【摘要】 针对水下目标辐射噪声中多线谱成分的检测与频率估计问题,提出一种基于加权交叠平均法和EM算法的水下目标辐射噪声线谱检测与频率估计方法。首先构造目标辐射噪声的数理模型,然后根据加权交叠平均(WOSA)谱估计方法不产生伪峰的特点,判断可能存在的线谱数量和线谱频率,确定数理模型参数,然后采用EM迭代算法得到线谱频率的极大似然估计。仿真结果表明:提出方法计算量较少,使得多线谱频率估计问题的复杂度大大降低,可用于检测和识别水下目标辐射噪声中的多线谱成分,且对于幅度较大的线谱频率估计时收敛速度较快。更多还原

【Abstract】 In order to detect the target radiated noise lines and estimate the frequency of lines,an algorithm based on the weighted overlapped segment averaging method and EM algorithm is presented.The method constructs a receiving noise model,whose parameters are determined by the weighted overlapped segment averaging method.The frequency parameters are estimated by a technique based on the Expectation-Maximization algorithm which,iteratively,produces good approximations to the maximum likelihood estimate.Experimental results of sea trial data show that the proposed method works well for many frequency components presented in target radiated noise.It is concluded that the proposed method converges faster on the component with largest amplitude but lower on the remaining amplitudes.更多还原