Improved system and corresponding method for active noise reduction in a vehicle passenger compartment

Abstract

An active noise reduction system for the passenger compartment (4) of a vehicle (1) includes: a reference sensor (16) designed to detect a noise reference signal indicating ambient noise to be eliminated; an error sensor (17) disposed within the passenger compartment (4) for detecting an error signal in a noise reduction area to achieve noise reduction feedback control; a loudspeaker (12) designed to reproduce sound within the passenger compartment (4); and a digital processing unit (14) designed to control the loudspeaker (12) to generate an anti-noise signal to be reproduced within the passenger compartment (4) through a noise reduction algorithm based on the reference signal, the error signal, and the determination of acoustic paths within the passenger compartment (4) to achieve noise reduction. In particular, the system has distributed intelligence, wherein the reference sensor (16) and / or the error sensor (17) have embedded digital processing capabilities and are configured to cooperate with the digital processing unit (14).

Description

Cross-reference to related applications

[0001] This patent application claims priority to Italian patent application No. 102024000028776, filed on December 17, 2024, the entire disclosure of which is incorporated herein by reference. Technical Field

[0002] This solution relates to an improved system and method for active noise reduction in the passenger compartment of a vehicle (especially a motor vehicle). Background Technology

[0003] The application of active road noise cancellation (ARNC) technology in modern motor vehicles is known, and it aims to reduce noise in the passenger compartment, which mainly originates from the rolling noise and aerodynamic noise of the vehicle's tires.

[0004] Active noise cancellation technology typically generates a signal whose waveform is reversed relative to the noise waveform (e.g., a signal with the same sound level or amplitude but reversed phase) through speakers in the passenger compartment, also known as an "inverted" or "anti-noise" waveform, thereby actively reducing acoustic noise.

[0005] Active noise cancellation systems typically employ one or more reference sensors to detect an external noise reference signal, generate an anti-noise signal based on this noise reference signal, and reproduce this anti-noise signal through one or more speakers within the passenger compartment. These anti-noise signals undergo destructive interference with the original noise signal to reduce the noise level reaching the ears of a listener (the driver or passenger of the vehicle). The active noise cancellation system also includes one or more error sensors located within the passenger compartment to detect residual noise after the destructive interference operation.

[0006] The aforementioned active noise cancellation techniques for use in the passenger compartment of motor vehicles require defining acoustic paths, particularly the so-called secondary acoustic paths, which represent the propagation path of the anti-noise signal from the loudspeaker emitting the anti-noise signal to the noise cancellation area typically located near the listener's ear (near the headrest); and may include the so-called primary paths, which represent the propagation path of the noise signal from the noise source to the aforementioned noise cancellation area. These techniques also require implementing noise cancellation algorithms based on the aforementioned acoustic paths, typically using "feedforward" hybrid control, which combines the typical structure of open-loop control with feedback of the error signal.

[0007] Typically, noise reduction algorithms are quite complex and computationally expensive.

[0008] Therefore, a trade-off usually needs to be made between the performance of the algorithm (in terms of noise reduction) and the computational cost required by the relevant processing units in the vehicle (which are usually also dedicated to other functions within the vehicle). Summary of the Invention

[0009] The overall objective of this solution is to provide an active road noise reduction system that can overcome or, in any way, constrain the aforementioned problems.

[0010] To achieve the above objectives, this solution provides an active noise reduction system, vehicle, and method for use in the passenger compartment of a vehicle.

[0011] According to one embodiment of the present invention, an active noise cancellation system for the passenger compartment of a vehicle is provided, comprising:

[0012] A reference sensor is configured to detect a noise reference signal indicating ambient noise;

[0013] An error sensor is installed in the passenger compartment and is configured to detect error signals in the noise reduction area to achieve noise reduction feedback control.

[0014] A loudspeaker is configured to reproduce sound within the crew compartment; and

[0015] The digital processing unit is configured to execute a noise reduction algorithm based on the noise reference signal, the error signal, and the determination of the acoustic path within the crew cabin, thereby controlling the loudspeakers to generate an anti-noise signal to be reproduced within the crew cabin to achieve noise reduction.

[0016] The system features distributed intelligence, wherein the reference sensor and / or the error sensor have embedded digital processing capabilities and are configured to cooperate with the digital processing unit to facilitate the execution of the noise reduction algorithm.

[0017] In one embodiment, the reference sensor and / or the error sensor includes: a sensing stage configured to provide a detection signal; and a processing stage operatively coupled to the sensing stage and configured to perform processing on the detection signal to generate a preprocessed signal provided to the digital processing unit as input, based on the detection signal and the noise reduction algorithm.

[0018] In one embodiment, the processing level includes a digital signal processor (DSP), a microcontroller, a machine learning core processor (MLC), or a similar digital processing unit integrated with the sensing level.

[0019] In one embodiment, the digital processing unit is configured to receive the preprocessed signal from the reference sensor and / or the error sensor, and perform further processing based on the noise reduction algorithm to generate a noise-reduced signal designed to be provided as input to one or more of the speakers for generating the anti-noise signal in a noise reduction area within the passenger compartment. The anti-noise signal is designed to cancel out noise signals to determine the residual noise level reaching the listener's ear.

[0020] In one embodiment, the processing stage is configured to filter the detected signal by evaluating a secondary acoustic path within the occupant compartment, wherein the secondary acoustic path is defined between a corresponding one of the loudspeakers and a corresponding one of the error sensors, and output the filtered noise reference signal as the preprocessed signal to the digital processing unit.

[0021] In one embodiment, the digital processing unit is configured to perform a noise reduction stage, the noise reduction stage including: a control block configured to output a control signal designed to drive at least one of the speakers to generate the noise-resistant signal; and an adaptive block configured to adapt the operation of the control block according to the filtered noise reference signal received from the processing stage.

[0022] In one embodiment, the processing stage is configured with digital signal processor functionality to perform one or more of the following signal processing operations: performing a Fast Fourier Transform (FFT) frequency domain transformation operation on the detected signal; a convolution operation to filter the detected signal by evaluating the secondary acoustic path within the crew cabin and outputting the filtered noise reference signal; and a multiple coherence calculation operation to assign appropriate weights to the noise reference signal acquired by the reference sensor for noise reduction operations.

[0023] In one embodiment, the processing stage is configured to execute artificial intelligence and machine learning algorithms and perform one or more of the following operations: merging the noise reference signals acquired by the reference sensors based on a deep learning model and according to a multicoherence criterion; synthesizing the filtered noise reference signal from the detection signal based on a deep learning model; and evaluating the contribution of each of the reference sensors to the total noise to be eliminated based on a deep learning model.

[0024] In one embodiment, the processing level is designed to be optimized "offline" in an initial phase before the system is used while the vehicle is in motion; and in a subsequent online use phase, i.e. while the system is running and the vehicle is in motion, the processing level is configured to update over time as the vehicle's operating and working conditions change.

[0025] In one embodiment, the processing stage is configured to facilitate the execution of an optimization algorithm for positioning the reference sensor and / or the error sensor at optimized positions relative to the vehicle for the noise reduction operation; the optimization algorithm aims to: determine an initial large set of positions for the reference sensor and / or the error sensor; and subsequently identify, from the initial large set of positions, an optimal subset of positions that minimizes the prediction error of the total noise to be eliminated.

[0026] According to one embodiment of the present invention, a vehicle is provided, including the above-described active noise cancellation system.

[0027] According to one embodiment of the present invention, an active noise cancellation method for an passenger compartment in a vehicle is provided, comprising: detecting a noise reference signal indicating ambient noise to be eliminated by a reference sensor; detecting an error signal in a noise reduction area by an error sensor disposed in the passenger compartment to achieve feedback control of the noise reduction; executing a noise reduction algorithm based on the noise reference signal, the error signal, and a determination of the acoustic path within the passenger compartment, thereby controlling a loudspeaker through a digital processing unit to generate an anti-noise signal to be reproduced in the passenger compartment to achieve the purpose of noise reduction.

[0028] The control includes the execution of distributed intelligence, wherein the reference sensor and / or the error sensor have embedded digital processing capabilities and are configured to cooperate with the digital processing unit to facilitate the execution of the noise reduction algorithm.

[0029] In one embodiment, the method generates a preprocessed signal based on a detected signal and the noise reduction algorithm using the reference sensor and / or error sensor; and further processes the preprocessed signal using the digital processing unit to generate a noise-reduced signal, the noise-reduced signal being designed to be provided as input to one or more of the speakers for generating the anti-noise signal in a noise reduction area within the passenger compartment, the anti-noise signal being designed to cancel-interference with the noise signal to determine the residual noise level reaching the listener's ear.

[0030] In one embodiment, the method further includes: filtering the detected signal by evaluating the secondary acoustic path within the crew cabin using the reference sensor and / or error sensor, and outputting the filtered noise reference signal as the preprocessed signal to the digital processing unit.

[0031] In one embodiment, the method includes executing an optimization algorithm for optimizing the positioning of the reference sensor and / or the error sensor relative to the vehicle at an optimized location for the noise reduction operation; wherein the optimization algorithm aims to: determine an initial large set of locations for the reference sensor and / or the error sensor; and subsequently identify an optimal subset of locations from the initial large set of locations that minimizes the prediction error of the total noise to be eliminated. Attached Figure Description

[0032] The invention will now be described with reference to the accompanying drawings, which illustrate non-limiting embodiments thereof, wherein:

[0033] Figure 1 A motor vehicle equipped with an active road noise reduction system is shown.

[0034] Figure 2 This is a schematic block diagram of an active noise reduction algorithm.

[0035] Figure 3 and Figure 4 A block diagram of the noise reduction system architecture according to one aspect of this solution is shown.

[0036] Figure 5 A flowchart illustrating the operation for optimizing the location of noise sensors in motor vehicles, according to another aspect of this scheme, is shown. Detailed Implementation

[0037] As described in detail below, one aspect of this solution generally requires the implementation of a system for active road noise reduction within a vehicle's passenger compartment. This system possesses distributed intelligence, thereby reducing the computational load required by the corresponding processing units within the vehicle and / or improving noise reduction performance. At least a portion of the processing required for the noise reduction algorithm is actually performed in an embedded manner within sensors (particularly reference sensors and / or error sensors), which are operatively coupled to the aforementioned processing units and used within the same active noise reduction system.

[0038] Figure 1 A vehicle, particularly a motor vehicle 1 (which may be conventional or combustion, hybrid or electric), is shown, which is equipped with a body 2 supported on the ground by wheels 3, the body 2 defining a passenger compartment 4 and including front and rear doors 5 that allow access to the passenger compartment 4 and a hatchback door 6 that allows access to the corresponding luggage compartment.

[0039] Inside the passenger compartment 4, the motor vehicle 1 includes a pair of front seats 8 arranged in front (relative to the direction of travel) for accommodating the driver and a passenger of the motor vehicle 1; and a rear seat 9 arranged rearward (relative to the direction of travel).

[0040] Specifically, the motor vehicle 1 includes an active road noise reduction system (hereinafter referred to as ARNC system) 10, which is generally part of the audio system of the motor vehicle 1 and is configured to reproduce sound signals within the passenger compartment 4. For this purpose, it includes a plurality of speakers 12, which have different frequency contributions in a known manner (e.g., midrange speakers, tweeters, etc.) and are generally arranged at the doors 5. In the illustrated embodiment, the audio system also includes a woofer arranged in the trunk.

[0041] ARNC system 10 includes a digital processing unit 14 (with a microprocessor, microcontroller, etc.) configured to control the aforementioned speaker 12 to generate a signal to be reproduced within the passenger compartment 4, particularly for the purpose of road noise reduction.

[0042] In one possible implementation, the digital processing unit 14 may overlap with or be part of the control management unit of the audio system, or typically overlap with or be part of the control management unit of the infotainment system of the motor vehicle 1, which is designed to control the generation of digital audio-video content (information and entertainment) in a known manner, and may control the activation, regulation, or monitoring of various functions of the motor vehicle 1 (e.g., management of air conditioning or control of data related to the operation of the motor vehicle 1).

[0043] The digital processing unit 14 is also operatively coupled to the electronic control unit (ECU) of the motor vehicle 1, which is designed to supervise the overall operation of the motor vehicle 1 (in a known manner, not shown herein).

[0044] The ARNC system 10 also includes: a reference sensor 16, such as an audio sensor or microphone, but more commonly an accelerometer in automotive applications, typically located outside the passenger compartment 4, configured to detect a reference signal indicating ambient noise to be eliminated; and an error sensor 17, such as an audio sensor or microphone, located inside the passenger compartment 4 of the vehicle 1, configured to detect an error signal in the noise reduction area (for noise reduction feedback control discussed below).

[0045] In one possible implementation, the error sensor 17 is arranged at the headrest of the front seat 8 (and possibly the rear seat 9) to detect noise in a noise-reducing area corresponding to the ear of the user (at least the driver) of the motor vehicle 1.

[0046] Reference sensor 16 may be positioned, for example, at the wheel 3 of motor vehicle 1 to provide detection of environmental noise, such as tire rolling noise or aerodynamic noise (further details regarding the possible location of such reference sensor 16 will be provided below).

[0047] The aforementioned digital processing unit 14 is configured to execute a noise reduction algorithm within the passenger compartment 4. In addition to other features, the noise reduction algorithm also needs to determine the acoustic path and the corresponding frequency transfer function within the passenger compartment 4 of the motor vehicle 1.

[0048] In particular, such as Figure 1 As illustrated schematically, these acoustic paths may include: primary acoustic paths, with P p This indicates that it defines the transfer function between the corresponding reference sensor 16 and the corresponding error sensor 17; and the secondary acoustic path, with P s This indicates that it defines the transfer function between the corresponding loudspeaker 12 and the corresponding error sensor 17 within the crew cabin 4 (the aforementioned primary acoustic path is usually determined indirectly by an algorithm).

[0049] Digital processing unit 14 is configured to perform an active noise reduction level based on these acoustic paths and the signals detected by the aforementioned reference sensor 16 and error sensor 17, particularly by means of the ARNC algorithm; Figure 2 One possible implementation of this noise reduction level is shown, denoted by 24.

[0050] Specifically, in Figure 2 In the first transfer block 30, the transfer function of the primary acoustic path P(z) is represented, which models the propagation of noise signals between the noise source and the noise reduction region (i.e., between the corresponding one in the reference sensor 16 and the corresponding one in the error sensor 17).

[0051] The first transmission block 30 schematically receives the noise reference signal x(n) provided by the aforementioned reference sensor 16 and outputs a primary interference signal d(n), which represents the noise to be eliminated in the noise reduction area.

[0052] The second transfer block 32 represents the transfer function of the secondary acoustic path S'(z), which models the audio signal propagation between the corresponding loudspeaker 12 and the aforementioned noise reduction region.

[0053] The second transmission block 32 schematically receives the control signal c(n) and outputs the noise-resistant signal y(n), wherein the control signal c(n) represents the signal provided to the aforementioned speaker 12.

[0054] Figure 2 A difference block 34 is also shown, which schematically receives the aforementioned primary interference signal d(n) at the summation input and the anti-noise signal y(n) at the differential input, thereby outputting an error signal e(n) (which is detected by the aforementioned error sensor 17).

[0055] In detail, the noise reduction stage 24 includes: a control filter block 35 configured to receive a noise reference signal x(n) as input and generate the aforementioned control signal c(n) as output; and an adaptive block 36 configured to appropriately adjust the filtering action performed by the control filter block W(z), for example by modifying the weights, coefficients and / or transfer function.

[0056] The adaptive block 36 can be operated, for example, by the Least Mean Squares (LMS) algorithm or a similar adaptive algorithm, and receives the aforementioned error signal e(n) (typically detected by one of the aforementioned error sensors 17) and the filtered noise reference signal x'(n) as input.

[0057] In this regard, the noise reduction stage 24 includes an evaluation block 37, which receives a noise reference signal x(n) as input and outputs a filtered noise reference signal x'(n), and performs a secondary acoustic path evaluation S'(z).

[0058] According to a specific aspect of this plan, such as Figure 3 As illustrated, the ARNC system 10 has distributed intelligence, wherein the sensors used in the same ARNC system 10 (in this embodiment, the aforementioned reference sensor 16 is shown as an example) are internally embedded with digital processing capabilities and cooperate with the aforementioned digital processing unit 14 to facilitate the execution of the aforementioned noise reduction level 24 and the corresponding noise reduction algorithm, performing at least some of the required processing.

[0059] Specifically, the sensor (in this example, the aforementioned reference sensor 16) includes a processing stage 42 in addition to the sensing stage 40. The sensing stage 40 is configured to provide a detection signal S based on the quantity to be detected (in this specific case, an external noise reference signal related to rolling noise and / or aerodynamic noise). d The processing stage 42 is operatively coupled to the sensing stage 40 and configured to process the detection signal S. d Perform appropriate processing based on the detection signal S d Generate preprocessed signal S p The preprocessed signal S p It is then provided as input to the digital processing unit 14 (representing the aforementioned noise reference signal x(n)).

[0060] The processing stage 42 can be digital, such as including a digital signal processor (DSP), microcontroller, machine learning core (MLC) processor, or other similar digital processing units configured in an embedded manner.

[0061] In addition to the first sensing structure (typically an accelerometer structure with one or more detection axes), the aforementioned sensing stage 40 may optionally include additional sensing structures for other physical quantities of interest, such as sensing structures defining microphones and / or gyroscopes.

[0062] In one possible implementation, the aforementioned reference sensor 16 is implemented as a micro-electro-mechanical system (MEMS) type smart sensor, employing microfabrication technology of semiconductor materials.

[0063] The digital processing unit 14 of the ARNC system 10 receives the preprocessed signal S from the aforementioned reference sensor 16. p And perform appropriate further processing to generate the noise-reduced signal SANC, corresponding to the aforementioned noise-resistant signal y(n).

[0064] These noise reduction signals S ANC The noise is provided as input to one or more of the aforementioned speakers 12 to generate an anti-noise signal in a noise reduction area within the crew cabin 4. The anti-noise signal is designed to cancel out the original noise signal, thereby reducing the residual noise level reaching the listener's ears.

[0065] Figure 4 The diagram illustrates one possible implementation of the architecture of the ARNC system 10 within the passenger compartment 4 of a motor vehicle 1. In this case, the system includes four smart sensors (reference sensor 16 in the example, located at the four wheels 3 of the motor vehicle 1, not shown herein), each of which is intelligent, i.e., the aforementioned processing stage 42 is embedded within it (shown schematically).

[0066] In the example shown, these reference sensors 16 can perform detection functions not only as accelerometers (“acc”) but also as microphones and / or gyroscopes (“gyro”), and are configured to detect physical quantities of interest, and based on the detection signal (S) d The preprocessed output preprocessed signal (S) p ).

[0067] The ARNC system 10 also includes, in the illustrated example, five actuators or speakers 12, arranged at the (front and rear) doors 5 and the trunk of the vehicle 1; and a digital processing unit 14 operatively coupled to a smart sensor (reference sensor 16 in the example) and receiving preprocessed signals (S... p (May also include the original detection signal S) d Simultaneously coupled to the speaker 12, the digital processing unit 14 transmits the noise reduction signal S ANCThe speaker 12 is driven to generate an anti-noise waveform.

[0068] Advantageously, distributed signal processing for noise reduction purposes, jointly performed by the digital processing unit 14 and the smart sensors, can achieve the following: reduced latency due to parallel data processing (through the processing stage 42 of the smart sensors and the digital processing unit 14); improved performance of the executed noise reduction algorithm, or, with the same performance of the noise reduction algorithm, implementation of the digital processing unit 14 using fewer computational execution components; and simplified architecture and wiring of the sensor network (e.g., the smart sensors can be interconnected in a "star center" pattern before being connected to the digital processing unit 14, and / or the smart sensors can be connected to the digital processing unit 14 via their respective single connection lines, regardless of the number of detection channels used).

[0069] More specifically, the aforementioned processing stage 42 of the reference sensor 16 may be configured with digital signal processing (DSP) functions, and the signal processing functions performed may include, for example, one or more of the following:

[0070] For the detection signal (S) d The signal is already in the frequency domain and is transferred to the digital processing unit 14. Therefore, the digital processing unit 14 does not need to perform these frequency domain transformation operations.

[0071] Convolution operations, such as evaluating the detection signal S via the secondary acoustic path S'(z), are used to perform convolution operations. d The signal is filtered, and the filtered noise reference signal x'(n) is output. The filtered noise reference signal x'(n) is used as the aforementioned preprocessed signal S. p Directly provided to the digital processing unit 14 (in the example, reference 14) Figure 2 The evaluation block 37 of the noise reduction stage 24 described herein can be directly implemented within the processing stage 42 embedded in the smart sensor in this case.

[0072] Multiple coherence calculation operations, such as assigning appropriate weights to the noise reference signals acquired by each reference sensor 16 for noise reduction operations.

[0073] Alternatively or additionally, the processing stage 42 of the smart sensor may be configured to execute artificial intelligence (AI) and machine learning algorithms and architectures, and the functions performed may include one or more of the following:

[0074] - A deep learning model is used to merge noisy reference signals based on the multicoherence criterion (thereby reducing the computational cost of the digital processing unit 14, which can receive a highly coherent preprocessed signal as input).

[0075] -Using a deep learning model capable of predicting the value of the filtered noise reference signal x'(n), from the detected signal S d The operation of synthesizing the filtered noise reference signal x'(n);

[0076] - Using a deep learning model, the contribution of each sensor to the overall noise reduction is evaluated (in this case, the digital processing unit 14 can combine these evaluation results to determine the control action).

[0077] According to one aspect of this scheme, the smart sensor (such as the aforementioned reference sensor 16) can be optimized "offline" in particular for the operating characteristics of the corresponding processing stage 42, that is, optimized in the preliminary steps before the ARNC system 10 is actually used while the motor vehicle 1 is in motion, and the computational load of this preliminary step may also be large.

[0078] In subsequent online usage steps, i.e. while the ARNC system 10 is running and the vehicle 1 is in motion, the reference sensor 16 can process information based on previous optimizations and adjustments.

[0079] Advantageously, the calibration operation of the noise reduction algorithm does not have to be completed in the aforementioned offline steps.

[0080] In this case, the ARNC system 10 can be updated over time as conditions change, continuously updating during the operation of the vehicle 1, thereby optimizing and calibrating (or “fine-tuning”) the noise reduction on the vehicle 1.

[0081] Specifically, sensor optimization can remain active during user operation, and the noise reduction algorithm can be updated accordingly.

[0082] If the processing stage 42 of the smart sensor executes an artificial intelligence algorithm, the training of the neural network used by the deep learning model can be performed in the aforementioned offline steps, and advantageously, it can also be performed continuously over time during online use.

[0083] The neural network can be trained in an offline training step to predict the noise to be eliminated, thereby minimizing the computational load required by the smart sensor during online use; specifically, the neural network can be trained based on a noise reference signal (the aforementioned detection signal S) that the reference sensor 16 is intended to acquire. d (This is used to predict the noise to be eliminated.)

[0084] Advantageously, training does not have to end in the offline step, but can continue during the operation of ARNC system 10, maintaining stable performance even if the operating conditions associated with the operation of motor vehicle 1 change.

[0085] Specifically, each neural network can be trained based on a dataset, which can be divided into:

[0086] - The training set, which is the portion of the dataset used for training, is collected through the measurement step;

[0087] - The test set, which is the portion of the dataset used to test the performance of the neural network, is also collected through the measurement process.

[0088] The training set can be further divided into input data and output data, which correspond to known inputs and known outputs. Therefore, the neural network can "learn" the relationship between inputs and outputs, thus enabling it to predict outputs corresponding to future inputs that are not part of the training dataset.

[0089] The performance can be verified in the second measurement step on motor vehicle 1.

[0090] According to one aspect of this solution, the use of artificial intelligence algorithms by intelligent sensors (such as the aforementioned reference sensor 16) can also optimize the positioning of the sensors in the motor vehicle 1, especially positioning them at locations optimized for subsequent noise reduction operations.

[0091] like Figure 5 As shown, the algorithm for optimizing this positioning can, in the first step 50, determine a first large set of locations (so-called “superset”) of the smart sensor (e.g., reference sensor 16) by means of the following techniques (known, but not described in detail herein): characterization of the main vibroacoustic path that causes noise to be structurally transmitted within the passenger compartment 4 of the motor vehicle 1; characterization of aeroacoustic contribution; and / or characterization of noise sources.

[0092] In subsequent step 52, an optimal subset of locations can be identified from the aforementioned location superset. This optimal subset of locations is achieved by a trained neural network implemented in the corresponding processing stage 42 of each smart sensor, which minimizes the prediction error of each smart sensor for the total noise to be eliminated.

[0093] Specifically, as shown in step 53, a neural network can be trained (for each sensor) to evaluate the individual contribution to multicoherence, thereby selecting the optimal subset of sensors.

[0094] Subsequently, in step 54, a neural network of smart sensors associated with the previously identified optimal subset can be trained to “compensate” the information contribution provided by the superset of sensors that were discarded by the optimization process.

[0095] Therefore, this optimization process can determine the optimal positioning of smart sensors used for noise reduction, thereby achieving the best balance between performance and resource utilization.

[0096] Based on the above, the advantages of this solution are obvious.

[0097] In any case, it should be pointed out again that this solution can reduce the computational cost of the digital processing unit 14 by allocating or dividing the noise signal processing operations between the smart sensor and the actual digital processing unit 14.

[0098] This can lead to the use of less computationally intensive execution components or increased complexity of the noise reduction algorithm (e.g., in terms of the number of channels, filter complexity, or the use of strategies that require complex processing), optimizing the performance of the noise reduction system, especially relative to the actual location of the driver or passenger in the crew compartment 4.

[0099] As mentioned earlier, at the cost of increased computational cost (which is still advantageous compared to traditional solutions), it is also advantageous to continuously perform calibration of the noise reduction algorithm over time, so that the system can be updated as the conditions of the vehicle 1 change (ensuring the stability of performance over time).

[0100] Finally, it is obvious that the above solutions can be modified and varied without exceeding the scope of protection of the invention as defined by the appended claims.

[0101] Specifically, it should be noted that, as mentioned above, in one possible implementation, the aforementioned distributed intelligence can be alternatively or additionally implemented through the error sensor 17 of the ARNC system 10, which can also be embedded with digital processing capabilities and cooperate with the aforementioned digital processing unit 14 to facilitate the execution of the noise reduction level 24 and the corresponding noise reduction algorithm.

[0102] In this case, these error sensors 17 will therefore be equipped with the aforementioned processing stage 42, which is operatively coupled to the corresponding sensing stage 40 and configured to respond to the corresponding detection signal S. d Perform appropriate processing according to the detection signal S d Generate preprocessed signal S p The preprocessed signal S p It is provided as input to the digital processing unit 14 (in a manner exactly similar to that described above for reference sensor 16).

[0103] Furthermore, it should be noted that the solutions disclosed herein can be used in a variety of applications. Even if they differ from the examples described in detail above, in general, the solutions can be used whenever noise reduction is required in the passenger compartment of a vehicle (of any type).

Claims

1. An active noise cancellation system (10) for use in the passenger compartment (4) of a vehicle (1), comprising: The reference sensor (16) is configured to detect a noise reference signal indicating ambient noise; An error sensor (17) is provided in the crew compartment (4). The error sensor (17) is configured to detect error signals in the noise reduction area to achieve noise reduction feedback control. A loudspeaker (12) is configured to reproduce sound within the crew compartment (4); as well as The digital processing unit (14) is configured to execute a noise reduction algorithm based on the noise reference signal, the error signal, and the determination of the acoustic path within the crew cabin (4), thereby controlling the loudspeaker (12) to generate an anti-noise signal to be reproduced within the crew cabin (4) to achieve noise reduction. The system is characterized by distributed intelligence, wherein the reference sensor (16) and / or the error sensor (17) are embedded with digital processing capabilities and are configured to cooperate with the digital processing unit (14) to facilitate the execution of the noise reduction algorithm.

2. The system according to claim 1, wherein, The reference sensor (16) and / or the error sensor (17) include: a sensing stage (40) configured to provide a detection signal (S d ); and a processing stage (42), which is operatively coupled to the sensing stage (40) and configured to process the detection signal (S). d ) performs processing to perform processing based on the detection signal (S) d Based on the noise reduction algorithm, a preprocessed signal (S) is generated and provided to the digital processing unit (14) as input. p ).

3. The system according to claim 2, wherein, The processing level (42) includes a digital signal processor (DSP), a microcontroller, a machine learning core processor (MLC), or a similar digital processing unit integrated with the sensing level (40).

4. The system according to claim 2 or 3, wherein, The digital processing unit (14) is configured to receive the preprocessed signal (S) from the reference sensor (16) and / or the error sensor (17). p ), and perform further processing based on the aforementioned noise reduction algorithm to generate a noise-reduced signal (S). ANC The noise reduction signal (S) ANC The noise cancellation signal is designed to be provided as an input to one or more of the speakers (12) for generating the noise cancellation signal in the noise reduction area within the crew cabin (4), the noise cancellation signal being designed to cancel out the noise signal to determine the residual noise level reaching the listener's ear.

5. The system according to any one of claims 2 to 4, wherein, The processing stage (42) is configured to evaluate the detected signal (S) by assessing the secondary acoustic path (S'(z)) within the crew cabin (4). d The secondary acoustic path (S'(z)) is defined between a corresponding one of the loudspeakers (12) and a corresponding one of the error sensors (17), and the filtered noise reference signal (x'(n)) is used as the preprocessed signal (S). p The output is sent to the digital processing unit (14).

6. The system according to claim 5, wherein, The digital processing unit (14) is configured to execute a noise reduction stage (24), which includes: a control block (35) configured to output a control signal (c(n)) designed to drive at least one of the speakers (12) to generate the noise-resistant signal; and an adaptive block (36) configured to adapt the operation of the control block (35) to the filtered noise reference signal (x'(n)) received from the processing stage (42).

7. The system according to any one of claims 2 to 6, wherein, The processing stage (42) is configured with digital signal processor functionality to perform one or more of the following signal processing operations: processing the detected signal (S d Perform a Fast Fourier Transform (FFT) frequency domain transformation operation; perform a convolution operation to evaluate the detected signal (S) by assessing the secondary acoustic path (S'(z)) within the crew cabin (4). d ) is filtered, and the filtered noise reference signal (x'(n)) is output; The multiple coherence calculation operation assigns corresponding weights to the noise reference signal acquired by the reference sensor (16) for noise reduction operation.

8. The system according to any one of claims 2 to 7, wherein, The processing stage (42) is configured to execute artificial intelligence and machine learning algorithms and perform one or more of the following operations: merging the noise reference signal acquired by the reference sensor (16) based on a deep learning model and according to the multicoherence criterion; and merging the noise reference signal acquired by the reference sensor (16) based on a deep learning model and the detection signal (S) d The operation of synthesizing the filtered noise reference signal (x'(n)); and the operation of evaluating the contribution of each of the reference sensors (16) to the total noise to be eliminated based on a deep learning model.

9. The system according to claim 7 or 8, wherein, The processing level (42) is designed to be "offline" optimized during the initial phase before the system (10) is used while the vehicle (1) is in motion; and during the subsequent online use phase, i.e. while the system (10) is running and the vehicle (1) is in motion, the processing level (42) is configured to be updated over time as the operating and working conditions of the vehicle (1) change.

10. The system according to any one of claims 2 to 9, wherein, The processing stage (42) is configured to facilitate the execution of an optimization algorithm for positioning the reference sensor (16) and / or the error sensor (17) at optimized positions relative to the vehicle (1) for the noise reduction operation; the optimization algorithm aims to: determine an initial large set of positions for the reference sensor (16) and / or the error sensor (17); and subsequently identify an optimal subset of positions from the initial large set of positions that minimizes the prediction error of the total noise to be eliminated.

11. A vehicle (1) comprising an active noise cancellation system (10) according to any one of claims 1 to 10.

12. An active noise reduction method for use in the passenger compartment (4) of a vehicle (1), comprising: The noise reference signal indicating the environmental noise to be eliminated is detected by the reference sensor (16); Error signals are detected in the noise reduction area by an error sensor (17) installed in the passenger compartment (4) to achieve feedback control of the noise reduction; a noise reduction algorithm is executed based on the noise reference signal, the error signal, and the determination of the acoustic path in the passenger compartment (4), thereby controlling the loudspeaker (12) through the digital processing unit (14) to generate an anti-noise signal to be reproduced in the passenger compartment (4) to achieve the purpose of noise reduction. The control includes the execution of distributed intelligence, wherein the reference sensor (16) and / or the error sensor (17) have embedded digital processing capabilities and are configured to cooperate with the digital processing unit (14) to facilitate the execution of the noise reduction algorithm.

13. The method of claim 12, comprising: Based on the detection signal (S) obtained by the reference sensor (16) and / or error sensor (17). d And based on the noise reduction algorithm, a preprocessed signal (S) is generated. p ); and through the digital processing unit (14), the preprocessed signal (S) p Further processing is performed to generate a noise-reduced signal (S). ANC The noise reduction signal (S) ANC The noise cancellation signal is designed to be provided as an input to one or more of the speakers (12) for generating the noise cancellation signal in the noise reduction area within the crew cabin (4), the noise cancellation signal being designed to cancel out the noise signal to determine the residual noise level reaching the listener's ear.

14. The method of claim 13, further comprising: The detected signal (S) is evaluated by assessing the secondary acoustic path (S'(z)) within the crew cabin (4) using the reference sensor (16) and / or the error sensor (17). d The noise reference signal (x'(n)) is filtered, and the filtered noise reference signal (x'(n)) is used as the preprocessed signal (S). p The output is sent to the digital processing unit (14).

15. The method according to any one of claims 12 to 14, further comprising executing an optimization algorithm for optimizing the positioning of the reference sensor (16) and / or the error sensor (17) relative to the vehicle (1) at an optimized position for the noise reduction operation; wherein, The optimization algorithm aims to: determine an initial large set of locations for the reference sensor (16) and / or the error sensor (17); and subsequently identify, from the initial large set of locations, an optimal subset of locations that minimizes the prediction error of the total noise to be eliminated.

Images