电力推进船舶电力负荷的多变量混沌局部预测

为提高电力推进船舶电力负荷预测精度,提出电力推进船舶电力负荷的多变量混沌局部预测.将相空间重构由单变量时间序列拓展到多变量时间序列,并依据电力推进船舶电力负荷及其相关因素构成的多变量时间序列进行相空间重构.针对每一分量时间序列采用互信息法进行最佳时间延迟的选择,最优嵌入维数则采用虚假邻点法进行确定.根据多变量混沌时序局部预测,提出基于正则化的电力推进船舶电力负荷多变量混沌局部预测.通过对实际船舶电力负荷的计算实例表明,基于多变量时间序列的预测方法比单变量预测具有较好的预测效果.

Chaos local Forecasting of Electric Propulsion Ship Power Load on Multivariate Time Series

Chaotic local forecasting of multivariate time series based on electric propulsion ship power load was proposed for improving the forecasting accuracy of electric propulsion ship power load. Scalar time series was extended to multivariate time series, and the phase space of multivariate time series was reconstructed. The good time delay was chosen for each scalar time series by mutual information. The method to get the minimum embedding dimension is based on the false neighbor. Moreover, a local forecasting model of multivariate chaotic timed series based on the regularized was put forward according to multivariate chaotic local forecasting. The power load of an electric propulsion ship was carried out by the method presented. Through the analysis of the obtained forecasting results it is shown that the forecasting effect of the method presented is better than that of the scalar time series.