CAJ | 학술논문

旋转不变子空间法和多重信号分类法需假设背景噪声为独立的高斯白噪声或自相关矩阵已知，当条件不满足时算法的性能急剧下降。针对这一问题，根据矢量水听器多通道输出的特点，提出了一种基于平行因子模型的单矢量水听器方位频率联合估计方法。首先利用矢量水听器各个通道t时刻和t+1时刻的输出数据，计算声压和各振速不同组合时的四阶累积量，并构建三阶平行因子模型；然后分析了 PARAFAC 模型低秩分解的唯一性条件并利用三线性交替最小二乘算法得到了单矢量水听器阵列流形和相位延迟估计，进而得到目标的方位和频率估计。与旋转不变子空间法和多重信号分类法相比，该方法不需要子空间估计和谱峰搜索，在高斯噪声和拉普拉斯噪声背景下对多目标的分辨能力好于ESPRIT算法。仿真和实测数据的分析结果证明了算法的有效性。
선전불변자공간법화다중신호분류법수가설배경조성위독립적고사백조성혹자상관구진이지，당조건불만족시산법적성능급극하강。침대저일문제，근거시량수은기다통도수출적특점，제출료일충기우평행인자모형적단시량수은기방위빈솔연합고계방법。수선이용시량수은기각개통도t시각화t+1시각적수출수거，계산성압화각진속불동조합시적사계루적량，병구건삼계평행인자모형；연후분석료 PARAFAC 모형저질분해적유일성조건병이용삼선성교체최소이승산법득도료단시량수은기진렬류형화상위연지고계，진이득도목표적방위화빈솔고계。여선전불변자공간법화다중신호분류법상비，해방법불수요자공간고계화보봉수색，재고사조성화랍보랍사조성배경하대다목표적분변능력호우ESPRIT산법。방진화실측수거적분석결과증명료산법적유효성。
@@@@ESPRIT (Estimation of Signal Parameter via Rotational Invariance Techniques) and MUSIC (Multiple SIgnal Classification) DOA finding algorithms are based on the hypothesis that the additive noise is independent, identically distributed white Gaussian noise or the noise’s auto correlation matrix is known. When the condition is not satisfied, the performance of the algorithms would drop dramatically. Based on this, according to the multi-output characteristics of a single vector hydrophone, a multi-source DOA (Direction-Of-Arrival) and frequency estimation algorithm based on PARAFAC (PARAllel FACtor) model is proposed. By using the hydrophone’s output data and time-delays between adjacent snapshots, the fourth-order cumulants of sound pressure and velocity data are calculated. And a third-order PARAFAC model is constructed. The uniqueness of high-dimension matrix’s low rank decomposition is analyzed and TALS (Tri-linear Alternating Least Square) method is used to estimate the manifold of the vector hydrophone and phase delays between adjacent outputs. Finally, the sources’ DOA’s and frequencies are obtained. Compared with ESPRIT and MUSIC algorithms, the proposed one doesn’t need subspace estimation or peak searching, and has a better resolution under Gaussian and Laplace noises. The simulation and experiment results verify the efficiency of the algorithm.