基于总体最小二乘法的混合滤波器组误差补偿算法

模拟分析滤波器的实现误差以及各子带间的增益、相位失配误差的存在都将影响混合滤波器组（HFB）的完美重构效果。针对以往矫正方法对信号源稳定度要求高、调试复杂的缺点,提出一种模数混合的HFB失配误差矫正系统结构。利用伪随机噪声覆盖信号全谱且功率谱密度为常数的特点作为矫正信号源,根据其二阶统计量计算期望的补偿器频域响应。提出一种新的基于总体最小二乘（TLS）算法的IIR形式补偿器设计方法,推导了以IIR滤波器系数为未知变量的线性方程组,并利用TLS算法有效抑制设计误差以及量化噪声等扰动因素影响,优化滤波器系数,减小失配误差影响。仿真结果表明该算法的结构简单、易于实现,同时能有效改善HFB的重构效果。

Hybrid Filter Banks Compensation Design Using Total Least Squares Solution

The problems caused by analogy analysis filters realization error and gain phase error usually impact the reconstruction performance of the hybrid filter banks（HFB）.A novel approach for designing analogy-digital compensator system is given to replace the old method which is very complicated and need higher stabilized signal recourse.The additive Gaussian white noise is used as the input signal and calculating the transform of compensator utilizing the property of power spectrum density is a constant and throughout the whole spectrum.A novel criterion for designing IIR-type compensator based on Total Least Squares（TLS） is proposed,which can optimize the coefficients of filters.The disturbers which caused by realization and random noise can be restrained by means of the TLS.Simulation results show the effectiveness and reliability of this approach.