【摘要】 本文利用风浪实验数据,采用Welch、MTM、MVDR等方法估计了波面位移和铅直流速的凝聚谱,以期得到适合于波流信号的交叉谱估计方法。数据分析结果表明,三种方法给出的凝聚谱估计在以谱峰频率为中心的较宽频段上基本一致,均高于临界值,表明波流信号在主导波频段上存在显著的相关性;在高频段,MVDR方法估计存在一个类似单谱的指数衰减趋势,而另外两种方法的估计值持续高于临界值。数值模拟结果亦表明MVDR方法有更强的相关信号识别能力和频率分辨能力,且不易出现伪峰,因而估计结果更加可靠。更多还原

【Abstract】 We employed and compared three methods: Welch, MTM, and MVDR, to estimate wave-current cross-spectra, and analyzed three pairs of data of wave and current from our wind-wave experiment. The results show that the magnitude-squared coherence estimated by the three methods is similar in peak frequency, implying that coherence between wave and current in dominant frequency band is significant. In high-frequency band, the estimated magnitude-squared coherence by MVDR method decays as wave spectra and current spectra do, unlike the coherence from the other two methods that stay significant even in very high frequency band. In addition, numerical simulation showed that MVDR method is more capable of distinguishing adjacent cohere frequency without spurious peak, and its result could be more reliable.更多还原