摘要
针对目前船舶极短期运动预报中以真实值与各理论预报算法的均方差最小来确定最优预报算法可能导致真正的最优算法被排除的情况,采用波形匹配和仿射变换提出一种改进的精度分析方法,用于船舶极短期运动预报的最优算法确定。该方法首先识别预报信号的周期和波数等关键波形参数,通过仿射变换为一个与真实信号同周期同相位的新信号,进而计算船舶运动时间序列间真实值与新信号的均方差,从而确定船舶极短期运动预报的最优算法。实际应用表明,该方法可以更合理地确定船舶极短期运动预报的最优算法。
To determine the best prediction method for very-short term prediction of ship motions in a specific condition, the accuracies of the candidate methods are usually evaluated by comparing their standard deviations. However, this method may exclude the best one ＇de facto＇ in some cases. In this paper, a method is proposed for the determination by matching the waveform of the predicted signal and through affine transformation. The major characteristics such the period and the wave numbers of a predicted signal are identified. Then the qualified predictions are converted via affine transformation to one with the same period and same phase angle with the actual signal. Finally the best prediction method is then determined by comparing the standard deviations. A validation case shows that the proposed method is able to more reasonably determine the best method for very-short term prediction of ship motions.
出处
《船舶工程》
CSCD
北大核心
2017年第9期40-43,共4页
Ship Engineering
基金
黄岛区科技局自主创新重大专项支持(编号:2015-1-6)
关键词
船舶运动
极短期预报
最小均方差
波形匹配
仿射变换
ship motions
very-short term prediction
standard deviation
waveform matching
affine transformation
