摘要
由于横向时延滤波和IQ正交滤波在稳定性、收敛速度和误差率等方面的性能差异对于最小均方算法(LMS,Least Mean Squares)在工程上的实际应用具有参考意义,采用数值仿真的方法模拟了两者实现自适应最小均方算法的具体过程,并对比分析了它们的滤波性能。仿真结果显示IQ正交滤波具有更高的稳定性、更快的收敛速度、更小的误差率,其滤波性能优于横向时延滤波。IQ正交滤波器的阶数为2阶,而横向时延滤波器的阶数通常>10,滤波器阶数越多引入的权噪声也会相对增加,对于稳定性、收敛速度和误差率都是不利的,因此实际应用中优先选择阶数较少的IQ正交滤波器。
Adaptive filter can adjust self parameters according to actual circumstances, and experience knowledge about signals and interferences is not necessary when adaptive filter is designed. So adaptive filter is flexible and convenient, and it is applied in lots of technology fields. Adaptive algorithm is the core of adaptive filter, which determines its performance. LMS algorithm is adopted in engineering frequently due to some advantages such as simple form, small amounts of computation and short operation time. Transverse filtering and IQ filtering are two main approaches to actualize the LMS algorithm. It is of reference significance to research the performance differences on stability, convergence rate, and error between two approaches. Therefore it simulates the processes in which transverse filtering and IQ filtering are implemented to realize adaptive LMS algorithm, and compares their filtering performances. The simulating results reveal that IQ filtering has higher stability, faster convergence rate and smaller error. Its filtering performance is superior to that of transverse filtering. The rank of IQ filter is two, at the same time the rank of transverse filter generally exceeds ten. Weight noises will enlarge when the rank of filter increases, and this is adverse to improve the filtering performance such as stability, convergence rate, and error ratio. IQ filtering will be adopted first in practice, since it is of less rank.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2007年第5期148-151,共4页
High Voltage Engineering
基金
国防科技重点实验室基金(9140C210601070C2101)。~~