妹琴. 2005.离散时滞标准神经网络模型及其应用 [J].中国科学E辑,35(10): 1031-1048. Liu Meiqin. 2005. Standard discrete time-delay neural network model and its application [J]. Science in China Series E (in Chinese),35(10): 1031-1048.
[22]  吕纯濂,陈舜华,朱永提. 1996.多维动态关联模型在台风路径、强度和风速同时预报中的应用研究 [J].气象学报,54(6): 737-744. Lü Chunlian,Chen Shunhua,Zhu Yongti. 1996. Research and apply multiple dynamic interdependent model (MDIM) to predict typhoon track、intensity and wind-speed [J]. Acta Meteorologica Sinica (in Chinese),54(6): 737-744.
[23]  倪允琪, 金汉良, 薛家元. 1981. 西太平洋台风路径业务数值预告模式及其初步使用结果 [J]. 大气科学, 5 (3): 281-291. Ni Yinqi, Jin Hanliang, Que Jiayuan. 1981. An operational numerical forecasting scheme of typhoon tracks over the western Pacific and its preliminary results [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 5 (3): 281-291.
[24]  Meng Z Y, Chen L S, Xu X D. 2002. Recent progress on tropical cyclone research in China [J]. Advances in Atmospheric Science, 19 (1): 103-110.
[25]  North G R, Bell T L, Cahalan R F, et al. 1982. Sampling errors in the estimation of empirical orthogonal function [J]. Mon Wea Rev, 110: 699-706.
[26]  施能. 2008. 气象科研与预报中的多元分析方法 [M]. 北京: 气象出版社, 72-114. Shi Neng. 2008. Multiple Analysis Method in Scientific Research and Forecast of Meteorology (in Chinese) [M]. Beijing: China Meteorological Press, 72-114.
[27]  王正群, 陈世福, 陈兆乾. 2005. 并行学习神经网络集成方法 [J]. 计算机学报, 28 (3): 402-408. Wang Zhengqun, Chen Shifu, Chen Zhaoqian. 2005. A parallel learning approach for neural network ensemble [J]. Chinese Journal of Computers (in Chinese), 28 (3): 402-408.
[28]  魏鼎文, 张捷迁. 1978. 台风路径的某些模拟实验研究 [J]. 大气科学, 2 (3): 290-296. Wei Dingwen, Zhang Jieqian. 1978. Some of the simulation experiment for typhoon tracks [J]. Chinese Journal of Atmospheric Sciences (in Chinese), 2 (3): 290-296.
[29]  严军,刘健文. 2005.基于神经网络奇异谱分析的ENSO指数预测 [J].大气科学,29(4): 620-626. Yan Jun,Liu Jianwen. 2005. A study of ENSO index prediction based on neural network singular spectrum analysis [J]. Chinese Journal of Atmospheric Sciences (in Chinese),29(4): 620-626.
[30]  张韧, 洪梅, 王辉赞, 等. 2008. 基于遗传算法优化的ENSO指数的动力预报模型反演 [J]. 地球物理学报, 51 (5): 1346-1353. Zhang Ren, Hong Mei, Wang Huizhan, et al. 2008. Retrieval of the non-linear dynamic forecast model of El Nino/La Nina index based on the genetic algorithm optimization [J]. Chinese Journal of Geophysics (in Chinese), 51 (5): 1346-1353.
[31]  Zheng Deling, Liang Ruixin, Zhou Ying, et al. 2003. A chaos genetic algorithm for optimizing an artificial neural network of prediction silicon content in Hot Metal [J]. Journal of University of Science and Technology Beijing, 10 (2): 68: 68-71.
[32]  周家斌, 黄嘉佑. 1997. 近年来中国统计气象学的新进展 [J]. 气象学报, 55: 297-305. Zhou Jiabin, Huang Jiayou. 1997. The new development of meteorology in recent years in China [J]. Acta Meteorologica Sinica (in Chinese), 55: 297-305.
[33]  周秉荣, 李凤霞, 申双和, 等. 2009. 从MODIS资料提取土壤湿度信息的主成分分析方法[J]. 应用气象学报, 20 (1): 114-118. Zhou Bingrong, Li Fengxia, Shen Shuanghe, et al. 2009. Principal component analysis method acquiring soil moisture information from MODIS data [J]. Journal of Applied Meteorological Science (in Chinese), 20 (1): 114-118.
[34]  周志华, 陈世福. 2002. 神经网络集成 [J]. 计算机学报, 25 (1): 1-8. Zhou Zhihua, Chen Shifu. 2002. Neural network ensemble [J]. Chinese Journal of Computers (in Chinese), 25 (1): 1-8.

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133