NLMS与RLS算法的仿真比较及其在FECG提取中的应用

该文通过计算机仿真对比研究了归一化最小均方误差（NLMS）和递推最小二乘（RLS）两种自适应滤波算法，并将这两种算法用于胎儿心电图仪的自适应滤波器仿真设计中。该方法通过自适应滤波拾取理想的参考信号，再与腹部混迭信号相减抵消母亲心电图（MECG），从而提取出胎儿心电（FECG）信号。计算机仿真实验结果表明，这两种算法都能通过有效抑制MECG及其它各种干扰以实现FECG的检测。相比之下，RLS算法具有良好的应用性能，除收敛速度快于NLMS以及稳定性强外，还具有更高的起始收敛速率；更小的权失调噪声，更大的抑噪能力，但其计算复杂度高于NLMS算法。

Simulative Comparison of NLMS and RLS Algorithms and Their Application to FECG Abstraction

In this paper, the NLMS algorithms and RLS algorithms are studied briefly through computer simulation by comparison. Furthermore, the NLMS and RLS filtering algorithms have been applied for simulative design of adaptive filter in electrocardiograph. In this study, the desired reference signal is obtained by adaptive filtering, and is subtracted from the abdominal mixture signal to eliminate maternal ECG to abstract FECG. The results of computer simulation show that both NLMS and RLS algorithms can realize the detection FECG by restraining maternal ECG and other disturbances. In comparison the RLS algorithm performance is much better than NLMS algorithm, besides the convergence speed is much faster and the behavior of the RLS filter coefficients is much more stable, it has faster beginning convergence rate, lower misadjustment noise, and better robustness against noise and disturbance, but the RLS algorithm has high computational complexity in comparison.