QR分解的最小二乘格型自适应滤波算法在噪声主动控制中的应用

QR-decomposition-based least squares lattice adaptive filter algorithms for active noise control

在噪声主动控制系统中,滤波-x递归最小二乘(FxRLS)算法收敛速度快但计算量大。基于此,提出了格型联合估计滤波器结构与基于QR分解的最小二乘格型(QRD-LSL)自适应滤波算法相结合的噪声控制方法,该方法对联合估计过程进行了改进并得到了基于各阶估计误差的联合过程估计权系数更新关系,格型联合估计器结构简单,QRD-LSL自适应滤波算法数值稳定性好。仿真结果表明提出的噪声控制方法有良好的噪声控制效果,收敛速度快,计算量小,稳态误差小,跟踪性能好。

In active noise control systems, the Filtered-x Recursive Least Squares （FxRLS） algorithm has a fast convergence speed, but large computation cost. In this paper a combination method using the lattice structure of joint-process estimation and QR-Decomposition-based Least Squares Lattice （QRD-LSL） active filter algorithms is proposed to control unwanted noise. The proposed method develops the joint-process estimation and obtains the new weighted coefficients update relationship based on the individual error estimation. The Lattice structure of joint-process estimation is simple. The QRD-LSL active filter algorithm has good numerical properties. Simulation results show that there are good noise control performances in the pro- posed method, such as fast convergence speed, short computation time, small steady-state error and good tracking capability.