A robust active road noise control method based on coherence constraint

Abstract

The application discloses a kind of robust active road noise control methods based on coherence constraint.The steps include: (1) the hardware configuration of active road noise control system is carried out;(2) the multiple coherence coefficients of error signal and reference signal are calculated in real time, and it is used as the step constraint factor of control filter update;(3) the update of control filter is carried out by secondary path decomposition and reference signal whitening, combined with frequency domain filtering error minimum mean square algorithm;(4) the control filter is converted to time domain, and control signal is output in real time to drive cancellation loudspeaker to sound, and after propagating via secondary path, it is coherent superposition with road noise signal at error microphone, and quiet zone is generated near human ear.The application can be used in active road noise control system in multi-channel complex scene, and the stable protection method based on coherence constraint is obtained faster convergence speed and lower steady-state error in steady-state scene, and the robustness of system is improved in noisy scene.

Description

Technical Field

[0001] This invention belongs to the technical field of active noise control, specifically relating to a robust active road noise control method based on coherent constraints. Background Technology

[0002] Active Road Noise Control (ARNC) technology, through the combination of electronic components and multi-channel algorithms, can significantly reduce road noise inside the vehicle cabin and improve the passenger experience. The ARNC system requires an accelerometer installed in the vehicle chassis to collect road vibration information as a reference signal. This signal is then fed into a multi-channel digital signal processor (DSP), which filters the signal and outputs a control signal in real time to drive the cancellation speaker. The signal propagates through a secondary path and is coherently superimposed with road noise signals near the ears, achieving noise control. The DSP adaptively updates the control filter weights based on the reference signal and error signals picked up by the error microphone, using a multi-channel algorithm to optimize the system's noise reduction effect. The Decomposing and Whitening Frequency Domain Filtered-error Least Mean Square (DWFDFeLMS) algorithm, based on secondary path decomposition and reference signal whitening, has been proven to be a reliable technical solution for road noise control (Lian S, Li T, Gu J, et al. An online decoupling-whitening frequency domain filtered-error least meansquare algorithm for active road noise control[J]. The Journal of the Acoustical Society of America, 2024, 156(2): 1413-1424.). However, in practical applications, interference noise such as human voice communication and road bumps, after being picked up by the error microphone and fed into the algorithm update, will affect the convergence process of the control filter and may even cause the algorithm to diverge. Therefore, it is necessary to select a suitable protection algorithm to improve the robustness of the system.

[0003] Using a state transition strategy in noisy scenarios, stopping filter updates when interference noise is detected can avoid system divergence. However, the effectiveness of this method is highly dependent on the selection of state transition parameters (Shen X, Gan WS, Shi D. Alternative switching hybrid ANC[J]. Applied Acoustics, 2021, 173: 107712.). Using a weight constraint algorithm to limit the maximum value of the control filter update for impulse noise improves system robustness but affects the steady-state performance of the algorithm (Lan H, Zhang M, Ser W. A weight-constrained FxLMS algorithm for feedforward active noise control systems[J]. IEEE Signal Processing Letters, 2002, 9(1): 1-4.). Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a robust active road noise control method based on coherent constraints. This method, based on a stability protection method with coherent constraints, can achieve faster convergence speed and lower steady-state error in steady-state scenarios, and exhibits stronger robustness in noisy scenarios.

[0005] The technical solution adopted in this invention is as follows:

[0006] A robust active road noise control method based on coherence constraints, comprising the following steps:

[0007] Step 1: Configure the hardware for the active road noise control system, including placing an acceleration sensor on the vehicle chassis to collect road vibration information as a reference signal; and installing an active noise-canceling headrest with an error microphone and a cancelling speaker at the front passenger's head, wherein the error microphone is used to collect error signals at both ears.

[0008] Step 2: Calculate the multicoherence coefficients of the error signal and the reference signal in real time, and use the multicoherence coefficients as the step size constraint factor for controlling the filter update; control the large coherence coefficients of the error signal and the reference signal in the early stage, use a large step size to ensure fast convergence; control the small coherence coefficients in the later stage, use a small step size to reduce the steady-state error of the system; in noisy scenarios, the introduction of interference noise reduces the signal coherence, use a small step size to improve the robustness of the system.

[0009] Step 3: Update the control filter by combining secondary path decomposition and reference signal whitening with the minimum mean square algorithm for frequency domain filtering error.

[0010] Step 4: Convert the control filter to the time domain and output the control signal in real time to drive the cancellation speaker to emit sound. After propagating through the secondary path, the sound is coherently superimposed with the road noise signal at the error microphone, creating a quiet zone near the human ear and achieving the noise reduction function.

[0011] This invention predicts the maximum noise reduction of the system by calculating the multicoherence coefficients of the error signal and the reference signal in real time, and uses this as a step size constraint factor for updating the control filter. In the early stage of control, the coherence coefficient is relatively large, and a large step size is used to ensure rapid algorithm convergence; in the later stage of control, the coherence coefficient is relatively small, and a small step size is used to reduce the system's steady-state error. When interference noise exists in the system, the coherence of the error signal and the reference signal deteriorates, and the smaller step size constrains the weight update of the control filter, thereby improving the robustness of the ARNC system. Therefore, compared with existing technologies, this invention, in a road noise real-time control system, uses a stability protection method based on coherence constraints. In steady-state scenarios, it achieves faster convergence speed and lower steady-state error, while in noisy scenarios, it improves the system's robustness, demonstrating high practical value. Attached Figure Description

[0012] Figure 1 Here is an overall schematic block diagram of the method of the present invention. (a) is a system configuration diagram of a multi-channel ARNC system, and (b) is a flowchart of the multi-channel frequency domain FeLMS (Coherence-based DWFDFeLMS, C-DWFDFeLMS) algorithm based on secondary path decomposition and reference signal whitening after adding coherent constraint stability protection.

[0013] Figure 2 This is an update block diagram of the media filter based on secondary path decomposition and reference signal whitening.

[0014] Figure 3 Here are the hardware configuration diagrams in this embodiment: (a) is a layout diagram of the vehicle chassis acceleration sensor, and (b) is an installation diagram of the active noise-canceling headrest.

[0015] Figure 4 These are the noise reduction curves at the error microphone in the steady-state scenario of this embodiment. (a) is the noise reduction curve at the left ear microphone, and (b) is the noise reduction curve at the right ear microphone.

[0016] Figure 5 This is the step size variation curve of the C-DWFDFeLMS algorithm in the steady-state scenario of this embodiment.

[0017] Figure 6 These are the noise reduction curves at the error microphone in a noisy scene in this embodiment. (a) is the noise reduction curve at the left ear microphone, and (b) is the noise reduction curve at the right ear microphone.

[0018] Figure 7This is the step size variation curve of the C-DWFDFeLMS algorithm in a noisy scene in this embodiment. Detailed Implementation

[0019] This embodiment provides a robust active road noise control method based on coherence constraints, the main steps of which include the following parts:

[0020] 1. Hardware configuration of the ARNC system

[0021] 1) Place acceleration sensors on the vehicle chassis

[0022] In a feedforward ARNC system, accelerometers can be placed on the vehicle subframe, steering shaft, shock absorbers, tow arms, etc., to pick up road vibration information and obtain reference signals that are highly correlated with the road noise signals at the driver's ears (Zhang JA, Murata N, Maeno Y, et al. Coherence-based performance analysis on noise reduction in multichannel active noise control systems[J].The Journal of the Acoustical Society of America,2020,148(3):1519-1528.), which are used to update the control filter weights in the multichannel control algorithm.

[0023] 2) Active noise-canceling headrests are installed in the vehicle's seating area.

[0024] Installing active noise-canceling headrests at the passenger's head, using cancelling speakers for road noise control, and employing error microphones to pick up error signals for filter updates is a common approach in ARNC (Action Control Noise Control) systems (Jung W, Elliott SJ, Cheer J. Local active control of road noise inside a vehicle[J]. Mechanical Systems and Signal Processing, 2019, 121: 144-157.). On the one hand, a relatively flat frequency response should be ensured in the secondary path between the speaker and the microphone; on the other hand, the headrest should be as comfortable and aesthetically pleasing as possible to provide passengers with a better riding experience.

[0025] 3) Connect to a multi-channel DSP

[0026] The reference signal acquired by the accelerometer is fed into a multi-channel DSP. After being filtered by a control filter, a real-time control signal is output to drive the cancellation speaker to emit sound. The sound propagates through a secondary path to the binaural area and coherently superimposes with the road noise signal. The error signal, after being controlled, is picked up by an error microphone and sent to the DSP. The DSP calculates the multicoherence coefficients of the error signal and the reference signal to obtain the step size constraint factor for filter updates, and then updates the control filter weights using the C-DWFDFeLMS algorithm. During the convergence process, the noise in the binaural noise reduction area gradually decreases, thereby achieving the road noise control function.

[0027] 2. Derivation of the DWFDFeLMS Algorithm

[0028] Example: A multi-channel ARNC system containing R reference signals, C control signals, and M error signals, such as... Figure 1 As shown in (a). Define 2L discrete frequency points k = 0, 1, ..., 2L-1, to obtain the autospectral density matrix of the reference signal, and the cross-spectral density matrices of the path noise signal and the reference signal, respectively.

[0029]

[0030] In equation (1), To express the mathematical expectation, superscript is used. H Representing the conjugate transpose, x(k) and d(k) are the Fast Fourier Transforms (FFTs) of the reference signal and the path noise signal at 2L discrete frequency points, respectively.

[0031] 1) Secondary path decomposition and reference signal whitening

[0032] Define S(k) as the frequency domain response of the secondary path, and obtain the all-pass component S using internal and external integrals. all (k) and the minimum phase part S min (k)(Wesselink JM, Berkhoff AP. Fast affine projections and theregularized modified filtered-error algorithm in multichannel active noisecontrol[J]. The Journal of the Acoustical Society of America, 2008, 124(2):949-960.)

[0033] S(k)=S all (k)S min (k) (2)

[0034] The error signal is filtered through the full-pass filter to obtain the filtered error signal.

[0035]

[0036] In the formula, e(k) is the FFT of the error signal at 2L discrete frequency points.

[0037] The autospectral density matrix P of the reference signal xx (z) Perform spectral factor decomposition

[0038]

[0039] Wherein, the spectral density function F min (k) can be viewed as a transfer function containing the characteristics of the reference signal (Elliott SJ. Optimal controllers and adaptive controllers for multichannel feedforward control of stochastic disturbances[J]. IEEE Transactions on signal Processing, 2000, 48(4):1053-1060.). Therefore, F can be used min The inverse matrix of (k) performs whitening by preprocessing the reference signal, resulting in a set of uncorrelated whitened reference signals:

[0040]

[0041] Among them, superscript -1 This indicates the inverse operation.

[0042] 2) Implementation of the multi-channel frequency domain FeLMS algorithm

[0043] The paper (Lian S, Li T, Gu J, et al. An online decoupling-whitening frequencydomain filtered-error least mean square algorithm for active road noise control[J]. The Journal of the Acoustical Society of America, 2024, 156(2): 1413-1424.) points out that the optimal solution of the frequency domain controlled filter is

[0044]

[0045] In equation (6), {·}+ To extract the causal part within the parentheses, the control filter can be calculated as follows by defining the intermediate filter Ψ(k).

[0046]

[0047] Where, the optimal solution of Ψ(k) is Based on Newton's algorithm (Widrow B, Stearns SD. Adaptive signal processing prentice-hall[J]. Englewood Cliffs, NJ, 1985: 52.), the update equation for Ψ(k) can be written as:

[0048] Ψ new (k)=(1-β)Ψ old (k)+βΨ opt (k) (8)

[0049] In the above formula, β is the update step size of the medium filter.

[0050] For non-steady-state road noise signals, this invention uses a block update strategy and the mean of T frames of data to obtain a more accurate spectral density estimate. In this case, equation (8) can be rewritten as a P-iteration approximation of the optimal solution:

[0051]

[0052] In the above formula, p is the current iteration number. Let v be the filtering error signal of the t-th frame. t (k) is the whitening reference signal for frame t. The update block diagram of the media filter is as follows: Figure 2 As shown. The residual signal passes through The filter is applied to obtain the filtered residual signal; the reference signal is then filtered... The whitened reference signal is obtained by filtering. The conjugate transpose of the whitened reference signal is multiplied with the filtered residual signal in the frequency domain. After processing T-frame data, causal constraints are added, and the medium filter is updated. When p = 0, Ψ 0 (k)=Ψ old (k), When p>0 It is the filtered residual signal of the data in frame t during the p-th iteration:

[0053]

[0054] After completing P iterations, we have Ψ P+1 (k)=Ψ new (k). At this point, according to Ψ new (k) The update formula for the time-domain controlled filter is obtained as follows:

[0055]

[0056] Wherein, IFFT{·} + This represents the causal component of the Inverse Fast Fourier Transform (IFFT). During the filter update process, the multi-channel digital signal processor outputs control signals in real time to drive the cancelling speaker to emit sound, gradually reducing noise in the binaural noise reduction area, thereby achieving the road noise control function.

[0057] 3. Calculation of multicoherence coefficient and step size constraint factor

[0058] Without considering causality, the maximum noise reduction of the ARNC system can be predicted using the multicoherence coefficients of the road noise signal and the reference signal:

[0059]

[0060] Let k represent the multicoherence coefficients of the m-th path noise signal and the reference signal, and satisfy the following for all frequency points k. P xdm P is the cross-spectral density vector of the m-th path noise signal and the reference signal. xx (k) is the spectral density matrix of the reference signal, P dmdm This is the spectral density of the m-th road noise signal. However, the error microphone cannot directly acquire the road noise signal during the control process, therefore the multicoherence coefficients of the error signal and the reference signal can be calculated in real time.

[0061]

[0062] In equation (13), P xem P is the cross-spectral density vector of the m-th error signal and the reference signal. emem It is the spectral density of the m-th error signal. In the initial stage of control, e m (k)=d m (k), The value of is at its maximum. As the control process gradually converges, the coherence between the error signal and the reference signal gradually weakens. The value gradually decreases.

[0063] In a practical ARNC system, the signal picked up by the error microphone includes not only road noise related to the reference signal, but also in-vehicle interference noise a(n) unrelated to passenger voice. In this case, we define d'(n) = d(n) + a(n) for the noisy scenario, and the optimal solution for controlling the filter in equation (6) can be rewritten as...

[0064]

[0065] In equation (14) It can be represented as

[0066]

[0067] Where a(k) is the FFT of 2L discrete frequency points of the interference noise. Substituting equation (15) into equation (14) and comparing it with equation (6), it is found that the introduction of incoherent noise does not affect the optimal solution of the algorithm. However, when incoherent noise is fed into the algorithm iteration, it will cause the filter weights to deviate from the previous converged solution, which not only reduces the convergence speed of the algorithm, but may also cause the algorithm to diverge.

[0068] Define the multicoherence coefficient in a noisy scene as:

[0069]

[0070] Where e′ m (n) is the m-th noisy error signal. Considering that a(n) and x(n) are uncorrelated, the noisy error vector e′(n) can be written as...

[0071] e′(n)=e(n)+a(n) (17)

[0072] At this point, following the approximation method of equation (15), equation (16) can be written as:

[0073]

[0074] When the system is far from the optimal solution, the controllable part of the error signal (i.e., the part related to the reference signal) is larger. The value is relatively large; however, when the system approaches convergence... The value then decreases. Although the introduction of interference noise will increase the noisy error signal, it is uncorrelated with the reference signal and can still guarantee... The value is at a low level.

[0075] In a multi-channel system, the step size constraint factor for controlling the filter update is obtained by calculating the mean of M multicoherence coefficients.

[0076]

[0077] At this point, the update formula for the media filter based on coherence constraints is:

[0078]

[0079] The step size μ(k) is corrected by the step size constraint factor to

[0080]

[0081] The update block diagram of the C-DWFDFeLMS algorithm after adding stability protection with coherence constraints in this invention is as follows: Figure 1 As shown in (b). The upper dashed box represents the time-domain processing part, including real-time filtering of the control filter and point-by-point output control signals of the DSP; the lower dotted box represents the frequency-domain processing part, including FFT of the signal, calculation of coherence coefficients, and updating of the medium filter and control filter. First, the two reference signal data blocks in the time domain are connected in parallel, and an L-point all-zero data block is inserted before the error signal data block. The 2L-point reference data and error data are then converted to the frequency domain by FFT. Then, the multiple coherence coefficients are calculated according to equation (16), and substituted into equation (20) to update the medium filter. After the medium filter is updated, the weights of the time-domain control filter are calculated by substituting into equation (11), thus forming a time-frequency domain hybrid control structure without time delay. In the early stage of control, the coherence coefficients of the error signal and the reference signal are relatively large, and a large step size is used to ensure fast convergence of the algorithm; in the later stage of control, the coherence coefficients of the error signal and the reference signal are relatively small, and a small step size is used to reduce the steady-state error of the system. When interference noise exists in the system, the coherence between the noisy error signal and the reference signal deteriorates. The smaller step size constrains the weight update of the control filter, thereby improving the robustness of the ARNC system.

[0082] Example

[0083] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention. After reading the present invention, any modifications of the present invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0084] 1. Hardware configuration and algorithm parameter selection for the ARNC system

[0085] The vehicle used in this experiment is a Changan UNI-K. A total of four 3-axis accelerometers were deployed on the vehicle chassis to collect 12 reference signals. The placement of the accelerometers is as follows: Figure 3 As shown in (a); the active noise-canceling headrest consists of a left ear speaker and a right ear speaker, and a left ear microphone and a right ear microphone that simulate the noise reduction point of the human ear, as shown in (a). Figure 3As shown in (b), a Texas Instruments TMS320C6678 digital signal processor was used for noise control in the area to be denoised within the vehicle cabin. The sampling frequency was 2500Hz, the length of the control filter and the secondary path were both 256 orders, and the number of data frames T and the number of iterations P were both set to 10. Based on the 50% overlap preservation method, the size of the FFT was set to 512, and the A-weighted (dBA) noise reduction in the 20-1250Hz range was used as an evaluation index for algorithm performance. The performance of the existing DWFDFeLMS algorithm without coherent constraint stabilization protection and the C-DWFDFeLMS algorithm of this invention with coherent constraint stabilization protection were tested respectively, and the results were compared with the Wiener solution under the current driving conditions.

[0086] 2. Real-world road noise control performance test under steady-state conditions

[0087] Under steady-state conditions at 50 km / h, the update step size β = 0.3 was set for both the DWFDFeLMS and C-DWFDFeLMS algorithms. 30s of data were recorded at the error microphone with ARNC off and on, respectively, as the road noise signal before control (Primary noise) and the error signal after control. The noise reduction curves for different algorithms were statistically analyzed as follows: Figure 4 As shown in (a) and (b), the results show that, due to the use of the coherent constraint method, the method of this invention can quickly converge to the Wiener solution in about 11 seconds, with a noise reduction of 5.3 dBA for the left ear and 5.0 dBA for the right ear; while the DWFDFeLMS algorithm, due to its excessively large step size, cannot converge to the Wiener solution in the later stages. The step size variation curve of the C-DWFDFeLMS algorithm is shown in Figure [Figure number missing]. Figure 5 As shown, the step size gradually decreases as the system converges. The coherent constraint method ensures that a large step size is used in the early stages to accelerate the convergence speed, and a small step size is used in the later stages to reduce the steady-state error of the system. Furthermore, from... Figure 4 As can be seen from the road surface bumps indicated by the middle arrow, the C-DWFDFeLMS algorithm has better stability than the DWFDFeLMS algorithm.

[0088] 3. Real-world road noise control performance test in noisy scenarios

[0089] To construct a realistic noisy driving scenario, a 10-second voice signal was added after 20 seconds of driving at 50 km / h to evaluate the stability of different algorithms. The update step size β was set to 0.2 for both the DWFDFeLMS and C-DWFDFeLMS algorithms. 50 seconds of data were recorded at the error microphone with ARNC off and on, respectively, as the road noise signal before control and the error signal after control. The noise reduction curves of different algorithms were statistically analyzed as follows: Figure 6As shown in (a) and (b), the control performance of the DWFDFeLMS algorithm deteriorates after noise removal, while the C-DWFDFeLMS algorithm can quickly recover to a better control level and has better stability. The step size variation curve of the C-DWFDFeLMS algorithm is shown in Figure (b). Figure 7 As shown. The introduction of incoherent noise degrades the coherence between the noisy error signal and the reference signal. The C-DWFDFeLMS algorithm uses a small step size to limit the impact of noise on the update of the control filter. When the noise is removed, the coherence is restored, and the C-DWFDFeLMS algorithm can quickly converge to the Wiener solution by using a larger step size.

Claims

1. A robust active path noise control method based on coherence constraints, characterized in that, The method includes the following steps: Step 1: Configure the hardware for the active road noise control system, including placing an acceleration sensor on the vehicle chassis to collect road vibration information as a reference signal; and installing an active noise-canceling headrest with an error microphone and a cancelling speaker at the front passenger's head, wherein the error microphone is used to collect error signals at both ears. Step 2: Calculate the multicoherence coefficients of the error signal and the reference signal in real time, and use the multicoherence coefficients as the step size constraint factor for controlling the filter update; control the large coherence coefficients of the error signal and the reference signal in the early stage, use a large step size to ensure fast convergence; control the small coherence coefficients in the later stage, use a small step size to reduce the steady-state error of the system; in noisy scenarios, the introduction of interference noise reduces the signal coherence, use a small step size to improve the robustness of the system. Step 3: Update the control filter by combining secondary path decomposition and reference signal whitening with the minimum mean square algorithm for frequency domain filtering error. Step 4: Convert the control filter to the time domain and output the control signal in real time to drive the cancellation speaker to emit sound. After propagating through the secondary path, the sound is coherently superimposed with the road noise signal at the error microphone, creating a quiet zone near the human ear and achieving the noise reduction function.

2. The robust active road noise control method based on coherence constraints as described in claim 1, characterized in that, In step 2, the multicoherence coefficients of the error signal and the reference signal are calculated in real time, specifically including: Define 2L discrete frequency points k = 0, 1, ..., 2L-1, and the formula for calculating the multicoherence coefficient is as follows: P is the cross-spectral density vector of the m-th error signal and the reference signal. xx (k) is the spectral density matrix of the reference signal. It is the spectral density of the m-th error signal; superscript -1 and H These represent the inverse operation and the conjugate transpose operation, respectively.

3. The robust active road noise control method based on coherence constraints as described in claim 2, characterized in that, In step 2, the multicoherence coefficients are used as the step size constraint factor for controlling the filter update, specifically including: The step size constraint factor for controlling filter updates is obtained by calculating the mean of the M multicoherence coefficients. At this point, the update step size of the control filter can be written as: Where β is a constant.

4. The robust active path noise control method based on coherence constraints as described in claim 1, characterized in that, Step 3, through secondary path decomposition and reference signal whitening, specifically includes: Decompose the secondary path into a fully connected component S. all (k) and the minimum phase part S min (k), and the error signal is filtered using the full-pass portion to obtain the filtered error signal. Where e(k) is the Fast Fourier Transform of the error signal at 2L discrete frequency points; The spectral density matrix P of the reference signal xx (k) Perform spectral factorization to obtain the spectral density function F containing the characteristics of the reference signal. min (k), and use F min The whitened reference signal is obtained by filtering the reference signal using the inverse matrix of (k). Where x(k) is the Fast Fourier Transform of 2L discrete frequency points of the reference signal.

5. The robust active road noise control method based on coherence constraints as described in claim 4, characterized in that, In step 3, the control filter is updated using the minimum mean square algorithm for frequency domain filtering error, specifically including: Combining the minimum mean square algorithm for frequency domain filtering error, the update formula for the frequency domain medium filter is obtained as follows: In the above formula, p is the number of iterations, and T is the number of data frames used in each update. Let v be the filtering error signal of the t-th frame. t (k) is the whitening reference signal for the t-th frame; when p = 0, Ψ 0 (k)=Ψ old (k), When p>0 It is the filtered residual signal of the data in frame t during the p-th iteration: After completing P iterations, we have Ψ P+1 (k)=Ψ new (k); at this time according to Ψ new (k) The update formula for the time-domain controlled filter is obtained as follows: Wherein, IFFT{·} + This represents the causal component of the inverse fast Fourier transform.

Images