Summary
International Symposium on Nonlinear Theory and its Applications
2005
Session Number:3-1-4
Session:
Number:3-1-4-3
Predicting the wind using spatial correlation
Yoshito Hirata, Hideyuki Suzuki, Kazuyuki Aihara,
pp.634-637
Publication Date:2005/10/18
Online ISSN:2188-5079
Summary:
When we attempt to generate more electricity by controlling wind turbines, the first step is to predict the wind. However, it is difficult to predict the wind just from the past observations at that point because the wind does not have temporal correlation much. Here we set up an experimental observation and tried to predict the wind using spatial correlation. By constructing an embedding for multivariate time series and building a nonlinear model, we found that the wind is better predicted from the past observations at neighboring points rather than from those at the predicted point only.