Summary: | 碩士 === 國立成功大學 === 水利及海洋工程學系 === 85 === The wavelet transform was often applied to nonstationary
data to decomposethe signal into a frequency and time frame.
This study istrying to combinethis capability of wavelet
transform with traditional phase difference method (P.D.M.) to
estimate the short time velocity and direction of random
oceanwave. Also, the nonparametric density function
estimation was adopted todetermine the nonstationary
probability density function directly in order toeliminate the
influence of bin width of classical histogram.
For further understanding the meanings of wavelet transform, it
was firstto simulate the waves by superposient frequencies,
and then the wang two ormore regular waves of differvelet power
spectrum and phase spectrum were cal-culated. It was shown
that the wavelet power spectrum/phase spectrum couldonly be
represented by "peak" frequencies for a specific temporal
location.However, the peak frequencies could be the original
input frequencies if thewidth of time-frequency window is not
overlaid, otherwise they will be theinteracting ones of
different frequencies. It is why we calculated the shorttime
wav e velocity and direction for peak frequencies in this study.
Finally, the mean values and probability density function of
short timevelocity and direction of random wave were
calculated and compared to theresults obtained by P.D.M..
Also, the probability density function of shorttime direction
were compared to the directional spreading function estimatedby
Bayesian approach parameters method (B.A.P.M.). The results
show that themean values of short time direction were close to
the values of P.D.M. and themean direction of B.A.P.M., but the
shape of short time direction is lessspreading than the
directional spreading function of B.A.P.M.. The mean valuesof
short time velocity were slower than the estimation of P.D.
M. and alsoslower than the linear wave velocit y of peak
frequency in the ratio of 0.8.
|