基于WT和MGFPE混合模式的船舶横摇运动长期预测

为了获得横摇运动在不同时间尺度下的演变规律,提出了利用小波变换和均生函数周期外推混合模式进行船舶横摇运动非线性时间序列长期预测的方法.该方法通过小波变换对横摇运动时间序列进行多分辨率分析,将原序列分解为多个相对简单的准周期信号,信号的趋势项、周期项和随机项被分离出来,然后采用均生函数周期外推预报模式对这些准周期信号进行预报和集成.仿真结果表明,此方法能有效地提高预报长度,并能获得较高的建模及预报精度,该方法亦可用于其他非线性时间序列长期预测.

Long-term prediction of ship roll motion based on hybrid model of WT and MGFPE

A hybrid prediction approach combining wavelet transform(WT) and mean-generating function period extrapolation(MGFPE) is presented and applied to the long-term prediction of nonlinear time sequences of ship roll motion.In order to obtain the evolvement rule of ship roll motion under different time scales,wavelet transform is employed to the multi-resolution analysis(MRA) of time sequences of ship roll motion.The original sequences are decomposed into relatively simple quasiperiodic signals according to the scale.The trend terms,periodic terms and stochastic terms are separated from the original sequences,then the mean-generating function period extrapolation prediction mode is employed to predict and integrate these quasiperiodic signals.Simulation results indicate that this method can improve prediction time and has better precision.This method can also be used for long-term prediction of other nonlinear time sequences.