逆ノッチフィルタと線形予測器を用いた適応ノッチゲインの実現

Abstract

This paper proposes an adaptive notch filter which automatically adjusts the notch frequency and the notch gain, where they are the null frequency and the depth of the null, respectively. The proposed notch filter consists of an inverse notch filter and a linear prediction filter, and gives an appropriate filter gain which can eliminate a sinusoidal noise and extract only a wide-band signal even if they share the same frequency. We compare the capability of the proposed notch filter with traditional methods. Simulation results show that the proposed method improves the SNR in comparison to the conventional methods.