| Underwater sound is the only information carrier in the long-range underwater detection and communications. And thus the underwater acoustic plays an important role in the ocean exploration, the ocean development, the ocean utilization, and the marine military defense etc.Underwater acoustic signal is often non-linear and non-stationary, and Hilbert-Huang Transform (HHT) proposed by Dr. Norden E. Huang is an effective method to analyze the non-linear, non-stationary signal. So this dissertation focuses on the application of HHT to the underwater acoustic signal processing, and the research fix attention on the following three aspects:at first, the physical significance of the HHT has been explored, and the related issues of its realization have been discussed. Secondly, the HHT has been used to process the vector signals. And then, the HHT method has applied to detect and identify the defect in the concrete models, and by analyzing experimental data to estimate the effect of using the method to process the underwater acoustic signal. Accordingly, this paper arranged as follows:1. Research has been carried on the related concepts of HHT and its realization. The important concepts using in HHT were expounded in detail, which were the instantaneous frequency, single-component signal, multi-component signal, the time scale and the intrinsic mode function (IMF) and so on. Empirical mode decomposition (EMD) and the Hilbert spectrum have been explained. And some problems involved in the implementation of EMD, such as the curve fitting, the end points forecast and the stop criteria were discussed.2. The practicability research of using EMD in the underwater acoustic processing has been carried. A few problems encountered in signal analysis using EMD has been discussed in this thesis, and some new solutions have been proposed. And the solutions have been verified though analysis the simulation data. And this part includes the following three aspects:1) An adaptive de-noise algorithm has been proposed using the statistical properties of the white noise after EMD analysis, which is the product of the energy density of IMFs and its corresponding average period is a constant. The simulations shown that the proposed method has a good effect in signal de-noise, and it needs not to choose the parameters as the wavelet analysis.2) The research of the frequency resolution of EMD has been carried using EMD to analyze the two-tone signal (the sum of two tone signals). And the conditions of EMD to correctly separate the two-tone signals were given.3) For the mode mixing problem occurred possibly in EMD, a new approach has been put forward which combines the differential operator and the cumulative sum. And the new method has been compared to other algorithm though the simulation and it can achieve good results relatively.3. Some problems using HHT analyze vector hydrophone signal has been discussed. There were three methods to realize the plural EMD, which is positive and negative frequency method, the first-order difference vector method, the projection method. And the three methods have been contrasted though simulation, and the advantages and disadvantages in their implementation were pointed out. Complex EMD was applied to estimate the azimuth of the multi-targets with a single vector. The different plural signal forms have been discussed, and its influence to the Complex EMD has been also discussed; and the application condition of the various signal forms has been obtained.4. The feasibility study of using HHT to detect and identify the defect in the concrete model has been processed. Combined the task of dam defect detection for the underwater robot, a test system has been designed, and it see to collect and store the underwater acoustic echo signal. The EMD algorithm and other methods were used to analysis the data and extract the characteristics of the signals, and the identification of the different defects with concrete model has been realized initially.
|
PDF Full Text Request