多维非线性时间序列的投影寻踪学习逼近

The Projection Pursuit Learning to Approximate Multiple Dimensional Non-linear Time Series

建立了多维非线性时间序列投影寻踪学习网络结构及算法，证明了投影寻踪学习网络可以以任意精度逼近多维非线性时间序列，解决了基于投影寻踪学习的多维非线性时间序列的建模和预测问题，实际应用例子表明该算法可行。

The projection pursuit learning network to approximate multiple dimensional non-linear time series and the algorithm are presented in this paper, which can approximate the modes of multiple dimensional non-linear time series at any accuracy. The modeling and forecasting of multiple dimensional non-linear time series based on projection pursuit are solved, and a real example indicates that the algorithm is feasible.