Active noise control method for eliminating and reducing noise

Abstract

The invention discloses an active noise control method for eliminating and reducing noise. An active noise control device is subjected to the operation in the following steps of: (1) identifying a transfer function between a secondary source and an error sensor by using a common LMS (Least Mean Square) algorithm to obtain a transfer function estimating value of a signal transmitting channel; (2) obtaining an output signal y1(n) of an FXLMS (Filtered-x Least Mean Square) algorithm self-adapting filter W(n) according to a primary noise source signal x(n); (3) obtaining an output signal y2(n) of a DFT-FSF (Discrete Forurier Transform-Fast Forurier Transform) algorithm self-adapting filter according to a filtering reference signal of the signal transmitting channel; (4) adding y1(n) and y2(n) to obtain y(n) and respectively obtaining an expected signal d(n) and an error signal e(n) at the error sensor position and an estimating value of d(n); and (5) updating the DFT-FSF algorithm self-adapting filter V(n) and adjusting the FXLMS self-adapting filter W(n); and repeating the steps (2) to (5) and gradually reducing e(n) to achieve the aim of noise reduction at the error sensor position. The invention has the advantage of combining and utilizing the favorable benefits that an FXLMS algorithm is suitable for narrowband noise reduction and a DFT-FSF algorithm is suitable for broadband noise reduction and greatly reduces computation burden.

Description

technical field The present invention relates to a noise reduction method, in particular to an active noise control method for noise reduction. Background technique The existing active noise control uses the superposition principle of sound waves to generate a group of sound waves with the same amplitude and opposite phase as the primary noise through the secondary sound source, and the two are added at the error sensor to achieve the effect of noise reduction. For decades, people have proposed many noise control algorithms, the most famous of which is the filter-X LMS algorithm (FXLMS) proposed by Widrow. In view of the slow convergence speed of the FXLMS algorithm, which is only effective for narrowband noise and poor for broadband noise, people have proposed algorithms based on subband adaptive filters and frequency domain FXLMS based on the broadband characteristics of the noise spectrum, but they More time delay is introduced on the primary reference noise path, so th...

AI Technical Summary

Problems solved by technology

However, this method also has some shortcomings: if the noise signal is a single-frequency signal or multiple single-frequency signals, the convergence effect will be unstable; this method requires two discrete Fourier transform DFT operations and one inverse DFT operation for each sampling point , although DFT can be completed with FFT, the total amount of calculation is also very considerable, which puts too high demands on the processor

Images