| Detection of Envelope Modulation on Noise and feature extraction of surface and underwater targets are key technologies for target recognition,and they are also one of the research contents that have attracted much attention in the field of underwater acoustics in recent years.Underwater Unmanned Vehicle has attracted much attention due to its advantages such as small environmental restrictions and high flexibility.Sonar is the eyes and ears of UUV.The paper focuses on the characteristics of unmanned systems that cannot be manually interfered.The method for passive detection and estimation of target radiated noise modulation signal based on vector arrays suitable for unmanned platforms lays a foundation for the realization of passive recognition of underwater targets based on UUV platforms.In the processing of DEMON spectrum,the thesis firstly modeled the modulated signal with the characteristics of ship radiation noise,and discussed the DEMON spectrum characteristics of different envelope models.Research the optimal detection algorithm for the vector array of underwater unmanned platform.Combining the idea of taking the local mean value in the sliding window with the adaptive bidirectional filtering technology,the estimated DEMON spectrum is de-continuous spectrum processed,and the DEMON line spectrum is extracted by combining three thresholds.The above processing flow is unattended and can be used to realize the autonomous detection of DEMON spectrum on the UUV platform,which is conducive to the subsequent realization of feature extraction of the target DEMON spectrum.Based on the conventional power spectrum estimation algorithm-average periodogram method,combined with the characteristics of the underwater unmanned platform’s strong mobility and high system real-time requirements,the paper focuses on several processes applicable to the DEMON spectrum detection of unmanned platforms Methods: 1.Aiming at the shortcomings of the average periodogram method that cannot obtain long-term stable signals in actual detection,the researched all-pole AR model spectrum estimation method;2.In order to obtain higher-order signal characteristics in addition to the power spectrum,the research is based on 3/2-dimensional DEMON spectrum estimation algorithm for diagonal slices of third-order cumulants;3.For ocean noise and signals that generally do not completely conform to the Gaussian distribution,the GG-DEMON estimator based on generalized Gaussian probability density distribution and Its improved MGG-DEMON estimator.Several algorithms are compared and analyzed through simulation and long and short sequence sea trial data to select the best DEMON spectrum detection method suitable for UUV vector array.Feature extraction is the key to underwater passive target recognition.Combining the ideas of greatest common divisor and doubling frequency detection,the paper studies a target speed feature extraction algorithm for unmanned platforms with strong anti-interference ability and stability.On the basis of existing research,the energy structure characteristics of the first 10 harmonics of two-blade to nine-blade propellers are given,and the number of blades of the target is calculated by combining the matching degree of the expert system.Finally,based on the sea trial data,the energy characteristics of wavelet packet decomposition under different decomposition levels of A,B,C and D targets are given.It lays the foundation for target recognition using the DEMON spectrum characteristics of ship radiated noise in underwater unmanned platforms. |
PDF Full Text Request