Predicting multivariate time series in real time with confidence intervals: Applications to renewable energy

详细信息    查看全文

• 作者：Yoshito Hirata (1) (2) (3)
  Kazuyuki Aihara (1) (2) (3)
  Hideyuki Suzuki (1) (2) (3)
  
• 刊名：The European Physical Journal - Special Topics
• 出版年：2014
• 出版时间：October 2014
• 年：2014
• 卷：223
• 期：12
• 页码：2451-2460
• 全文大小：324 KB
• 参考文献：1. Y. Hirata, T. Yamada, J. Takahashi, H. Suzuki, Phys. Lett. A 376, 3092 (2012) CrossRef
  2. Y. Hirata, T. Yamada, J. Takahashi, K. Aihara, H. Suzuki, Renew. Energy 67, 35 (2014) CrossRef
  3. F. Kwasniok, L.A. Smith, Phys. Rev. Lett. 92, 164101 (2004) CrossRef
  4. T. Yamada, J. Takahashi, K. Aihara, Reliability Eng. Ass. Jpn. 28, 489 (2006) (in Japanese)
  5. H. Kantz, T. Schreiber, / Nonlinear Time Series Analysis (Cambridge University Press, 2004)
  6. D. Fran?ios, / High-dimensional Data Analysis (VDM Verlag, 2008)
  7. E. Lorenz, Proc. Seminar Predictability, ECMWF 1 (1996)
  8. J.A. Hansen, L.A. Smith, J. Atmos. Sci. 57, 2859 (2000) CrossRef
  9. L. Cao, A. Mees, K. Judd, Physica D 121, 75 (1998) CrossRef
  10. S. Boccaletti, D.L. Valadares, L.M. Pecora, H.P. Geffert, T. Carroll, Phys. Rev. E 65, 035204(R) (2002) CrossRef
  11. S.P. Garcia, J.S. Almeida, Phys. Rev. E 72, 027205 (2005) CrossRef
  12. Y. Hirata, H. Suzuki, K. Aihara, Phys. Rev. E 74, 026202 (2006) CrossRef
  13. I. Vlachos, D. Kugiumtzis, Phys. Rev. E 82, 016207 (2010) CrossRef
  
• 作者单位：Yoshito Hirata (1) (2) (3)
  Kazuyuki Aihara (1) (2) (3)
  Hideyuki Suzuki (1) (2) (3)
  
  1. Institute of Industrial Science, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo, 153-8505, Japan
  2. Department of Mathematical Informatics, Graduate School of Information Science and Technology, The University of Tokyo, Tokyo, 153-8505, Japan
  3. CREST, JST, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan
  
• ISSN：1951-6401

文摘

We extend our earlier work on predicting a univariate time series in real time with confidence intervals (Hirata, et al., Renew. Energy 67, 35 (2014)) to a multivariate time series. We realize this extension by using the -em class="a-plus-plus">p-norm-where p is smaller than 1. We compare the performance when p is 0.5 with that when p is 2 using solar irradiation data and wind data measured all over Japan.