A method and system for active control of road noise in a vehicle based on wideband spatial modal decoupling

Abstract

The application discloses a kind of based on wideband space modal decoupling in-car road noise active control method and system.The method is aimed at the problem of computing power explosion and space water bed effect faced by in-car multi-channel active noise reduction, discards the traditional frequency point by frequency point SVD decomposition in offline stage, constructs weighted wideband energy covariance matrix respectively at loudspeaker end and error microphone end and carries out global eigenvalue decomposition, extracts global constant time domain coding and decoding matrix with optimal full-band space energy distribution.In online adaptive control stage, the above coding and decoding matrix is used to decouple and reconstruct the multi-channel cross-coupling physical system into multiple parallel independent modal single-channel control loop, and adaptive filtering algorithm is executed in parallel in diagonalized modal space to update weight, and finally the modal control instruction is mapped back to physical space to drive loudspeaker to sound.The application reduces the adaptive filtering complexity from the order of magnitude of the number of physical channels to the linear order of magnitude of the number of modes, not only greatly reduces the computing overhead of vehicle-mounted processor, clears the front-mounted deployment obstacle of low-cost MCU chip, but also cuts off the mutual interference between channels from the physical mechanism, completely eliminates the space water bed effect, and improves the robustness of noise reduction.

Description

Technical Field

[0001] This invention relates to the field of active noise control (ANC) technology for vehicles, and more specifically, to a highly efficient in-vehicle road noise active control method and system based on broadband spatial mode decoupling. Background Technology

[0002] With the rapid global adoption of electric vehicles, the broadband masking noise from traditional internal combustion engines has disappeared, making low-frequency road noise in the 20–200 Hz range a major source affecting driving comfort. In this frequency band, traditional passive noise reduction technologies such as sound insulation and damping are limited by the "law of mass," making it difficult to achieve significant noise reduction while ensuring vehicle lightweighting. Therefore, onboard active road noise control (RANC) technology has become a core solution. Currently, multi-channel adaptive filtering algorithms (such as multi-channel MIMO-FxLMS algorithm) widely used in industry face a dual bottleneck in actual vehicle deployment: First, there are bottlenecks in computational complexity and hardware cost. When the control range covers the entire vehicle cabin, multiple sensors and secondary speakers must be deployed. Due to the cross-coupling between physical channels, the computational overhead of the filter increases explosively, on the order of the product of the number of channels (O(M × S × R × L)). This makes it extremely difficult to deploy existing road noise algorithms on a large scale on low-cost automotive-grade MCU single-core chips. Secondly, there is the spatial "waterbed effect" caused by physical coupling. The strong cross-coupling of channels inside the vehicle causes the secondary path matrix to exhibit ill-conditioned characteristics, which can easily lead to algorithm divergence. Traditional control methods based on zero-forcing discrete measurement points are prone to inducing a sharp amplification of noise in non-control areas or adjacent spaces, i.e., producing a severe spatial "waterbed effect." Passengers will feel noise attenuation and degradation even if they deviate slightly from the control point. While existing technologies have attempted spatial decoupling using analytical wave domain methods or frequency-by-frequency singular value decomposition (SVD), analytical methods rely on regular geometric assumptions that cannot adapt to complex vehicle cabins. Furthermore, traditional narrowband frequency-by-frequency SVD encounters severe phase ambiguity, non-causality, and unacceptable real-time iterative computing power bottlenecks in time-domain reconstruction when faced with continuous broadband road noise. Summary of the Invention

[0003] The purpose of this invention is to provide an active control method and system for in-vehicle road noise based on broadband spatial mode decoupling, so as to solve the problems of excessively high multi-channel computational complexity and spatial waterbed effect. According to a first aspect of the present invention, an active in-vehicle road noise control method based on broadband spatial mode decoupling is provided, comprising: S1: In offline mode, inject a broadband excitation signal into the secondary speaker to identify the secondary path impulse response matrix in the physical space; S2: Construct a weighted broadband energy covariance matrix on the speaker input side and the error microphone output side respectively, and perform eigenvalue decomposition to extract the dimension-reduced and realized global output mode basis as the encoding matrix and the global input mode basis as the decoding matrix; S3: In online control mode, the acquired residual error signal is projected onto the modal error space through the encoding matrix to obtain the modal residual error signal; S4: For each independent modal channel, execute the adaptive filtering algorithm in parallel and independently to update the modal weights and generate modal control commands; S5: The modal control command is reconstructed into a physical drive signal through the decoding matrix, which drives the secondary speaker to produce sound to cancel out road noise. According to a second aspect of the present invention, an active in-vehicle road noise control system based on broadband spatial mode decoupling is provided, comprising: Signal acquisition system: includes chassis reference sensors and in-vehicle error microphones; Controller: Embedded with an offline computing module and an online real-time control module, it executes the above-mentioned modal decoupling algorithm; Secondary speaker sound generation system: produces sound based on the decoded physical signal. Optionally, the present invention also provides an electronic device for performing the control method and a computer-readable storage medium. The in-vehicle road noise active control method and system disclosed herein have the following technical effects: Overcoming hardware computing power bottlenecks: This invention abandons the traditional frequency-point SVD decomposition and extracts a global constant encoding / decoding matrix through broadband energy aggregation, decoupling and reconstructing the originally massive physically cross-coupled system into K independent single-channel control loops. The single-step core computational load of the algorithm is drastically reduced from the physical channel product level to modal linear correlation, which can reduce the filtering computation load of the vehicle processor by more than 60%, clearing the computing power obstacle for the deployment of active road noise control on low-cost single-core automotive-grade MCU chips. Overcoming the spatial waterbed effect: This invention, through spatial orthogonal transformation, completely severs the antagonism and unnecessary electroacoustic loss between physical channels at the underlying mathematical mechanism, ensuring that each modal channel does not interfere with each other. This not only greatly relaxes the boundary of the algorithm's stable step size, but also successfully extends the local "discrete point zero-forcing" into a regional "silent basin," maintaining a stable and excellent noise reduction effect even if the occupant's head is off-center from the microphone area. Ensuring causality and physical realizability: By introducing frequency-domain weighted energy covariance and maximum energy phase alignment techniques, this invention losslessly projects complex space modes into a time-domain constant matrix, completely solving the time-domain non-causality and phase ambiguity problems in traditional methods, and possessing extremely strong physical vehicle deployability. Other features and advantages of the invention will become clear from the following detailed description of exemplary embodiments of the invention with reference to the accompanying drawings. Attached Figure Description

[0004] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the invention and, together with their description, serve to explain the principles of the invention. Figure 1 is a schematic diagram of the overall control topology architecture of the method provided in the embodiment of the present invention; Figure 2 is a schematic diagram of the system hardware space topology layout provided in an embodiment of the present invention; Figure 3 is a schematic diagram comparing the time-domain residual sound pressure noise reduction effect before and after applying the embodiments of the present invention. The following labels are shown in Figure 2: 100-Vehicle cabin; 101-Reference sensor; 102-Secondary speaker; 103-Error microphone; 104-Controller. Detailed Implementation

[0005] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification. In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. This invention proposes an embodiment of in-vehicle road noise active control (BMD-FxLMS) based on broadband spatial mode decoupling. Figure 1 shows the overall control topology architecture of the method provided by this embodiment, which is clearly divided into an offline mode recognition stage (within the upper wireframe) and an online adaptive control stage (within the lower wireframe). Figure 2 shows a schematic diagram of the system hardware spatial topology layout provided by this embodiment, wherein the reference sensor 101 is arranged in the chassis or wheel area, the secondary speaker 102 is arranged in the passenger compartment, the error microphone 103 is arranged in the occupant head area, and the controller 104 is centrally located. Referring to the above figures, the specific operating steps of this embodiment are as follows: During the offline phase, secondary path identification and broadband mode generation are primarily performed: Step S1: With the cabin stationary or in a calibration state, broadband excitation signals are sequentially injected into the control speaker, and the error microphone picks up the responses, thereby identifying all secondary path impulse response matrices G in the physical space. ω. Step S2: To enable the controller to handle the non-uniform energy distribution of broadband noise in the 20–200 Hz range, a frequency domain energy weighting function w(ω) is introduced. k This function is set based on the power spectral density (PSD) of typical road noise. To simultaneously perform orthogonal mapping of spatial energy on both the input (speaker end) and output (microphone end) sides, a broadband energy covariance matrix is ​​constructed: The input-side weighted broadband energy covariance matrix is: Eigenvalue decomposition is performed on it, and the global constant space input mode basis V is extracted as the complete decoding matrix. The weighted broadband energy covariance matrix on the output side is: Extract the conjugate transpose U of its global output mode basis. H As the complete encoding matrix, considering the sparsity of low-frequency acoustic modes, by calculating the cumulative energy proportion of eigenvalues, only the dominant mode vectors corresponding to the top K largest eigenvalues ​​are extracted to obtain a dimensionless real matrix. and V K . During the real-time online adaptive control phase, parallel noise reduction, primarily performed using single-channel isolation (SISO), is conducted. Step S3 (Error Space Transformation): The real-time acquired in-vehicle multi-channel physical residual error signal vector e(n) is encoded using a real number encoding matrix. Projection yields a K-dimensional modal residual error vector. . Step S4 (Independent Modal Adaptive Filtering): Decouple the large multi-channel coupled network into K mutually independent scalar channels. For the k-th modal channel, its equivalent filtered reference signal x' k,r (n) is directly generated by scalar modal equivalent path filtering. Its adaptive control weights w k,r (n) is updated rapidly iteratively according to the Normalized Least Mean Square (NLMS) rule. Since the modes are mutually orthogonal and independent, a larger step size can be assigned to the dominant mode to accelerate convergence. Step S5 (Modal Synthesis and Physical Output): Each parallel modal controller calculates the modal control command y. modal (n) is then processed by the real number decoding matrix V. K Mapping and reconstruction generate the actual driving signal y(n) = V that drives the physical speaker. K y modal (n) is finally sent to the speaker to generate noise-canceling waves. It should be noted that the information interaction and execution process between the above-mentioned devices / modules are based on the same concept as the method embodiments of this application, and are systems corresponding to the above-mentioned control methods. All implementation methods in the above-mentioned method embodiments are applicable to the embodiments of this system. For details on their specific functions and the resulting technical effects, please refer to the method embodiment section, which will not be repeated here. To further verify the beneficial effects of the method and system described in this invention, a system-level verification was conducted using a benchmark 5-seat cabin of a compact sedan (A / B class) as an example. With a complex hardware configuration employing M=4 error microphones, S=6 secondary speakers, and R=4 reference sensors, this invention extracted K=4 dominant modes. Verification results show that compared to the traditional fully coupled MIMO-FxLMS algorithm, this invention significantly reduces the single-step multiply-accumulate computation (MACs) by approximately 66.7%. When deployed on an automotive-grade MCU chip (such as the NXP S32K344), the theoretical CPU utilization of its core filtering stage plummeted from 46.1% to 15.4%. Meanwhile, in terms of noise reduction performance, this invention achieves a steady-state average noise reduction of 19.77 dB. In the virtual evaluation area where the occupant's head is 15 cm away from the center of the physical microphone, the noise reduction remains stable and does not exhibit the spatial noise reduction degradation phenomenon caused by traditional zero-forcing algorithms. It successfully expands discrete "silent points" into regional "silent basins" and significantly suppresses the spatial waterbed effect. Furthermore, while this embodiment focuses on road noise suppression applications within automotive cabins, those skilled in the art should understand that the proposed broadband spatial mode decoupling method is not limited to this. Its high portability makes it fully applicable to low-frequency noise control in any confined, enclosed space with multiple-input multiple-output (MIMO) characteristics, including but not limited to: high-speed train carriages, aircraft cockpits, ship cabins, wind turbine nacelles, and soundproof enclosures for large industrial equipment. The above description of the structure, features, and effects of the present invention is based on the embodiments shown in the figures. However, the above are only preferred embodiments of the present invention. It should be noted that the technical features involved in the above embodiments and their preferred methods can be reasonably combined and matched by those skilled in the art to form a variety of equivalent solutions without departing from or changing the design concept and technical effects of the present invention. Therefore, the present invention is not limited to the scope of implementation shown in the figures. Any changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments, that do not exceed the spirit covered by the specification and figures, should be within the protection scope of the present invention.

Claims

1. A method for active control of in-vehicle road noise based on broadband spatial mode decoupling, characterized in that, Includes the following steps: S1: In offline mode, a broadband excitation signal is injected into the secondary speaker, and the response is collected through the error microphone to identify the secondary path impulse response matrix from the secondary speaker to the error microphone in the physical space. S2: Based on the secondary path impulse response matrix, a weighted broadband energy covariance matrix is ​​constructed on the speaker input side and the error microphone output side respectively, and eigenvalue decomposition is performed to extract the dimension-reduced global constant time-domain output mode basis as the encoding matrix and the global constant time-domain input mode basis as the decoding matrix. S3: In online control mode, the residual error signal at the error microphone in the space is collected in real time, and projected onto the modal error space through the encoding matrix to obtain the modal residual error signal; S4: Acquire vibration reference signals from reference sensors, and for each parallel independent modal channel, independently execute adaptive filtering algorithms based on their respective modal residual error signals and equivalent filtered reference signals to update the modal controller weights and generate modal control commands; S5: The modal control command is reconstructed into a physical drive signal through the decoding matrix, which drives the secondary speaker to output a noise-canceling sound wave to acoustically cancel out the original low-frequency noise.

2. The method according to claim 1, characterized in that, In step S2, the process of constructing the weighted broadband energy covariance matrix and performing eigenvalue decomposition specifically includes: Introducing the frequency domain energy weighting function w(ω) k The weighting function is set based on the power spectral density (PSD) of typical road surface excitation or the target noise spectral envelope; On the speaker input side, a weighted broadband energy covariance matrix Q is constructed and its eigenvalues ​​are decomposed: Extract the global input mode basis V as the complete decoding matrix; where G(ω) k ) is the secondary path frequency response matrix, N is the number of discrete frequency points, and Λ is the eigenvalue diagonal matrix; Similarly, on the error microphone output side, a weighted broadband energy covariance matrix Q is constructed. e And extract the conjugate transpose U of its global output mode basis. H As a complete encoding matrix.

3. The method according to claim 2, characterized in that, Step S2 also includes modal space truncation and cooperative diagonalization: Extract the dominant mode vectors corresponding to the top K largest eigenvalues ​​that meet the energy percentage threshold, where the number of truncated modes K ≤ min(M, S), and M is the number of error microphones and S is the number of secondary speakers; This yields the actual dimensionality-reduced encoding matrix. and the actual decoding matrix V K ; Find the element with the highest energy in each eigenvector and rotate its phase to the real number axis. Then, the complex space modal basis is losslessly projected and mapped to a real constant matrix for use in online time-domain control.

4. The method according to claim 3, characterized in that, The specific mathematical representations of steps S3 and S5 are as follows: The real-time physical residual error signal vector e(n) is encoded by a dimension-reduced real-number encoding matrix. Encoded as modal residual error vector Each modal controller independently generates a modal control command vector y modal (n) via real number decoding matrix V K Decoded into a physical drive signal vector y(n):

5. The method according to claim 1, characterized in that, In step S4, the specific method for independently executing the adaptive filtering algorithm is as follows: The system is decoupled into K parallel independent single-channel control loops, and the single-step computational complexity of the adaptive filtering of the K control loops is linearly related to K. For the k-th modal channel, its controller weight vector w k,r The update of (n) follows the Normalized Least Mean Square (NLMS) rule: Where, x' k,r (n) is the equivalent filtered reference signal, e modal,k (n) represents the modal error, μ k The step size is the adaptive step size, and δ is the regularization constant.

6. An active in-vehicle road noise control device based on broadband spatial mode decoupling, characterized in that, include: The offline processing unit includes a secondary path identification module and a mode generation and dimensionality reduction module, which are used to identify the impulse response matrix of the secondary path in the offline state, and construct the weighted broadband energy covariance matrix of the input side and the output side respectively to extract the dimensionality-reduced constant time-domain encoding matrix and decoding matrix. The online control unit includes an error space projection module, an independent adaptive filtering module, and a physical signal reconstruction module. In online control mode, it projects the real-time residual error signal onto the modal error space through the encoding matrix, performs adaptive filtering in parallel for each modal channel to generate modal control commands, and reconstructs them into physical drive signals through the decoding matrix to drive the speaker to output silencing sound waves.

7. An in-vehicle road noise active control system based on broadband spatial mode decoupling, characterized in that, include: The signal acquisition system includes a reference sensor located in the chassis or wheel area, and an error microphone located near the ear in an enclosed space. The controller includes the modules of the active control device as described in claim 6; The secondary loudspeaker sound generation system receives physical drive signals and outputs secondary sound waves in the physical space to cancel out low-frequency noise.

8. An electronic device comprising a memory and a processor, characterized in that, When the processor executes the program in the memory, it implements the steps of the method as described in any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 5.

Images