摘要
传统的变步长最小均方误差(VS-LMS)算法在自适应稳态阶段步长变化较大,为有效解决这一问题,对算法中的参数进行了修改,提出了一种改进的变步长自适应谱线增强算法。新算法根据步长调整原则,通过改进步长因子与误差信号之间的非线性函数关系,使其初始阶段和时变阶段步长自适应增大、稳态阶段步长减小。经理论分析和仿真实验验证,新算法的时域和频域波形信息均优于原算法,具有良好的降噪性和稳定性以及较快的收敛速度,可更好地应用于自适应系统中。
With traditional variable step-size least mean square（VS-LMS） error algorithm,the step size in adaptive steady state phase changes a lot.To effectively solve this problem,the revision is made for the parameters in algorithm,and the improved variable step-size adaptive spectral-line enhancement algorithm is proposed.In new algorithm,according to the step-size adjustment criteria,through improving the nonlinear functional relationship between step-size factor and error signal,to make the step size in initial phase and time varying phase increase adaptively,while the step size in steady phase decrease.The theoretical analysis and simulated experiment verifies that the waveform information of new algorithm in time domain and frequency domain are better than those of original algorithm.The new algorithm features excellent denoising performance and stability,and faster convergence speed,is can be well applied in adaptive systems.
出处
《自动化仪表》
CAS
北大核心
2011年第3期26-28,共3页
Process Automation Instrumentation
关键词
自适应滤波
谱线增强
降噪
最小均方误差
非线性
Adaptive filtering Spectral-line enhancement Denoising Least mean square error Nonlinear
