期刊
MEASUREMENT
卷 45, 期 6, 页码 1350-1361出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.measurement.2012.03.004
关键词
Multiple-notch filters; Digital IIR filters; Powerline interference; Transient behavior; Parameter-varying systems; ECG signal
资金
- Ministry of Science and Higher Education of the Republic of Poland [N N505 484740]
In signal processing it is often required to have multiple-notch filters which simultaneously possess a very selective magnitude response and a transient response of short duration. Multiple-notch filters are usually used to remove the powerline hum and its harmonics. It is known, that sharper notches are obtained by either using more coefficients in FIR filters, or by placing the pole closer to the unit circle in IIR filters. However, increasing the filter selectivity also increases the duration of the transient process in the filter after the action of the excitation. This paper presents and compares two methods for suppressing the transient response in multiple-notch filters. As an example, these methods have been used to eliminate the powerline noise from the ECG signal. Both filtering techniques are characterized by improved transient behavior compared to the traditional multiple-notch filters. In the first part of the paper a concept of digital IIR multiple-notch filters whose pole radius changes with time is proposed. Owing to a temporary change in the value of the pole radius, the transient can be considerably reduced. In the second part of this review the technique which uses the vector projection to find better initial values for time-invariant multiple-notch filter is presented. Simulations verifying the effectiveness of the proposed IIR multiple-notch filter with time-varying pole radius and the filter with non-zero initial conditions are presented and compared to the performance of the traditional time-invariant filter using an ECG signal with unwanted fixed-frequency harmonic interferences as a study case. (C) 2012 Elsevier Ltd. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据