CN-122024688-A - Multi-channel self-adaptive active wind noise reduction automobile seat and noise reduction method thereof

CN122024688ACN 122024688 ACN122024688 ACN 122024688ACN-122024688-A

Abstract

The application discloses a multichannel self-adaptive active wind noise reduction automobile seat and a noise reduction method thereof, wherein the wind noise reduction automobile seat comprises a multichannel self-adaptive active wind noise reduction system, the multichannel self-adaptive active wind noise reduction system comprises a reference microphone, a loudspeaker set, an error microphone set and a multichannel controller, the reference microphone is arranged near a noise source and in a wind environment and is used for collecting wind noise signals as reference signals, the loudspeaker set is arranged in a headrest or near the headrest of the automobile seat, the error microphone set is arranged in the headrest or near the headrest of the automobile seat and is used for collecting residual noise signals of a noise reduction area as error signals, the input end of the multichannel controller is respectively connected with the reference microphone and the error microphone set, the output end of the multichannel controller is connected with the loudspeaker set, and the multichannel controller is used for inhibiting wind noise through a self-adaptive filter and a decoupling filter and can remarkably improve the active noise reduction performance of the wind noise environment in the automobile.

Inventors

YAN YUHANG
YAN SHIJUN
GONG ENJIE
Chu Diejing

Assignees

曲阜天博汽车电器有限公司

Dates

Publication Date: 20260512
Application Date: 20260213

Claims (9)

1. The multichannel self-adaptive active wind noise reduction automobile seat is characterized by comprising a multichannel self-adaptive active wind noise reduction system, wherein the multichannel self-adaptive active wind noise reduction system comprises a reference microphone, a loudspeaker set, an error microphone set and a multichannel controller, wherein the reference microphone is arranged nearby a noise source and in a wind environment and is used for collecting wind noise signals as reference signals, the loudspeaker set is arranged in a headrest of the automobile seat or nearby the headrest, the error microphone set is arranged in the headrest of the automobile seat or nearby the headrest and is used for collecting residual noise signals of a noise reduction area as error signals, the input end of the multichannel controller is connected with the reference microphone and the error microphone set respectively, the output end of the multichannel controller is connected with the loudspeaker set, and the multichannel controller suppresses wind noise through a self-adaptive filter and a decoupling filter.
2. The multi-channel adaptive active wind noise reduction car seat of claim 1, wherein the multi-channel controller comprises: The initialization module is used for initializing the internal parameters and system state variables of the multichannel controller and measuring the secondary channel response from each loudspeaker in the loudspeaker group to the corresponding error microphone; a decoupling filter for inversely modeling a secondary channel response from each of the speakers to a corresponding error microphone based on the IIR filter for generating a predicted inverse wind noise signal; the adaptive filter is based on a variable step length adaptive method of the FIR filter, and the coefficients of the FIR filter are dynamically adjusted according to the reference signal and the error signal; And the output module is used for inputting the output signal of the adaptive filter into the decoupling filter, obtaining a control signal and outputting the control signal to a corresponding loudspeaker so as to drive the loudspeaker to generate a secondary sound field for counteracting wind noise.
3. The multi-channel adaptive active wind noise reduction car seat of claim 2, wherein the coefficients of the multi-channel controller specifically comprise: The IIR filter has pole coefficients For said decoupling filtering obtained by calculation of the device The number of the channels is K, and each channel coefficient comprises an L-order modeling coefficient; The FIR filter has K channel coefficients calculated by the adaptive filter, each channel coefficient comprising a J-order zero coefficient.
4. A multi-channel adaptive active wind noise reduction car seat as defined in claim 3, wherein the reference microphone is provided at a window frame outside a car window, or at a junction of a window glass and a car body, or in close proximity to an inner car surface of the window glass, for directly or indirectly collecting a wind noise signal transmitted into the interior of the car via a window structure as the reference signal.
5. The multi-channel adaptive active wind noise reduction car seat of claim 4, wherein the error microphone set and the speaker set are symmetrically arranged on the left and right sides of the headrest; the loudspeaker group is arranged in the structure with the two sides of the headrest bent forwards, and the sound radiation direction of the loudspeaker group faces to the left ear area and the right ear area of the passenger; the error microphone is arranged at the front side of the corresponding side loudspeaker and is closer to the ear of the passenger relative to the corresponding loudspeaker, and is used for collecting residual noise signals near the corresponding ear of the passenger.
6. A multi-channel adaptive active wind noise reduction method for wind noise reduction by a multi-channel adaptive active wind noise reduction car seat according to any one of claims 3 to 5, characterized in that the multi-channel adaptive active wind noise reduction method comprises the following steps: S10, the multichannel controller sends out an instruction for acquiring wind noise, the reference microphone picks up the wind noise and feeds the wind noise back to the multichannel controller, and the multichannel controller generates a predicted inverse wind noise signal through the adaptive filter; and S20, the multichannel controller sends the predicted inverse wind noise signals to the decoupling filter, then sends the generated filtered inverse wind noise signals to the loudspeaker group, and feeds back the real-time wind noise error signals acquired by the error microphones to the adaptive filter, and the inverse wind noise sent by the loudspeaker group reaches the error microphone group through a secondary channel to inhibit wind noise at the error microphones.
7. The multi-channel adaptive active wind noise reduction method according to claim 6, wherein the filtering method of the decoupling filter comprises: the secondary channel decoupling IIR filtering specifically comprises the following steps: Step 1, generating an inverse model estimation signal, and inputting an identification signal to the speaker group through the initialization module at the initial stage of system operation And collect responses using the error microphone set A secondary channel transfer function matrix from the speaker set to the error microphone set Wherein each element is Subscript of Indicating the number of the secondary speaker, Indicating the error microphone number; Step 2, inverse model estimation and iterative update, based on identification Calculate its approximate inverse model So that its component Will (i) be Modeling is carried out as an autoregressive model, and coefficients are updated in real time through a recursive least square method.
8. The method for adaptive active wind noise reduction in multiple channels of claim 7, wherein the inverse model estimation and iterative update in step 2 specifically comprises: at each sampling instant : Step 2.1, calculating the error of the system, the inverse filter output is calculated as Wherein For a modeling coefficient vector of order L, the inverse model estimation error signal is: ; step 2.2, coefficient of Updating according to a recursive least square method, wherein an updating formula is as follows: ; ; ; Wherein, the In order to drive the signal vector, Is that The equivalent inverse matrix of the moment driving signal vector autocorrelation matrix is obtained by updating the formula iteration, Is a unit matrix of corresponding dimension; Step 2.3 modeling coefficients obtained by steps 2.1 and 2.2 As secondary channel inverse model coefficients, the output signal of the adaptive filter is used as the input of the decoupling filter and passed through the secondary channel inverse model Filtering to obtain decoupled independent signals of all channels, and feeding back the signals to the corresponding loudspeakers.
9. The multi-channel adaptive active wind noise reduction method according to claim 7, wherein the filtering method of the adaptive filter comprises: Step 1, collecting residual noise signals through the error microphone group in each control period Wherein For the current moment of time, Initializing J-order coefficient vector of each channel adaptive filter for channel sequence number And setting an initial step length Step length variation parameter Minimum step size And maximum step size ; Step 2, the reference signal is processed The pass coefficient is Generates a secondary control signal The secondary control signal After being filtered by the decoupling filter, the sound is fed back to a power amplifier so as to drive the loudspeaker group to generate a secondary sound field for counteracting wind noise; step3, calculating the instantaneous power of the error signal Wherein Is a smoothing factor; dynamic adjustment of step size according to error power Wherein As a step-size forgetting factor, Is a gain coefficient; Clipping step size ; Step 4, using variable step length Updating adaptive filter coefficients Wherein In the form of a vector of the order J, Pre-filtering the reference signal for current and past periods of time A vector of the components; Step 5, updating the filter coefficient For filtering calculation at the next moment, repeating the steps 2 to 4 to enable the residual noise signal The mean square value of (c) is gradually reduced, and the system tends to be converged and stable.

Description

Multi-channel self-adaptive active wind noise reduction automobile seat and noise reduction method thereof Technical Field The invention relates to the technical field of automobile noise reduction, in particular to a multichannel self-adaptive active wind noise reduction automobile seat and a noise reduction method thereof. Background With the development of the automobile industry and intelligent cabin technology, noise control in the cabin has become a key topic for improving riding experience and comfort. The active noise control technology has effective suppression capability on low-frequency noise, and has important potential in-vehicle acoustic environment optimization. The multichannel active noise control system is able to create a larger mute area near the human ear by arranging multiple speakers and error microphones around the seat perimeter than with the single channel approach. However, multichannel active noise control for car seat scenes still faces a series of practical challenges. In a multichannel active noise control system, although the centralized control is excellent in overall noise reduction effect and system stability, the calculation amount is large, and the real-time requirement of a vehicle-mounted system is difficult to meet. In contrast, the distributed control architecture is more computationally efficient and easier to implement on embedded platforms. However, the secondary channel coupling phenomenon that is common in a distributed system affects not only the control accuracy, but also may reduce the overall performance due to local instability, particularly in cabin environments with limited space and complex reflection. Aiming at the secondary channel coupling problem, an active noise reduction system with an inverse filter structure is proposed to overcome the limitation （Bouchard M. and Feng Y., Inverse structure for active noise control and combined active noise control/sound reproduction systems, IEEE Transactions on Speech and Audio Processing, 2001）. of the traditional multi-channel active system in the aspect of crosstalk, however, the method often introduces longer delay time, is limited by the length of an inverse filter, has higher crosstalk removal residual error and limits the application of the active noise reduction system in the actual environment. On the other hand, in order to cope with the problem of sound source movement, a hybrid system combining an adaptive filter and a time domain convolution cyclic network has been proposed, a similar work of decoupling nonlinear components （Chen D., Cheng L., Yao D., et al, A secondary path decoupled active noise control algorithm based on deep learning, IEEE Signal Processing Letters, 2022）. in a secondary speaker through a neural network has also been proposed, and modeling is performed on secondary path inverse response by using a double-gating convolution cyclic network, so that a controller only needs to track main path change （Park J., Choi J.H., Kim Y., et al, HAD-ANC: A hybrid system comprising an adaptive filter and deep neural networks for active noise control, INTERSPEECH 2023, 2023）., which improves system adaptability, but mainly expands around a single channel at present, and the potential of a multichannel structure in terms of coverage and spatial balance is not fully exerted. Disclosure of Invention The embodiment of the application provides a multichannel self-adaptive active wind noise reduction automobile seat and a noise reduction method thereof, which can effectively reduce modeling errors and crosstalk interference, overcome the layout limitation of headrest speakers, track the change of a noise source in real time and remarkably improve the active noise reduction performance of a wind noise environment in a vehicle. An embodiment of the application provides a multi-channel adaptive active wind noise reduction automobile seat, which comprises a multi-channel adaptive active wind noise reduction system, wherein the multi-channel adaptive active wind noise reduction system comprises a reference microphone, a loudspeaker set, an error microphone set and a multi-channel controller, wherein the reference microphone is arranged near a noise source and in a wind environment and is used for collecting wind noise signals as reference signals, the loudspeaker set is arranged in a headrest of the automobile seat or near the headrest, the error microphone set is arranged in the headrest of the automobile seat or near the headrest and is used for collecting residual noise signals of a noise reduction area as error signals, the input end of the multi-channel controller is respectively connected with the reference microphone and the error microphone set, the output end of the multi-channel controller is connected with the loudspeaker set, and the multi-channel controller suppresses wind noise through an adaptive filter and a decoupling filter. In one possible implementation, the multi-channel controller inc