由于传声器阵列通常对阵元失配误差较为敏感,因此稳健波束形成器的设计己成为传声器阵列处理领域的研究热点之一。概率密度法是目前传声器阵列稳健波束形成器设计中的一类重要方法,但该方法所需的阵元失配误差的概率密度信息在实际中较难获取。针对这一问题,本文研究了基于阵元失配误差低阶统计量的稳健波束形成器设计方法,该方法仅利用在实际中较易获取的阵元失配误差的一阶和二阶统计量信息。本文分别研究了基于阵元失配误差低阶统计量的固定权和变加权最小二乘波束形成器设计,给出了两种波束形成器的相关设计理论。理论和仿真分析表明,在小误差条件下,低阶统计量法所设计的波束形成器仍保持与概率密度法相当的性能。
Since microphone arrays are highly sensitive to microphone mismatch errors, robust beamformer design has been studied extensively in the microphone array processing field. At present, the probability density function (PDF) based method is one of the important methods in robust beamformer design for microphone arrays. However, the PDF of microphone mismatch errors cannot be easily obtained in practice. To overcome this drawback, this paper studies a robust design method based on lower-order statistics of microphone mis- match errors. The proposed method uses only the first- and second-order statistics which are easier to obtain in practice. The design methodologies for fixed-weighted and varying-weighted least squares beamformers based on lower-order statistics of microphone mismatch errors are presented, respectively. Theoretical and simulation analysis show that the performance of the proposed beamformers is comparable to that of the beamformers designed via the PDF method under the condition of small microphone mismatch errors.