小波多尺度时间序列在船舶和海洋平台运动极短期预报中的应用

Application of Multi-scale Wavelet Theory and Time Series Model in Extreme Short Term Prediction of Ship and Offshore Platform Motions

运用小波多尺度理论,将非平稳时间序列分解为若干层近似意义上的平稳时间序列,使用混沌时间序列Volterra自适应预报模型对每层的单支重构信号进行预报,综合每层的预报值得到原时间序列的预报值,讨论分解层数、小波类型对小波多尺度时间序列法预报效果的影响。仿真结果表明,此方法相比较于传统的时间序列法在预报精度上有了明显的提高。

The multi-scale wavelet theory was used to decompose the non-stationary time series into several layers of stationary time series approximately.The Volterra adaptive prediction model of chaotic time series was applied to predict the signal of each layer,and each predicted layer was integrated to reconstruct the prediction of original time series.The effect of prediction about the choice of decomposed layers and type of wavelet is also discussed.Compared with linear time series method,the method presented has improved the prediction accuracy significantly.