Research On Higher-Order Statistics Based Underwater Moving Target Dynamic Spectrum Feature Enhancement

In this dissertation, the theories, algorithms and applications of the underwater moving target(MOT)-radiated dynamic spectrum feature enhancement are deeply investigated. The main contributions of the dissertation are summarized as follows:1 Based on the nonstationarity of the MOT-radiated noise, the fourth order cumulant (FOC) expression method of the MOT-radiated line spectrum signals is investigated. A conception of phase dynamic coupling is defined. The FOC of the MOT-radiated line spectrum signals is theoretically derived. The result has shown that it includes phase dynamic non-coupling component, pairing phase dynamic coupling one and cubic phase dynamic coupling.2 Optimization algorithm of calculating higher-order statistics(HOS) are studied. In this work, quadratic-to-cubic phase coupling is firstly defined. The feature of the fifth order cumulants is investigated. The result shows that the quartic phase coupling component and quadratic-to-cubic phase coupling one are only extracted using the fifth order cumulants of complex signals. In order to decrease the computational complexity of higher-order statistics(HOS), novel approximate estimation theorems of the FOC and the fifth order cumulants are created and proved in theory. The simulation result shows that employing the estimation theorems to optimize the calculation of higher order statistics is an effective method.3 Fourth order cumulant(FOC) different slice-based adaptive dynamic line enhancement algorithms are presented. The convergences of these algorithms are proved in theory. Their output signal-to-noise(SNR) is derived and bigger than that of the traditional adaptive line enhancer(ALE). Simulation with the torpedo-radiated data indicates that the proposed algorithms are much better than the ALE and the short-time correlation based adaptive line enhancer (SCBALE) in suppressing Gaussian noise and enhancing spectrum feature.4 Three new fast algorithms of higher order cumulant (HOC) based adaptive dynamic spectrum enhancer are presented. Based on the advantage of the sign algorithm, a fast algorithm of FOC diagonal slice sign iterative adaptive dynamic line enhancer is proposed. A fast algorithm of weighted HOC slice based adaptive dynamic line enhancer is established. This algorithm integrates the performance of the FOC different slices in handling Gaussian noise. The weighted HOC slice in this algorithm is updated by united sign iteration algorithm. The performance of this algorithm in suppressing Gaussian noise is improved by adjusting the weighting coefficients of the FOC slices. Based on advantages of sign algorithm and coherent integrator algorithm, a fast algorithm of FOC sign coherent integrator based adaptive line enhancer is presented. This algorithm has better tracing feature, steady output, and low complexity.The decreasing computational load of the above-presented each algorithm is more than 50% in comparison with non-sign algorithm. The above algorithms are testified by simulation with real torpedo-radiated noise.5 An algorithm of SKew Signal-to-Noise Ratio (SKSNR) based quadratic phase dynamic coupling signal extraction is created. Slice Signal-to-Noise Ratio(SSNR) and the SKSNR are firstly defined and their mathematical formulas are obtained. The theoretical analysis and simulation show that the SSNR and the SKSNR are equal to or greater than the original Signal-to-Noise Ratio(SNR), and that employing third order cumulant slices can effectively extract the MOT-radiated quadratic phase dynamic coupling signals from asymmetrical distribution noise.6 An algorithm of Kurtosis Signal-to-Noise Ratio(KSNR) based the MOT-radiated noise polyspectrum feature extraction is presented. In this algorithm, a KSNR associated with kurtosis of input signal is defined and its calculating formula is given. The simulation with the MOT-radiated data indicates that this algorithm has strong ability to suppress colored non-Gaussian noise and big signal-to-noise ratio increment.The a...