多项式结构宽带波束形成器的性能分析及其改进

首先从理论上分析了多项式结构宽带波束形成器的鲁棒性能。分别从传声器失配误差影响下阵列波束图扰动的统计分析以及白噪声增益分析两个方面揭示了影响多项式结构宽带波束形成器鲁棒性能的内在规律。发现当波束主瓣指向处于可调范围边缘时,与指向中心相比,多项式结构波束形成器的鲁棒性能变差。然后分析了影响多项式结构宽带波束形成器阵列指向性指数的主要因素,指出了多项式结构的截断误差会导致宽带波束形成器的主瓣指向出现偏差,进而造成阵列指向性指数性能下降。为了克服现有设计结构存在的缺陷,提出了一种具有波束主瓣指向自校正能力的改进设计结构。与现有设计结构不同,所提出的设计结构中的调向功能模块能够充分反映出波束期望指向和实际指向之间存在的非线性关系,避免了现有设计方法因采用线性调向功能模块而造成较大波束指向偏差的问题,有效改善了多项式结构宽带波束形成器的性能。      

融合插值点优化的多项式结构宽带波束形成器设计方法*

多项式结构设计方法是主瓣指向可调宽带波束形成器设计的一类重要方法。多项式结构的阶数是有限的,导致主瓣实际指向与期望指向之间存在偏差,因而影响了波束形成器的指向性指数。针对这一问题,该文提出了一种基于插值点优化的多项式结构宽带波束形成器设计方法。首先,引入多项式结构插值点处阵列响应的空间导数约束,以减小主瓣指向偏差;进而利用粒子群优化算法对多项式结构中的插值点进行优化,以充分利用插值点位置提供的自由度进一步提升多项式结构宽带波束形成器的性能。优化设计结果表明,与现有设计方法相比,该文提出的方法不仅降低了主瓣的指向偏差,同时也提高了指向性指数,有效改善了多项式结构宽带波束形成器的性能。     

稳健宽带波束形成器设计的低阶统计量法*

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已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。     

协方差矩阵重构的稳健自适应波束形成算法

针对协方差矩阵含有期望信号成分以及波束指向角失配时,传统自适应波束形成器性能严重下降的问题,提出了协方差矩阵重构的稳健自适应波束形成算法。该算法将全空域划分成若干互不重叠的区域,分别对应干扰区域与信号区域,先利用Capon波束形成器对干扰区域积分,由此构造出干扰协方差矩阵。然后,利用标准Capon波束形成器的波束域MUSIC谱估计法对信号区域积分,重构出信号协方差矩阵,以其主特征向量作为期望信号导引向量估计。由于算法重构了干扰加噪声协方差矩阵并对导引向量进行了修正,保证了自适应波束形成器的性能。理论分析和仿真实验结果表明,算法在训练数据含有期望信号成分和波束指向角度失配情况下具有良好的性能。      

频率不变波束形成器抽头稀疏化设计的交替方向乘子法

针对具有空间响应变化函数约束的频率不变波束形成器设计问题,提出了采用交替方向乘子法实现抽头稀疏设计的优化算法。该算法利用交替方向乘子法能够将原始优化问题进行分裂处理的特点,通过引入替代变量和指示函数,使得表征波束形成器抽头稀疏度量的非凸L₀范数与阵列响应约束分离,进而将问题分裂到元素层级并给出近邻算子的解。对于指示函数的近邻算子求解,在分裂到元素层级后则退化为简单的双边约束问题,因而降低了优化求解的计算复杂度。仿真分析表明,提出的方法比现有的L₁范数方法在宽频带条件下的抽头稀疏度能够提升6%~13%,通带最大波动误差减小了约2 dB,并且优化消耗时间更短。实验结果进一步验证了所提方法在实现高抽头稀疏度波束形成的同时,对声信号造成的失真更小。因此,所提出的方法在降低传声器阵列波束形成器的实现复杂度以及保持阵列响应的频率不变性能方面更具有优势。      

一阶指向可调差分阵列的前后向比分析与优化设计

一阶指向可调差分传声器阵列具有尺寸小、主瓣指向灵活可调、以及阵列响应不随频率变化等优点,因而在音频处理领域得到了重要关注。但差分传声器阵列对阵元失配误差较为敏感,在实际设计中需要予以考虑。前后向比直接反映了传声器阵列对后向噪声干扰的抑制程度,是差分传声器阵列设计中的重要指标。本文从理论上深入剖析了失配误差对一阶指向可调差分传声器阵列前后向比性能的影响,揭示了其中蕴含的规律。在此基础上,针对实际中同时存在随机失配误差情况,提出了一种最差性能优化设计方法。仿真实验和实测实验验证了本文理论分析的正确性和所提优化设计方法的有效性。      

阵列指向性二次型约束稳健波束形成算法

提出了一种基于阵列指向性二次型约束的优化算法,以提高最小方差无畸变响应(MVDR)波束形成对方向矢量的稳健性。通过控制阵列波束指向附近的小区域的自适应波束图与期望波束图的加权平方误差,使波束主瓣区域内的信号畸变最小,同时保持了对约束区间外的干扰信号的抑制能力。数值结果表明:在理想阵形和阵形畸变情况下,无方位失配时,利用本文算法的阵列输出信号与干扰噪声比(SINR)与新近出现的球面约束稳健Capon方法相当,而存在方位失配时,本文算法均优于后者。     

一种加权稀疏约束稳健Capon波束形成方法

为了克服标准Capon波束形成器旁瓣级高以及存在角度失配时性能急剧下降等缺点, 在稀疏约束Capon波束形成器的基础上, 提出了一种加权稀疏约束Capon波束形成器. 该方法利用波束响应的稀疏分布特性, 在标准Capon波束形成优化模型中加入旁瓣区域波束响应稀疏约束(l₁ 范数约束), 使旁瓣区域波束响应向量中非零元素的个数最小化; 通过阵列采样数据协方差矩阵特征分解得到信号子空间及噪声子空间, 利用信号子空间与噪声子空间的正交特性, 构造加权矩阵对稀疏约束进行加权, 使得稀疏重构时波束响应向量中不同角度对应的元素得到不同程度的约束. 该方法有效地抑制了Capon波束形成器的高旁瓣级, 加深了干扰方位零陷, 提高了阵列输出信干噪比. 由于稀疏约束, 波束响应向主瓣集中, 期望信号方向附近的波束响应都较大, 从而也提高了阵列抗导向矢量角度失配的能力. 数值仿真和水池实验验证了所提方法的有效性.     

小尺寸声矢量平面阵三维波束形成方法

提出了一种基于声场多极子展开的小尺寸声矢量平面阵三维波束形成方法。该方法利用了声场的垂直振速信息,并结合球谐波函数与多极子模态之间的联系,可用二维的阵列结构合成指向任意方向的三维波束图,简化了三维波束形成所需的阵列结构。以3×3均匀矩形阵为例的仿真结果表明,在相邻阵元间距小于0.2倍信号波长时,所提方法引入的差分近似误差可被忽略,所得阵列波束图接近理论波束图。相比于声压球面阵三维波束形成方法,所提方法在低频段能在获得相同阵增益的同时有着更好的稳健性,为实现小尺寸阵列的三维波束形成提供了一种可靠的方法。     

基于干扰协方差矩阵重构的恒定束宽鲁棒自适应波束形成

针对宽带波束形成中的恒定束宽波束响应优化设计问题与鲁棒性问题展开研究。首先,提出一种基于相位补偿的恒定束宽全局优化设计方法,通过对阵列流形向量进行相位补偿来设计恒定束宽波束,与现有的一些方法相比,该方法不仅能获得全局最优解,而且物理实现简单。同时,还提出一种基于协方差矩阵重构的鲁棒自适应宽带波束形成算法。该算法采用Capon估计器估计样本数据的空间一频率谱密度函数,然后对期望信号波达方向之外的角度区间进行积分来重构干扰加噪声协方差矩阵,最后利用重构的协方差矩阵设计自适应波束形成器权系数。该波束形成器设计问题被表述成凸优化问题求解。仿真结果表明,在整个输入信噪比范围内,该算法几乎都能获得接近理想值的输出信干噪比。      

Design of robust broadband beamformers using worst-case performance optimization for time-domain spherical microphone arrays

Spherical array modal beamformers are known to be sensitive to microphone mismatches. To combat the problem, robust design of spherical broadband beamformers with the time-domain implementation structure using the Worst-Case Performance Optimization(WCPO) is studied. It is shown that the conservativeness of the upper error bound on beamforming response is the primary factor to degrade the performance of the existing WCPO-based approach. Then an improved design approach for spherical broadband beamformers is presented using a stricter upper error bound on beamforming response. Theoretical analysis shows that the cost function of the proposed design is smaller and the range of feasible set of the proposed design is also larger when compared with the existing design. Moreover, a scheme on constraint specification has been proposed to reduce fluctuation of beamforming response over frequency at steering direction to alleviate signal distortion. Simulation results show that the proposed design outperforms its existing counterparts and exhibits a lower sidelobe level under the same condition of microphone mismatches.     

Design of broadband beamformers based on an expected response interpolation technique

A method for designing broadband beamformers based on an expected response interpolation technique was proposed, with the aim of reducing the heavy optimization burden in the broadband beamformer design problem. In the method, some typical frequencies in the designing frequency band are selected and the optimal array weights are designed on these typical frequencies, based on which a broadband expected response function is constructed. Then, FIR filters are designed to realize the broadband beamforming over the whole designing frequency band, via the interpolation technique which is utilized to the expected responses of the FIR filters. Broadband beamforming can be fulfilled by applying these filters to each of the sensors in the array. By using the proposed method, designers need not to divide the whole designing frequency band into narrow subbands so as to make the narrow band assumption valid, which not only reduces the heavy designing burden, but also releases the requirement for measurements at fine frequency grids in the robust beamformer design. Results of both computer simulations and watertank experiments verify the effectiveness of the proposed method.     

Nearfield broadband array design using a radially invariant modal expansion

This paper introduces an efficient parameterization for the nearfield broadband beamforming problem with a single parameter to focus the beamformer to a desired operating radius and another set of parameters to control the actual broadband beampattern shape. The parameterization is based on an orthogonal basis set of elementary beampatterns by which an arbitrary beampattern can be constructed. A set of elementary beamformers are then designed for each elementary beampattern and the desired beamformer is constructed by summing the elementary beamformers with frequency and source-array distance dependent weights. An important consequence of our result is that the beamformer can be factored into three levels of filtering: (i) beampattern independent elementary beamformers; (ii) beampattern shape dependent filters; and (iii) radial focusing filters where a single parameter can be adjusted to focus the array to a desired radial distance from the array origin. As an illustration the method is applied to the problem of producing a practical array design that achieves a frequency invariant beampattern over the frequency range of 1:10 (which is suitable for speech acquisition using a microphone array), and with the array focused either to farfield or nearfield where at the lowest frequency the radial distance to the source is only three wavelengths.     

基于自适应加权约束最小二乘法的麦克风阵列稳健频率不变波束形成算法

为了解决麦克风阵列通道失配时波束形成算法的稳健性问题, 提出一种基于自适应加权约束最小二乘法的麦克风阵列稳健频率不变波束形成算法. 该算法在分析无通道失配和通道失配时阵列模型特点基础上, 深入研究了通道失配时约束最小二乘频率不变波束形成算法存在的问题及其产生的原因; 将麦克风特性的概率密度函数作为稳健因子加入到约束最小二乘频率不变波束形成算法后, 其频率不变性的稳健性得到了一定的提高, 但稳健性仍较差. 为了进一步提高约束最小二乘法频率不变波束形成算法的稳键性, 通过定义代价函数中控制频率不变性的动态加权系数来调节旁瓣频谱能量, 大大提高了频率不变波束形成算法的稳键性, 将频率不变的频带范围内同一到达角度上不同频率所形成的阵列响应的最大值与最小值之比定义为波动误差, 并作为比较本文算法与约束最小二乘稳健波束形成算法和minmax稳健波束形成算法在通道失配时频率不变性稳键性的评价指标. 算法实例验证结果表明, 在麦克风阵列通道失配时, 本文算法的波动误差最小、频率不变波束形成稳健性最好, 而且适用于任意结构的阵列.     

Optimal array pattern synthesis for broadband arrays

Broadband beamformers with constant mainlobe response over the frequency of interest are desirable in many applications including underwater acoustics, ultrasonics, acoustic imaging and communications, and so on. Solutions to this problem have been presented for specific array geometry often requiring a larger number of sensors. And the array pattern synthesis error minimization is employed for the whole field of view, which leads to suboptimal designs. In this paper, a broadband array pattern synthesis approach to designing time-domain constant mainlobe response beamformer is proposed. By imposing constraints both on the mainlobe spatial response variation over frequency and on the sidelobes of the beamformer, several optimization criteria are presented and the corresponding convex second-order cone programming implementations are given. In this approach, no preliminary desired beampattern is required and the beam responses variation minimization is employed only in the mainlobe region and not in the sidelobe regions, which improves the beamformer mainlobe synthesis accuracy. Equally, one can obtain lower sidelobes at the same mainlobe synthesis accuracy. This approach is applicable to arrays with arbitrary geometry. Simulation and experimental results are presented to illustrate the effectiveness of this approach. Performance comparisons of the proposed beamformers and the existing beamformer are also provided.     

采用球谐分解的二范数广义逆波束形成

基于传声器阵列的声成像技术是解决噪声源识别的有效途径之一.该文提出了一种基于球谐分解的L2范数广义逆波束形成算法,并对此算法在分布式球形阵列布放方案下进行了定位精度及鲁棒性的对比分析研究.仿真结果显示,此算法对低频相干声源具有较高的空间定位精确度,且阵元位置误差对此算法性能的影响有限.通过在半消声室进行实验进一步证明了...     

Acoustic analysis by spherical microphone array processing of room impulse responses

Spherical microphone arrays have been recently used for room acoustics analysis, to detect the direction-of-arrival of early room reflections, and compute directional room impulse responses and other spatial room acoustics parameters. Previous works presented methods for room acoustics analysis using spherical arrays that are based on beamforming, e.g., delay-and-sum, regular beamforming, and Dolph-Chebyshev beamforming. Although beamforming methods provide useful directional selectivity, optimal array processing methods can provide enhanced performance. However, these algorithms require an array cross-spectrum matrix with a full rank, while array data based on room impulse responses may not satisfy this condition due to the single frame data. This paper presents a smoothing technique for the cross-spectrum matrix in the frequency domain, designed for spherical microphone arrays, that can solve the problem of low rank when using room impulse response data, therefore facilitating the use of optimal array processing methods. Frequency smoothing is shown to be performed effectively using spherical arrays, due to the decoupling of frequency and angular components in the spherical harmonics domain. Experimental study with data measured in a real auditorium illustrates the performance of optimal array processing methods such as MUSIC and MVDR compared to beamforming.