Active noise control method based on reference signal weighting

By using an algorithm based on reference signal weighting in the active noise control system, the spacing of the reference signal is increased and the adaptive filter weight is updated, which solves the problem of slow convergence speed and low steady-state noise reduction in the active noise control system. , achieving more efficient noise control effect.

CN119028305APending Publication Date: 2024-11-26TIANJIN UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202310594845.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-25
Publication Date
2024-11-26

Smart Images

  • Figure CN119028305A_ABST
    Figure CN119028305A_ABST
Patent Text Reader

Abstract

The invention relates to an active noise control method based on reference signal weighting, and the method comprises the following steps: 1, obtaining a reference signal x (n) and an error signal e (n) needed by a system, x (n) = [x (n), x (n-1),..., x (n-L + 1)] T, and n is a positive real number and represents a time sequence; the error signal e (n) is a signal obtained by offsetting a desired signal d (n) of the area to be muffled and a secondary sound source signal; a second step of generating a secondary sound source signal through the collected signal, and performing weight updating on an adaptive filter which is composed of four parts, namely a secondary sound source signal generation module, a reference signal secondary path filtering module, a reference signal weighting module and an adaptive filter weight updating module; and step 3, evaluating and adjusting the active noise control method based on reference signal weighting by adopting the average noise reduction amount.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of active noise control, and in particular relates to a noise reduction method of an active noise control system based on reference signal weighting. Background Art

[0002] Noise pollution can be seen everywhere in social life. Elevators and air conditioners in living environments are common noise source signals; engine noise and intake and exhaust noise generated inside the car during driving are also inevitable noise pollution in the transportation process; in addition, industrial noise pollution in scenes such as construction sites and construction sites is also uncomfortable. According to relevant data, when people work and live in a noise environment with high sound pressure levels for a long time, they will not only experience symptoms such as deafness and tinnitus, but may also induce high blood pressure and nervous system diseases. Therefore, considering the prevalence and severity of noise pollution, effective noise control is a necessary measure to solve noise pollution.

[0003] The methods of noise control can be divided into passive noise control (Passive Noise Control, PNC) and active noise control (Active Noise Control, ANC) according to the principles adopted. The principle of passive noise control is mainly to deal with noise through the interaction between noise sound waves and acoustic sound insulation materials or structural designs. The specific manifestations in normal life are adding sound-absorbing materials or silencers to vehicles and closing doors and windows. This method mainly targets the noise source or the propagation path of the noise. However, due to the wavelength characteristics of low-frequency noise, it can penetrate the sound-absorbing materials or silencers in passive noise control, resulting in passive noise control generally only having a good noise reduction effect on medium and high-frequency noise, and has limited control over low-frequency noise.

[0004] Compared with passive noise control systems, active noise control systems have higher flexibility and better noise reduction effects. Since the 1970s, many scholars at home and abroad have devoted themselves to studying a large number of active noise control systems and algorithms. Among them, the Filtered-X Least Mean Square (FxLMS) algorithm is a classic algorithm commonly used in active noise control systems. The Filtered-x Least Mean Square / Fourth (FxLMS / F) algorithm proposed on the basis of the FxLMS algorithm improves the tracking performance of the active noise control system in dealing with noise. However, the algorithm still has the problems of slow convergence speed and low steady-state noise reduction. In view of the above problems, the present invention proposes an active noise control method based on reference signal weighting, and applies it to the improved convex combination active noise control structure to further improve the noise reduction performance and convergence performance of the ANC system. Summary of the invention

[0005] The present invention aims to solve the problems of slow convergence speed and low steady-state noise reduction of current active noise control, and proposes an active noise control algorithm based on reference signal weighting. The present invention can further improve the noise reduction performance and convergence performance of the ANC system. The technical solution of the present invention is as follows:

[0006] The active noise control method based on reference signal weighting comprises the following steps:

[0007] The first step is to obtain the reference signal x(n) and error signal e(n) required by the system, where x(n) = [x(n), x(n-1), ..., x(n-L+1)] T , n is a positive real number, representing a time series; the error signal e(n) is the signal after the expected signal d(n) in the area to be silenced is offset by the secondary sound source signal;

[0008] In the second step, the active noise control system generates a secondary sound source signal through the collected reference signal x(n) and error signal e(n), and updates the weights of the two adaptive filters under the convex combination structure as follows:

[0009] (1) Generate secondary sound source signal:

[0010] The collected reference signal is convolved with the weight coefficients of two adaptive filters under the convex combination structure, and a secondary sound source signal is generated through a secondary microphone to suppress impulse noise.

[0011] (2) Reference signal secondary path filtering:

[0012] To ensure the normal operation of the system, the reference signal x(n) is filtered using a secondary path modeling filter to obtain a corrected reference signal x′(n);

[0013] (3) Reference signal weighting:

[0014] According to the characteristics of the reference signal, the Spacing Increasing Discretization (SID) method is used to assign different weights to the reference signal. Under the convex combination structure, two different adaptive reference signals are weighted using different intervals to finally obtain the weighted reference signal x s (n);

[0015] (4) Adaptive filter weight update:

[0016] Under the convex combination structure, two filters are based on the reference signal x with different weights. s The weights of the filter are updated using the LMS / F algorithm. The two filters work in parallel to control the noise reduction speed and the amount of noise reduction of the noise reduction system respectively.

[0017] In the third step, the average noise reduction (ANR) is used to evaluate and adjust the active noise control method based on the weighted reference signal. The present invention designs an active noise control algorithm based on the weighted reference signal. The algorithm improves the noise reduction performance of the algorithm by weighting the reference signal using the spacing increase discretization in the logarithmic space, and applies it to the improved convex combination active noise control structure to further improve the convergence speed of the algorithm, thereby improving the noise reduction performance and convergence performance of the active noise control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The complete algorithm block diagram of the present invention

[0019] Figure 2 Magnitude frequency response plot of the acoustic path

[0020] Figure 3 Phase frequency response plot of the acoustic path

[0021] Figure 4 ANR curve under composite sinusoidal noise

[0022] Figure 5 ANR curve under broadband Gaussian noise

[0023] Figure 6 ANR curve under broadband pink noise

[0024] Figure 7 ANR curve chart under path change DETAILED DESCRIPTION

[0025] In the first step, the reference microphone collects the noise source signal x(n) and inputs it into the active noise control system, and the error microphone collects the error signal e(n) into the active noise control system.

[0026] The second step is to generate a secondary sound source signal and update the weights of the two adaptive filters under the convex combination structure. The signal generation process of the present invention is described in detail below;

[0027] (1) Generate secondary sound source signal

[0028] The reference signal x(n) is input into the control filter for convolution operation to obtain the secondary sound source signal. The output response of the two filters under the convex combination structure can be expressed as The output of the overall filter is y(n) = λ(n)y1′(n) + (1-λ(n))y2′(n), where w(n) = [w0(n), w1(n), ..., w L-1 (n)] T is the weight coefficient of the filter at time n. y1′(n), y2′(n) are the output responses of the filter after passing through the secondary path. λ(n) is the combination function of the two filters. This parameter controls the combination of the two filters in each iteration and is given by the sigmoidal function. Its update formula is a(n) is updated based on the gradient descent algorithm. In order to ensure the continuous update range of the algorithm, its value range is limited to [-4, 4]. The update equation is as follows

[0029]

[0030] (2) Reference signal secondary path filtering

[0031] The reference signal x(n) is filtered by the secondary path modeling filter to obtain the signal x′(n). The signals after secondary path modeling under the convex combination structure are

[0032] (3) Reference signal weighting

[0033] Different weights are assigned to the reference signal after secondary path identification by means of SID. The reference signal after weight change can be expressed as x s (n) = x′(n)⊙SI, where ⊙ represents the Hadamard product, and SI represents the weighted operation on the reference signal. The reference signal weight assignment formula is:

[0034]

[0035] Thus, the weighted reference signal x can be obtained s (n)=[t k-1 x′(n),t k-2 x′(n-1), ...t0x′(n-L+1)] T The weighted reference signal collected by the two control filters under the convex combination structure is expressed as x s1 (n)=x′1(n)⊙SI1,x s2 (n)=x′2(n)⊙SI2.

[0036] (4) Adaptive filter weight update

[0037] The control filter updates the weights through the error signal and the weighted reference signal. The weight update formula of the filter under the convex combination structure is:

[0038] In the formula is a threshold parameter, μ represents the step size factor in the algorithm, and the weight update formula of the overall filter is w(n)=λ(n)w1(n)+(1-λ(n))w2(n).

[0039] The third step is to evaluate and adjust the algorithm performance.

[0040] The average noise reduction (ANR) is used to evaluate the performance of the algorithm. The definition of ANR is:

[0041]

[0042] Among them A e (n) = κA e (n)+(1-κ)|e(n)|,A d (n) = κA d (n)+(1-κ)|d(n)|. In the initial state, A e (0) = A d (0)=0, κ is the forgetting factor, and its value is set to 0.999.

[0043] The comparison between the present invention and the existing active noise control method under different noise conditions is shown in FIG. Figure 4 , 5 As shown in , 6 and 7, the active noise control method proposed in the present invention has better noise reduction performance.

Claims

1. An active noise control method based on reference signal weighting comprises the following steps: The first step is to obtain the reference signal x(n) and error signal e(n) required by the system, where x(n) = [x(n), x(n-1), ..., x(n-L+1)] T , n is a positive real number, representing a time series; the error signal e(n) is the signal after the expected signal d(n) in the area to be silenced is offset by the secondary sound source signal; In the second step, the active noise control system generates a secondary sound source signal through the collected reference signal x(n) and error signal e(n), and updates the weights of the two adaptive filters under the convex combination structure as follows: (1) Generate secondary sound source signal: The collected reference signal is convolved with the weight coefficients of two adaptive filters under the convex combination structure, and a secondary sound source signal is generated through a secondary microphone to suppress impulse noise. (2) Reference signal secondary path filtering: To ensure the normal operation of the system, the reference signal x(n) is filtered using a secondary path modeling filter to obtain a corrected reference signal x′(n); (3) Reference signal weighting: According to the characteristics of the reference signal, the Spacing Increasing Discretization (SID) method is used to assign different weights to the reference signal. Under the convex combination structure, two different adaptive reference signals are weighted using different intervals to finally obtain the weighted reference signal x s (n); (4) Adaptive filter weight update: Under the convex combination structure, two filters are based on the reference signal x with different weights. s The weights of the filter are updated using the LMS / F algorithm. The two filters work in parallel to control the noise reduction speed and the amount of noise reduction of the noise reduction system respectively. In the third step, the average noise reduction (ANR) is used to evaluate and adjust the active noise control method based on the reference signal weighting.

Citation Information

Cited By

  • Self-adaptive pruning multi-reference noise reduction method, device and system and storage medium

    CN116597806A

  • Adaptive pruning multi-reference noise reduction method, device, system and storage medium

    CN116597806B