A Comparative Study Of Stationary Waves Spectrum Analysis

Wave spectrum is a description of the basic characteristics of wave process parameters. It has important theoretical and practical significance to describe spectra and get waves feature using spectra. Many work in fact all need to be done combined with wave spectra, such as, observation and analysis of the waves, wave forecast, marine environmental studies, water port and shipbuilding engineering and so on. Wave spectra is one of the central issues of the current wave research. Due to the complexity of the waves, a small part of the wave spectra are derived from the observed wave elements together with some certain assumptions, while most of the wave spectrum are obtained using site wavefront by special spectral analysis method. Many studies in and abroad about spectra estimation analysis discussed the sampling characteristics and spectral estimation method parameters using the variable controlled method separately, but they did not consider the relationship that may exist between the two. Because previous studies discussions are generally based on field measurement wave data, they also failed to analyze the effect on the calculation result induced by the target spectrum. Based on this, relevant calculation is done in this paper,In this paper, effects of different characteristics on different spectra estimation methods, such as FTA method, Welch method and AR model method, are analyzed by computing some of the random wavefront generated by the linear wave superposition method with different characteristics using different methods. The results show that the factors that influence the spectral estimation are not only characteristic spectral estimation parameters and sample quantity estimation method, but also the characteristics of the target wave spectrum to study to a certain extent. It may be occurred that no matter how the maximum lag of FTA law, the number of segments of Welch law, and the order of AR model law change, we are not satisfied with the spectral estimation results. And when characteristic parameter values are selected appropriately, satisfactory results can be achieved by either three methods. Spectra width of the target spectra has a great effect on the precision curve obtained by Welch method and AR model method, while the effect is less when by the FTA method. The peak of the target spectra impacts the estimation accuracy of all three methods. AR model law fit the target spectra with larger peak better while FTA method and Welch method fit the smaller ones better. Sampling duration has a less effect on the stability of estimation result computed by Welch law and AR model method. Besides by analysis it's found that there is certain relationship between the sampling duration and the best number of segments and the best model order. Sampling length reduction will lead to the loss of the sTab. ility of the estimation result by FTA method. As long as sampling frequency meets the Nyquist sampling frequency standard, it has no significant effect on the estimation result by FTA method and Welch law while there is certain relationship between the optimal order value of AR model method and the sampling frequency.