Multichannel Adaptive Filter with Assured Stability and Its Applications to Active Noise Control

在具有多个主噪声源、次噪声源、参考传声器和误差传声器的多通道主动噪声控制中，研究了一类新的稳定保证自适应算法，以更新前馈控制器，产生人工控制声音以消除误差传声器处的噪声。结果表明：(1)该鲁棒自适应算法对主信道的任何变化都具有稳定性保证，该算法假定副信道是先验已知的，并给出了可加性不确定性的上界。(2)提出了两种基于辨识的自适应方案来补偿次要路径矩阵和主要路径矩阵的不确定性变化。在第一种方案中，通过对识别的两个等效路径矩阵对应的FFT和逆FFT，可以在线计算二次控制声音。在第二种方案中，我们在适当的时间更新误差系统以调整自适应前馈控制器。(3)在主源噪声为周期性时，研究了一种新的具有稳定性的plug-iii型自适应算法。这种方法不需要参考麦克风，因此将为工业应用提供一个有吸引力的方案。(4)提出了一种新的自适应滤波算法，提高了对时变未知参数的跟踪性能。传统的自适应方法均采用LMS算法，假设未知参数不变。然而，该方法包含一个描述时变参数动态行为的内部模型，可以跟踪未知参数的快速变化。(5)最后，通过风道自适应主动降噪和运动点自适应主动噪声控制两个实验研究，验证了所提自适应方案的有效性。结果与其他常规方法进行了比较。所提出的自适应前馈控制方案也可以推广到受地震加速度激励的结构主动抑制振动控制的应用中。少

In multichannel active noise control with a multiple number of primary noise sources, secondary control sources, reference microphones and error microphones, a class of new stability-guaranteed adaptive algorithms have been investigated to update the feedforward controllers for generating artificial control sounds to cancel the noises at the error microphones. The results obtained by this project can be summarized as follows : (1) The obtained robust adaptive algorithm is stability-guaranteed for any changes of the primary channels by assuming the secondary channels are known a priori and an upper bound of additive uncertainties is given. (2) Two identification-based adaptive schemes have been proposed to compensate for uncertain changes of the secondary path matrices as well as the primary ones. In the first scheme the secondary control sounds can be calculated in an on line way by taking the FFT and inverse FFT corresponding to the identified two equivalent path matrices. In the seco … More nd scheme, we update the error system in adequate timings to adjust the adaptive feedforward controller. (3) A new plug-iii type of adaptive algorithms with assured stability have also been investigated when the primary source noises are periodic. This approach requires no reference microphones and then will provide an attractive scheme for industrial applications. (4) A new adaptive filtering algorithm is derived to improve the tracking performance for time-varying unknown parameters. All of the conventional adaptive approach adopt LMS algorithm assuming that unknown parameters are constant. However, the proposed approach involves an internal model describing the dynamic behavior of time varying parameters and can track rapid changes of the unknown parameters. (5) Finally, we have examined the effectiveness of the proposed adaptive schemes in two experimental studies, for instance, adaptive active noise canceling in an air duct and adaptive active noise control for the moving points. The results are compared with other conventional approaches. The proposed adaptive feedforward control schemes can also be extended to applications of the active vibration suppression control of structures excited by seismic accelerations. Less