CAJ | 학술논문

基于加权零吸引因子最小均方算法（RZA-LMS），提出了一种应用于系统辨识的新型自适应滤波算法（ARZA-LMS）。RZA-LMS通过在标准LMS算法迭代过程中添加零吸引因子，促进了滤波器小权系数的收敛，从而在辨识稀疏系统时，加快了算法的整体收敛速度。但是RZA-LMS算法中的零吸引因子，选择了固定的e，过于武断，降低了算法的鲁棒性。通过在参数e与误差信号e之间建立非线性关系，使零吸引因子在最小化MSE更具有灵活性，提出了一种改进的RZA-LMS，提高了对系统辨识的收敛速度和稳定性。最后，计算机仿真验证了新算法的性能明显优于原算法和若干现有稀疏系统辨识的方法。
기우가권령흡인인자최소균방산법（RZA-LMS），제출료일충응용우계통변식적신형자괄응려파산법（ARZA-LMS）。RZA-LMS통과재표준LMS산법질대과정중첨가령흡인인자，촉진료려파기소권계수적수렴，종이재변식희소계통시，가쾌료산법적정체수렴속도。단시RZA-LMS산법중적령흡인인자，선택료고정적e，과우무단，강저료산법적로봉성。통과재삼수e여오차신호e지간건립비선성관계，사령흡인인자재최소화MSE경구유령활성，제출료일충개진적RZA-LMS，제고료대계통변식적수렴속도화은정성。최후，계산궤방진험증료신산법적성능명현우우원산법화약간현유희소계통변식적방법。
Based on RZA-LMS,a novel adaptive algorithm is presented for sparse system identification. The RZA-LMS algorithm generates a zero attractor in the LMS iteration due to the penalty item on coefficients,and the zero attractor promotes sparsity in taps during the filtering process,therefore convergence can be acceler-ated when identifying sparse systems. For the parameter e of the e-law compression in the zero attractor is con-stant,the algorithm is not robust. The proposed approach adaptively establishes nonlinear relationship between the parameter e and the error signal e,which makes the algorithm more flexible in an attempt to minimize the MSE. Simulation results demonstrate the advantages of the proposed filter in both convergence rate and steady-state behaviors under sparsity assumptions on the true coefficient vector.