Noise reduction method, device and vehicle

Abstract

This application relates to the field of noise reduction technology, disclosing a noise reduction method, apparatus, and vehicle. The method includes: determining a theoretical noise reduction signal corresponding to a narrowband noise signal; determining a first virtual second-order filter based on the noise frequency and target mapping relationship of the narrowband noise signal; determining a target noise reduction signal for the narrowband noise signal using the first virtual second-order filter; and performing noise reduction using the target noise reduction signal. The target mapping relationship characterizes the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when their simulation performance regarding the signal change of the theoretical noise reduction signal on the secondary path is consistent. The secondary path is the path from the transmitter of the theoretical noise reduction signal to the target noise reduction location, and the order of the reference filter is greater than 2. The technical solution provided by this application reduces the computational load and complexity in the noise reduction process and improves noise reduction efficiency.

Description

Technical Field

[0001] This application relates to the field of noise reduction technology, specifically to a noise reduction method, device, and vehicle. Background Technology

[0002] Noise reduction equipment is used to reduce aerodynamic noise, road tire noise, and power system noise (such as engine / motor).

[0003] In related technologies, when performing active noise reduction on narrowband noise, in order to reduce the impact of secondary paths on the noise reduction signal, it is necessary to pre-filter the narrowband noise signal through a secondary path model, and then adjust the noise reduction signal based on the pre-filtered narrowband noise signal and the residual sound pressure signal at the target noise reduction location.

[0004] However, current methods for modeling secondary paths involve a large amount of computation and have certain errors, resulting in low noise reduction efficiency and accuracy. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this application is to provide a noise reduction method, device and vehicle that reduces the amount of computation and computational complexity in the noise reduction process and improves the noise reduction efficiency.

[0006] In a first aspect, embodiments of this application provide a noise reduction method, the method comprising: determining a theoretical noise reduction signal corresponding to a narrowband noise signal; determining a first virtual second-order filter based on the noise frequency and target mapping relationship of the narrowband noise signal; determining a target noise reduction signal of the narrowband noise signal using the first virtual second-order filter; and performing noise reduction using the target noise reduction signal.

[0007] The target mapping relationship characterizes the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the simulated performance of the first virtual second-order filter and the reference filter is consistent with respect to the signal variation of the theoretical denoised signal on the secondary path. The secondary path is the path from the transmitter of the theoretical denoised signal to the target denoising position. The order of the reference filter is greater than 2, and the target denoising signal is used to eliminate noise at the target denoising position.

[0008] The technical solution provided in this application replaces the higher-order reference filter with a second-order filter, effectively reducing the computational load and complexity of the noise reduction process by translating the convolution operation of a broadband path of hundreds of orders onto a single-frequency signal into a convolution operation of a second-order filter onto a single-frequency signal. Furthermore, the technical solution determines a virtual second-order filter based on the target mapping relationship. This virtual second-order filter has the same simulation performance as the higher-order filter and is error-free, enabling a more efficient and economical solution to achieve the noise reduction target, thus improving the practicality and widespread applicability of noise control.

[0009] One possible implementation involves using a first virtual second-order filter to determine the target denoised signal for the narrowband noise signal. Specifically, this can be achieved by: using the first virtual second-order filter to correct the narrowband noise signal, obtaining a corrected narrowband noise signal; determining the denoising error signal at the target denoising location; and optimizing the theoretical denoising signal based on the denoising error signal and the corrected narrowband noise signal to obtain the target denoised signal. The correction operation is used to eliminate the influence of secondary paths on denoising, and the denoising error signal is the residual sound pressure level after the theoretical denoising signal cancels out the narrowband noise signal.

[0010] One possible implementation is that the target noise reduction location is at the human ear position. Determining the noise reduction error signal at the target noise reduction location can be specifically achieved by: determining the noise at the target receiving device, and determining the noise reduction error signal at the target noise reduction location based on the noise at the target receiving device. Wherein, the distance between the target receiving device and the target noise reduction location is less than a first preset distance.

[0011] One possible implementation method for determining the noise at the target receiving device can be as follows: determine the intermediate noise reduction signal from the theoretical noise reduction signal to the target receiving device, and determine the noise at the target receiving device based on the noise reduction error signal and the intermediate noise reduction signal measured at the target receiving device.

[0012] One possible implementation is to determine the noise reduction error signal at the target noise reduction location based on the noise at the target sound receiving device. Specifically, this can be achieved by: determining the noise at the target noise reduction location based on the noise at the target sound receiving device location, and determining the noise reduction error signal at the target noise reduction location based on the signal changes of the noise at the target noise reduction location and the theoretical noise reduction signal simulated by the first virtual second-order filter on the secondary path.

[0013] One possible implementation is to determine the noise at the target noise reduction location based on the noise at the target sound receiving device location. Specifically, this can be achieved by: obtaining a pre-simulated noise change relationship between the target sound receiving device and the target noise reduction location, and determining the noise at the target noise reduction location based on the noise at the target sound receiving device and the noise change relationship.

[0014] One possible implementation, the process of determining the target mapping relationship, includes: representing the filtered output signal of the reference filter on the input signal as a sum of a first sine signal and a first cosine signal; representing the filtered output signal of the first virtual second-order filter on the input signal as a sum of a second sine signal and a second cosine signal; solving the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the preset conditions are met; and using the solved correspondence to determine the target mapping relationship.

[0015] The frequency of the input signal is the noise frequency. The first sine signal and the first cosine signal are both related to the order coefficients of the reference filter. The second sine signal and the second cosine signal are both related to the order coefficients of the first virtual second-order filter. The preset condition is that the first sine signal is equal to the second sine signal and the first cosine signal is equal to the second cosine signal.

[0016] One possible implementation, the process of determining the reference filter, includes: using multiple single-frequency sinusoidal signals of fixed frequencies as excitation sources, determining the phase response and amplitude response of the reference filter at each fixed frequency; calculating the phase response and amplitude response of the reference filter within a target frequency range based on the phase response and amplitude response at the fixed frequencies; and fitting the coefficients of the reference filter based on the phase response and amplitude response within the target frequency range. Here, the fixed frequencies are multiple frequencies determined at fixed frequency intervals within the target frequency range, and the target frequency range is the frequency range to which the narrowband noise signal belongs.

[0017] The above implementation method uses a single-frequency signal instead of a white noise signal as the excitation source. The single-frequency signal has the characteristics of single frequency and concentrated energy, resulting in higher excitation intensity and identification accuracy. It can accurately excite and observe the response characteristics of non-causal systems (active noise reduction) at specific frequency points, clearly capture the amplitude response and phase lag (or lead) relationship of the system at specific frequency points, improve the prediction performance of the reference filter, and effectively solve the problem of filter model error under non-causal conditions caused by white noise excitation.

[0018] One possible implementation method for determining the theoretical denoising signal corresponding to the narrowband noise signal is as follows: input the narrowband noise signal into an adaptive filter to obtain the theoretical denoising signal. The theoretical denoising signal is a signal with opposite phase and equal amplitude to the narrowband noise signal.

[0019] One possible implementation is to optimize the theoretical denoising signal based on the denoising error signal and the corrected narrowband noise signal to obtain the target denoising signal. Specifically, this can be achieved by optimizing the weight coefficients of the adaptive filter based on the denoising error signal and the corrected narrowband noise signal, and then using the optimized adaptive filter to generate the target denoising signal.

[0020] One possible implementation is that the intermediate noise-reduced signal from the theoretical noise-reduced signal to the target receiving device is determined using a second virtual second-order filter. This second virtual second-order filter simulates the signal changes along the physical path of the theoretical noise-reduced signal from the transmitter to the target receiving device.

[0021] Secondly, this application provides a noise reduction device, which includes a processing module and a noise reduction module.

[0022] The aforementioned processing module is used to determine the theoretical noise reduction signal corresponding to the narrowband noise signal.

[0023] The aforementioned processing module is also used to determine the first virtual second-order filter based on the noise frequency and target mapping relationship of the narrowband noise signal.

[0024] The aforementioned processing module is also used to determine the target noise-reduced signal of the narrowband noise signal using the first virtual second-order filter.

[0025] The aforementioned noise reduction module is used to perform noise reduction using the target noise reduction signal.

[0026] The target mapping relationship characterizes the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the simulated performance of the first virtual second-order filter and the reference filter is consistent with respect to the signal variation of the theoretical denoised signal on the secondary path. The secondary path is the path from the transmitter of the theoretical denoised signal to the target denoising position. The order of the reference filter is greater than 2, and the target denoising signal is used to eliminate noise at the target denoising position.

[0027] One possible implementation, the aforementioned processing module, is specifically used for: correcting the narrowband noise signal using a first virtual second-order filter to obtain a corrected narrowband noise signal; determining the noise reduction error signal at the target noise reduction location; and optimizing the theoretical noise reduction signal based on the noise reduction error signal and the corrected narrowband noise signal to obtain the target noise reduction signal. The correction operation is used to eliminate the influence of secondary paths on noise reduction, and the noise reduction error signal is the remaining sound pressure level signal after the theoretical noise reduction signal cancels out the narrowband noise signal.

[0028] One possible implementation is that the target noise reduction location is at the human ear position. The aforementioned processing module is specifically used to: determine the noise at the target receiving device, and determine the noise reduction error signal at the target noise reduction location based on the noise at the target receiving device. Wherein, the distance between the location of the target receiving device and the target noise reduction location is less than a first preset distance.

[0029] One possible implementation is that the above processing module is specifically used to: determine the intermediate noise reduction signal of the theoretical noise reduction signal reaching the target sound receiving device, and determine the noise at the target sound receiving device based on the noise reduction error signal and the intermediate noise reduction signal measured at the target sound receiving device.

[0030] One possible implementation is that the above processing module is specifically used to: determine the noise at the target noise reduction location based on the noise at the target sound receiving device location, and determine the noise reduction error signal at the target noise reduction location based on the noise at the target noise reduction location and the signal change of the theoretical noise reduction signal simulated by the first virtual second-order filter on the secondary path.

[0031] One possible implementation is that the above processing module is specifically used to: obtain the noise change relationship between the pre-simulated target sound receiving device and the target noise reduction position, and determine the noise at the target noise reduction position based on the noise at the target sound receiving device and the noise change relationship.

[0032] One possible implementation is that the above processing module is specifically used to: represent the filtered output signal of the reference filter on the input signal as a sum of a first sine signal and a first cosine signal; represent the filtered output signal of the first virtual second-order filter on the input signal as a sum of a second sine signal and a second cosine signal; solve the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the preset conditions are met; and use the solved correspondence to determine the target mapping relationship.

[0033] The frequency of the input signal is the noise frequency. The first sine signal and the first cosine signal are both related to the order coefficients of the reference filter. The second sine signal and the second cosine signal are both related to the order coefficients of the first virtual second-order filter. The preset condition is that the first sine signal is equal to the second sine signal and the first cosine signal is equal to the second cosine signal.

[0034] One possible implementation, the aforementioned processing module, is specifically used for: using multiple fixed-frequency single-frequency sinusoidal signals as excitation sources, determining the phase response and amplitude response of the reference filter at each fixed frequency; calculating the phase response and amplitude response of the reference filter within a target frequency range based on the phase and amplitude responses at the fixed frequencies; and fitting the coefficients of the reference filter based on the phase and amplitude responses within the target frequency range. Here, the fixed frequencies are multiple frequencies determined at fixed frequency intervals within the target frequency range, and the target frequency range is the frequency range to which the narrowband noise signal belongs.

[0035] One possible implementation is that the aforementioned processing module is specifically used to: input the narrowband noise signal into an adaptive filter to obtain a theoretical denoising signal. The theoretical denoising signal is a signal with opposite phase and equal amplitude to the narrowband noise signal.

[0036] One possible implementation is that the above processing module is specifically used to: optimize the weight coefficients of the adaptive filter based on the noise reduction error signal and the corrected narrowband noise signal, and generate the target noise reduction signal using the optimized adaptive filter.

[0037] One possible implementation is that the intermediate noise-reduced signal from the theoretical noise-reduced signal to the target receiving device is determined using a second virtual second-order filter. This second virtual second-order filter simulates the signal changes along the physical path of the theoretical noise-reduced signal from the transmitter to the target receiving device.

[0038] The technical effects of any implementation method in the second aspect can be found in the technical effects of any implementation method in the first aspect mentioned above, and will not be repeated here.

[0039] Thirdly, this application provides a vehicle that includes the noise reduction device in any of the embodiments of the second aspect, or the vehicle applies the noise reduction method in any of the embodiments of the first aspect to reduce noise.

[0040] Fourthly, this application provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the noise reduction method in any of the embodiments of the first aspect described above.

[0041] Fifthly, this application provides a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the noise reduction method in any of the embodiments of the first aspect described above.

[0042] The solutions provided in the third to fifth aspects above can realize the noise reduction method in any of the embodiments of the first aspect above, and their specific implementations will not be described in detail here. The technical effects corresponding to any implementation of the solutions provided in the third to fifth aspects above can be found in the technical effects corresponding to any implementation of the first aspect above, and will not be described in detail here.

[0043] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application will be described below.

[0045] Figure 1 This is a schematic diagram of the structure of a noise reduction device provided in an embodiment of this application; Figure 2 A flowchart illustrating a noise reduction method provided in an embodiment of this application; Figure 3 A schematic diagram of an amplitude response curve provided for an embodiment of this application; Figure 4 A schematic diagram of a phase response curve provided in an embodiment of this application; Figure 5 A schematic diagram illustrating a method for determining a transfer function provided in an embodiment of this application; Figure 6This is a schematic diagram illustrating how a transfer function determined using white noise as an excitation source is used to predict a signal, as provided in an embodiment of this application. Figure 7 This is a schematic diagram illustrating how a transfer function determined using a single-frequency signal as the excitation source is used to predict a signal, as provided in an embodiment of this application. Figure 8 A schematic diagram of the amplitude response curve of a secondary path provided in an embodiment of this application; Figure 9 A schematic diagram of the phase response curve of a secondary path provided in an embodiment of this application; Figure 10 A schematic diagram of the structure of a vehicle provided in an embodiment of this application; Figure 11 A schematic diagram illustrating a method for determining the transfer function of the physical path between a microphone and the ear, provided in an embodiment of this application. Figure 12 A schematic diagram illustrating the framework of a noise reduction method provided in an embodiment of this application; Figure 13 This is a schematic diagram of the structure of a noise reduction device provided in an embodiment of this application; Figure 14 A schematic diagram of a comparison curve of the number of multiplication operations provided in an embodiment of this application; Figure 15 A schematic diagram of a comparison curve of the number of multiplication operations provided for an embodiment of this application; Figure 16 This application provides a schematic diagram of the change curve of vehicle rotation speed over time. Figure 17 A time-domain distribution diagram of an in-vehicle noise signal provided in an embodiment of this application; Figure 18 A frequency domain distribution diagram of an in-vehicle noise signal provided in an embodiment of this application; Figure 19 A comparative diagram illustrating the noise reduction effect provided in an embodiment of this application; Figure 20 A schematic diagram of the mean square error curve of an error signal provided in an embodiment of this application; Figure 21 This is a schematic diagram of the mean square error curve of another error signal provided in an embodiment of this application. Detailed Implementation

[0046] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0047] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0048] In the embodiments of this application, the words "exemplary," "for example," or "e.g.," are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "e.g.," in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "e.g.," is intended to present the relevant concepts in a specific manner.

[0049] The embodiments of this application are described below with reference to the accompanying drawings.

[0050] This application provides a vehicle.

[0051] Alternatively, a vehicle may also be referred to as a vehicle, mobile carrier, electric vehicle (EV), hybrid electric vehicle (HEV), plug-in hybrid electric vehicle (PHEV), fuel cell vehicle (FCV), autonomous vehicle, intelligent and connected vehicle (ICV), driverless vehicle, or new energy vehicle. In this application embodiment, the vehicle may be a sedan, sport utility vehicle (SUV), truck, electric vehicle, motorcycle, tricycle, special vehicle (such as ambulance, fire truck, police car, etc.), driverless taxi, intelligent connected bus, autonomous logistics vehicle, electric truck, etc. The method provided in this application embodiment is also applicable to various special-purpose vehicles, such as agricultural vehicles, mining vehicles, forestry vehicles, airport vehicles, and port vehicles; this application does not impose specific limitations on these.

[0052] In some embodiments, the vehicle includes noise reduction equipment.

[0053] Among them, noise reduction equipment is used to reduce aerodynamic noise, road tire noise and power system (such as engine / motor) noise, so as to improve speech clarity, reduce driver and passenger fatigue, and thus improve the user's driving experience.

[0054] For example, vehicle cabin noise mainly includes engine noise, intake and exhaust noise, and transmission system noise, with low-frequency periodic noise generated by the engine accounting for the largest proportion. Noise reduction equipment can control mid-to-high frequency noise through passive noise reduction methods such as sound insulation and vibration isolation, and can also control low-frequency noise through active noise reduction. Specifically, active noise reduction generates a signal with opposite phase and equal amplitude to the noise signal at the target noise reduction location, which is then superimposed on the noise signal to achieve cancellation.

[0055] As an example, such as Figure 1 As shown, the noise reduction device includes a sensor 100, a controller 110, a speaker 120, and a microphone 130. The sensor 100 acquires noise signals, for example, broadband noise signals acquired via a microphone or narrowband noise signals acquired via a speed sensor. The controller 110 generates a noise reduction signal based on the noise signal to cancel it out. The speaker 120 plays the noise reduction signal generated by the controller 110. The microphone 130 detects the residual sound pressure level after the noise signal is canceled at the target noise reduction location, so as to adjust the noise reduction signal generated by the controller 110 based on the residual sound pressure level to maximize the cancellation of the noise signal. Optionally, the controller 110 includes a digital signal processor, an FPGA, a microcontroller, etc. The speaker 120 includes a dynamic speaker, a planar speaker, etc. The microphone 130 includes an electret condenser microphone, a MEMS microphone, a condenser microphone, a piezoelectric microphone, etc. When reducing noise in a vehicle, the speaker 120 can be installed in the vehicle door, and the microphone 130 can be installed in the vehicle roof near the ear. This application does not limit the number, type, or installation location of the noise reduction equipment, as long as it can achieve the noise reduction function.

[0056] Optionally, the noise reduction equipment provided in this application embodiment can also be applied to the aerospace field (such as noise reduction in aircraft cabins and helicopters), the industrial manufacturing field (such as noise reduction in motors), and the defense field (such as noise reduction in submarine engines).

[0057] In some embodiments, please refer to Figure 2 , Figure 2This is a flowchart illustrating a noise reduction method provided in an embodiment of this application. The noise reduction method provided in this application can be applied to the noise reduction device described above, or to other devices capable of achieving active noise reduction, such as... Figure 2 As shown, the method includes: Step S201: The noise reduction device determines the theoretical noise reduction signal corresponding to the narrowband noise signal.

[0058] Narrowband noise refers to noise whose power spectral density is concentrated in a relatively narrow frequency band around its center frequency. For example, noise with a center frequency of 1MHz and a bandwidth of 10kHz has a very small relative bandwidth (10kHz / 1MHz = 1%), therefore it is a narrowband noise signal. If the center frequency is 1kHz and the bandwidth is 10kHz, then the noise is a wideband noise signal. Furthermore, the spectrum of a narrowband noise signal will show one or several sharp spectral peaks, with energy concentrated around a few frequency points, while the spectrum of a wideband noise signal is relatively flat, without a clearly dominant single spectral peak.

[0059] For example, narrowband noise signals are mainly generated by rotating machinery, and in vehicles, narrowband noise signals mainly originate from the engine and transmission system, motor and electronic control system, etc. The energy of narrowband noise signals is concentrated near one or a few known frequencies (such as the order noise of the engine, the whistling sound of the motor). Therefore, it is only necessary to generate a sine / cosine signal corresponding to the known frequency to cover the energy of the narrowband noise signal. That is, the narrowband noise signal can be characterized by a sine / cosine signal with the same frequency as the narrowband noise.

[0060] The theoretical noise reduction signal is a signal with opposite phase and equal amplitude to the narrowband noise signal. When the theoretical noise reduction signal and the narrowband noise signal are superimposed at the target noise reduction location, they can achieve destructive interference, thereby achieving noise reduction.

[0061] For example, a narrowband noise signal is input into an adaptive filter to obtain a theoretically denoised signal. For instance, a sine and cosine signal with the same frequency as the narrowband noise are input into the adaptive filter to obtain the theoretically denoised signal. Specifically, the adaptive filter is typically an FIR filter, which can adjust the amplitude and phase of the input signal by adjusting its weight coefficients, thereby outputting a theoretically denoised signal. The frequency of the theoretically denoised signal is the same as the input signal of the adaptive filter, and the amplitude and phase of the theoretically denoised signal are related to the weight coefficients of the adaptive filter. Optionally, a second-order filter can achieve amplitude and phase adjustment for a single-frequency sine signal; in practice, higher-order filters can be used to handle small frequency variations and errors.

[0062] Because there is a certain distance between the theoretical noise reduction signal and the target noise reduction location—for example, a distance between the speaker transmitting the theoretical noise reduction signal and the user's ear (the target noise reduction location)—the theoretical noise reduction signal will be distorted when it propagates to the target noise reduction location due to the influence of the path (secondary path) between the transmitter and the target noise reduction location. To reduce the influence of the secondary path, it is necessary to model the secondary path. The narrowband noise signal is pre-filtered using the secondary path model. The theoretical noise reduction signal is then adjusted based on the filtered narrowband noise signal and the residual sound pressure level at the target noise reduction location to obtain the target noise reduction signal. This allows for more accurate noise reduction based on the target noise reduction signal.

[0063] Step S202: The noise reduction device determines the first virtual second-order filter based on the noise frequency and target mapping relationship of the narrowband noise signal.

[0064] The target mapping relationship is used to characterize the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the simulation performance of the first virtual second-order filter and the reference filter are consistent with the theoretical noise reduction signal on the secondary path.

[0065] Specifically, the reference filter is used to simulate the secondary path to compensate for its impact on the theoretically denoised signal. Since the reference filter is of order greater than 2, the computational cost and complexity of filtering a single-frequency signal using the reference filter are far greater than those of filtering a single-frequency signal using a second-order filter. A single-frequency sinusoidal signal filtered by an M-order reference filter (such as an FIR filter) still outputs a sinusoidal signal of that frequency, with only its amplitude and phase changing. Therefore, a second-order filter can also achieve the same amplitude and phase changes. Based on this, this application proposes a first virtual second-order filter that simulates the same performance as the reference filter but significantly reduces computational cost and complexity. By simulating the secondary path using this first virtual second-order filter, the computational cost and complexity of the entire active noise reduction process are effectively reduced.

[0066] For example, the process of determining the target mapping relationship includes: representing the filtered output signal of the reference filter on the input signal as a sum of a first sine signal and a first cosine signal; representing the filtered output signal of the first virtual second-order filter on the input signal as a sum of a second sine signal and a second cosine signal; solving the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when a preset condition is met; and using the solved correspondence to determine the target mapping relationship.

[0067] Wherein, the frequency of the input signal is the noise frequency; the first sine signal and the first cosine signal are both related to the order coefficients of the reference filter; and the second sine signal and the second cosine signal are both related to the order coefficients of the first virtual second-order filter. The preset condition is that the first sine signal equals the second sine signal and the first cosine signal equals the second cosine signal.

[0068] For example, suppose the input signal (single-frequency signal) is ,in ω is the angular frequency.

[0069] The output signal after the input signal is filtered by the M-order filter Represented as:

[0070] in, , , This represents the order coefficients of the M-order filter, specifically... This represents the weight on the m-th sampling delay of the M-order filter.

[0071] The input signal is filtered by a second-order g-filter. Represented as:

[0072] in, These represent the order coefficients of a second-order filter. and These represent the weights of the two components of the second-order filter.

[0073] exist = Under the condition that the preset conditions are met, the correspondence between the coefficients of the second-order filter and the coefficients of the M-order filter is expressed as follows:

[0074]

[0075]

[0076]

[0077] in, and These represent the order coefficients of a second-order filter. This represents the order coefficients of the M-order filter.

[0078] Based on the frequency of the narrowband noise signal, the known order coefficients of the reference filter, and the target mapping relationship determined by the above formula, the order coefficients of the first virtual second-order filter for simulating the secondary path can be determined. Optionally, the order coefficients of the first virtual second-order filter corresponding to narrowband noise signals of different frequencies can be pre-stored during the experimental phase. When applied, the first virtual second-order filter corresponding to the narrowband noise signal can be directly called to simulate the secondary path, making the active noise reduction process more convenient and efficient.

[0079] In this embodiment, a second-order filter is used instead of the reference filter. The computational complexity of M multiplications and M-1 additions after filtering is reduced to 2 multiplications and 1 addition. Furthermore, the simulation performance of the first virtual second-order filter is consistent with that of the reference filter.

[0080] For example, such as Figure 3 and Figure 4 As shown, Figure 3 and Figure 4 The horizontal axis in the graph represents frequency. Figure 3 The vertical axis represents the amplitude. Figure 4 The ordinate represents the phase. Figure 3 In the diagram, curve A represents the amplitude response of the 128th-order acoustic path filter (M-order filter), curve B represents the amplitude response of the second-order filter to a 200Hz signal, curve C represents the amplitude response of the second-order filter to a 50Hz signal, and curve D represents the amplitude response of the second-order filter to a 100Hz signal. Figure 3 It can be seen that for signals of 50Hz, 100Hz, and 200Hz, the amplitude response values ​​of the three second-order filters are consistent with the amplitude response value of the M-connected filter, indicating that for signals of 50Hz, 100Hz, and 200Hz (narrowband noise-reduced signals), the amplitude response of the second-order filters is consistent with the amplitude response of the M-connected filter. Figure 4 In the diagram, curve E represents the phase response of the 128th-order acoustic path filter (M-order filter), curve F represents the phase response of the second-order filter to a 50Hz signal, curve G represents the phase response of the second-order filter to a 100Hz signal, and curve H represents the phase response of the second-order filter to a 200Hz signal. Figure 4 It can be seen that for signals of 50Hz, 100Hz, and 200Hz, the phase response values ​​of the three second-order filters are consistent with the phase response value of the M-connected filter, indicating that for signals of 50Hz, 100Hz, and 200Hz (narrowband noise-reduced signals), the phase response of the second-order filters is consistent with the phase response of the M-connected filter. Therefore, the simulation performance of the first virtual second-order filter is consistent with that of the reference filter.

[0081] The method described above, which determines the equivalent second-order filter to the reference filter based on the mapping relationship, offers higher determinism, reproducibility, computational efficiency, and lower cost compared to the traditional two-step local modeling method. Specifically, the two-step local modeling method treats the transformation of the reference filter into a second-order filter as an identification problem. After obtaining the high-order reference filter through offline broadband identification, it is fed into a data post-processing system. A single-frequency excitation signal is input at each target frequency (noise frequency), and the NLMS algorithm is used to iteratively update the second-order filter coefficients, making its output at the target frequency approximate the target response as closely as possible. This method requires iterative training and step-size tuning at each frequency, and the convergence speed and steady-state error are significantly affected by data length, noise interference, and step-size selection. The results exhibit a certain degree of randomness and unavoidable approximation error, especially in multi-frequency line spectrum scenarios, leading to additional training costs and implementation complexity. In contrast, this application directly calculates the coefficients of the second-order filter through the mapping relationship, ensuring that the complex response of the second-order filter at the target frequency is completely consistent with the reference filter, resulting in higher stability. It eliminates the need for additional single-frequency signal injection for iterative training and avoids reliance on step size adjustment for convergence, significantly reducing implementation costs. Furthermore, this application only requires repeating the process of "determining the second-order filter coefficients based on the mapping relationship" for each target frequency point. This not only ensures strict amplitude-phase matching at the target frequency point but also reduces computational complexity, making it more suitable for platforms with high real-time requirements or limited computing power. Step S203: The noise reduction device uses a first virtual second-order filter to determine the target noise reduction signal for the narrowband noise signal.

[0082] The target noise reduction signal is used to eliminate noise at the target noise reduction location. Specifically, the target noise reduction signal is obtained by adjusting the theoretical noise reduction signal based on the filtered narrowband noise signal and the residual sound pressure signal at the target noise reduction location. By combining the residual sound pressure signal at the target noise reduction location and the influence of the secondary path, the residual sound pressure signal at the target noise reduction location can be reduced as much as possible, thus achieving accurate noise reduction.

[0083] For example, the method of determining the target noise-reduced signal of the narrowband noise signal using a first virtual second-order filter includes the following steps: Step 1: Use the first virtual second-order filter to correct the narrowband noise signal to obtain the corrected narrowband noise signal.

[0084] The correction operation is used to eliminate the impact of secondary paths on noise reduction.

[0085] Specifically, the first virtual second-order filter simulates the secondary path, and the narrowband noise signal is input into the first virtual second-order filter for filtering. The resulting corrected narrowband noise signal can eliminate the influence of the secondary path on noise reduction.

[0086] For example, when simulating the secondary path using the first virtual second-order filter, the transfer function of the secondary path needs to be determined first. For instance, during the experiment, white noise is used as the excitation source to test the signal change of white noise from the transmitting end of the theoretical denoising signal to the target denoising position, thereby determining the transfer function of the secondary path, and then simulating and fitting it using the first virtual second-order filter.

[0087] Optionally, when determining the transfer function of the secondary path, white noise or a single-frequency sine wave signal can be used as the excitation source.

[0088] For example, broadband white noise is used for excitation across the entire frequency band, and fitting is achieved through a reference filter (such as a high-order FIR filter). However, the energy of the line spectrum noise is mainly distributed at a few frequency points (target frequencies). Therefore, ensuring the energy fitting effect at the target frequencies during the fitting process guarantees the accuracy of the fitting result. For example, such as... Figure 5 As shown, during the training phase, white noise is used to excite the physical acoustic path. And based on the NLMS algorithm Perform a global model update, the update formula is expressed as:

[0089]

[0090] in, Indicates the updated , This represents the white noise tap vector. Indicates the global modeling step size. Indicates the global modeling order. Indicates white noise excitation respectively and error .

[0091] Then, The input data post-processing system uses a single-frequency signal at the target frequency. The local modeling update is performed using the excitation signal to obtain... The update formula is expressed as:

[0092]

[0093]

[0094] in, Indicates the updated , Indicates a single-frequency signal , Indicates the first Local modeling step size for each frequency, express In the Modeling error at each frequency component This indicates the order of the local modeling. For a single target frequency, the above modeling and fitting process can be achieved using a second-order FIR filter (the first virtual second-order filter).

[0095] Compared to single-frequency sinusoidal signals, white noise has lower excitation intensity and identification accuracy at specific frequencies. The determined transfer function (filter) is less precise in its response to those frequencies, thus reducing the predictive performance of the determined second-order filter. For example: Figure 6 and Figure 7 As shown, when predicting a 100Hz original signal using a transfer function determined by white noise as the excitation source, the predicted signal differs from the original signal. However, when predicting a 100Hz original signal using a transfer function determined by a single-frequency signal as the excitation source, the predicted signal is more consistent with the original signal. Furthermore, the signal-to-noise ratio (SNR) of the transfer function determined by white noise as the excitation source is higher than that of the transfer function determined by the single-frequency signal as the excitation source. Therefore, the prediction performance of the transfer function determined by the single-frequency signal as the excitation source is better.

[0096] Specifically, active noise reduction is a non-causal system. A characteristic of non-causal systems is that the current output signal depends not only on current and past input signals but also on future input signals. Therefore, it places stringent requirements on the frequency characteristics of the excitation signal and the identification logic. While white noise signals have the advantage of wide frequency coverage, enabling excitation over a broad frequency range, their uniform and non-specific frequency energy distribution makes it difficult to accurately match the system's response characteristics to future input signals under non-causal constraints. Consequently, white noise excitation methods struggle to capture the core characteristic of "future inputs affecting current outputs" in non-causal systems. This leads to an inability to accurately extract the phase information and amplitude characteristics required by the filter during the identification process, resulting in a filter that deviates from the true characteristics of the system, or even contains fundamental errors, rendering it unusable for subsequent system control or response prediction.

[0097] Single-frequency sinusoidal excitation signals, characterized by their single frequency and concentrated energy, can precisely excite and observe the response characteristics of non-causal systems at specific frequency points. Under non-causal conditions, continuous excitation with a fixed-frequency single-frequency sinusoidal signal can clearly capture the amplitude response and phase lag (or lead) relationship of the system at that frequency, accurately reflecting the dynamic characteristics of the non-causal system, and thus identifying an observation filter consistent with the true state of the system. Therefore, whether for response prediction at a single target frequency or subsequent filter model verification, the filter identified by single-frequency excitation can guarantee high accuracy, effectively solving the problem of filter model errors under non-causal conditions caused by white noise excitation.

[0098] In some embodiments, the coefficients of the reference filter are determined using a single-frequency sinusoidal signal as the excitation source, thereby determining the parameters of the first virtual second-order filter.

[0099] For example, a single-frequency excitation method is used to select single-frequency signals of different frequencies one by one in the entire frequency band, and perform excitation, data acquisition, model identification and other operations to determine the coefficients of the reference filter.

[0100] Full-band excitation requires conducting experiments separately at each frequency point, which not only consumes a significant amount of experimental time and equipment resources but also increases the computational overhead of data processing and model integration. Especially when the noise reduction system needs to cover a wide frequency range and requires high frequency resolution, the tediousness of point-by-point single-frequency identification will increase significantly, and may even fail to meet the modeling efficiency requirements of engineering applications.

[0101] Therefore, to avoid the increased cost and complexity caused by full-frequency point-by-point testing, multiple fixed-frequency single-frequency sinusoidal signals are selected as excitation sources for testing. The response of the secondary path (reference filter) to the fixed-frequency signal is determined. Then, the response of the secondary path (reference filter) to the signal in the target frequency range is determined by interpolation calculation. Finally, the coefficients of the reference filter are fitted by the phase response and amplitude response of the signal in the target frequency range of the secondary path (reference filter).

[0102] The target frequency range is the frequency range to which the narrowband noise signal belongs.

[0103] Optionally, multiple fixed frequencies are multiple frequency points selected at fixed frequency intervals (such as 10Hz or 5Hz) within the target frequency range.

[0104] For example, multiple fixed frequencies are set at 10Hz intervals. After determining the phase and amplitude responses of the secondary path for signals at the fixed frequencies, interpolation is used to calculate the phase and amplitude responses of signals at frequencies other than the fixed frequencies within the target frequency range. For example: Figure 8The figure shows the amplitude response curve of the secondary path for signals in the 100Hz-200Hz range, as shown below. Figure 9 The figure shows the phase response curve of the secondary path for signals in the 100Hz-200Hz range. Figure 8 and Figure 9 In the above, fixed frequencies are 100Hz, 110Hz, 120Hz, 130Hz, 140Hz, 150Hz, 160Hz, 170Hz, 180Hz, 190Hz, and 200Hz. The amplitude and phase responses of signals with frequencies other than the fixed frequencies within the range of 100Hz-200Hz are obtained by interpolation calculation.

[0105] The above-mentioned difference calculation method avoids the problem of filter model errors caused by white noise excitation under non-causal conditions, and also avoids the cumbersome process of point-by-point single-frequency identification across the entire frequency band. It can achieve accurate modeling of the entire frequency band by accurately identifying only a small number of frequency points and combining interpolation techniques. While ensuring the correctness of the filter model and the accuracy of identification, it significantly reduces experimental and computational costs, improves modeling efficiency, and perfectly solves the contradiction between "accuracy and efficiency" in filter identification under non-causal conditions.

[0106] Step 2: Determine the noise reduction error signal at the target noise reduction location.

[0107] Among them, the noise reduction error signal is the remaining sound pressure signal after the theoretical noise reduction signal cancels out the narrowband noise signal.

[0108] For example, the sound pressure signal at the target noise reduction location can be directly obtained through a microphone as a noise reduction error signal.

[0109] In some embodiments, the target noise reduction location is the location of the human ear, the noise at the target sound receiving device is determined, and the noise reduction error signal at the target noise reduction location is determined based on the noise at the target sound receiving device.

[0110] In this scenario, the distance between the target sound-receiving device and the target noise reduction location is less than a first preset distance. For example, if the target sound-receiving device is a loudspeaker, and the loudspeaker is installed on the door closest to the driver in the vehicle, the distance between the loudspeaker and the driver's ear position is less than the preset distance, reducing the influence of secondary paths and improving the noise reduction effect.

[0111] For example, such as Figure 10 As shown, there are four loudspeakers 120, which are respectively installed on the four doors of the vehicle and output the target noise reduction signal / desired noise reduction signal respectively. Microphones 130 are also installed on the ceiling of each of the four seats in the vehicle, at a distance of 20-50cm from the listener's ear. Microphones 130 are used to directly acquire the noise reduction error signal from the target sound receiving device. This application does not limit the number, type, or placement of the loudspeakers 120 and microphones 130.

[0112] For example, an intermediate noise-reduced signal is determined when the theoretical noise-reduced signal reaches the target receiving device. Based on the noise reduction error signal measured at the target receiving device and the intermediate noise-reduced signal, the noise at the target receiving device is determined. Specifically, the sum of the intermediate noise-reduced signal and the noise at the target receiving device is the noise reduction error signal measured at the target receiving device. Therefore, the noise at the target receiving device can be obtained by subtracting the noise reduction error signal measured at the target receiving device from the intermediate noise-reduced signal.

[0113] The intermediate noise-reduced signal, which is the theoretical noise-reduced signal reaching the target receiving device, is determined using a second virtual second-order filter. This second virtual second-order filter is used to simulate the signal changes along the physical path of the theoretical noise-reduced signal from the transmitter to the target receiving device.

[0114] For example, during the experiment, white noise is used as the excitation source to test the signal change of white noise from the loudspeaker to the microphone, determining the transfer function of the physical path between the loudspeaker and the microphone. The transfer function (path) of the physical path between the loudspeaker and the microphone can be simulated by a second virtual second-order filter. Optionally, the second virtual second-order filter and the first virtual second-order filter can be filters of the same type. The process of the second virtual second-order filter simulating the physical path between the loudspeaker and the microphone is consistent with the process of the first virtual second-order filter simulating the secondary path.

[0115] Optionally, when determining the physical path between the loudspeaker and the microphone, either white noise or a single-frequency sine wave signal can be used as the excitation source. For example: Figure 5 As shown, during the training phase, white noise is used to excite the physical acoustic path. And based on the normalized least mean square (NLMS) algorithm, Perform a global model update, the update formula is expressed as:

[0116]

[0117] in, Indicates the updated , Indicates the global modeling step size. This represents the white noise tap vector. Indicates the global modeling order. Indicates white noise excitation respectively and error .

[0118] Then, The input data post-processing system uses a single-frequency signal at the target frequency. The local modeling update is performed using the excitation signal to obtain... The update formula is expressed as:

[0119]

[0120]

[0121] in, Indicates the updated , Indicates a single-frequency signal , Indicates the first Local modeling step size for each frequency, express In the Modeling error at each frequency component This indicates the order of the local modeling. For a single target frequency, the above modeling and fitting process can be achieved using a second-order FIR filter (a second virtual second-order filter).

[0122] In some embodiments, the parameters of the second virtual second-order filter are determined using a single-frequency sinusoidal signal as the excitation source.

[0123] For example, to avoid the increased cost and complexity caused by full-frequency point-by-point testing, multiple fixed-frequency single-frequency sinusoidal signals are selected as excitation sources for testing. The response of the physical path between the loudspeaker and the microphone to the signal at a fixed frequency is determined. Then, the response of the physical path between the loudspeaker and the microphone to the signal in the target frequency range is determined by interpolation calculation. In this way, the coefficients of the higher-order filter (reference filter) corresponding to the second virtual second-order filter are determined, and finally the parameters of the second virtual second-order filter are obtained.

[0124] In some embodiments, the noise at the target noise reduction location is determined based on the noise at the target receiving device location, and the noise reduction error signal at the target noise reduction location is determined based on the noise at the target noise reduction location and the signal change of the theoretical noise reduction signal simulated by the first virtual second-order filter on the secondary path. Specifically, the noise change relationship between the target receiving device and the target noise reduction location is obtained through pre-simulation, and the noise at the target noise reduction location is determined based on the noise at the target receiving device and the noise change relationship.

[0125] For example, during the experiment, using white noise or a frequency-sweeping radiation signal from the engine ranging from low to high speed as the excitation source, the signal change of vehicle engine noise from the microphone to the listener's ear is tested to determine the transfer function of the physical path between the microphone and the listener's ear. This transfer function (path) can be simulated by a third virtual second-order filter. Optionally, the third virtual second-order filter and the first virtual second-order filter can be of the same type. The process of the third virtual second-order filter simulating the physical path between the microphone and the listener's ear is consistent with the process of the first virtual second-order filter simulating the secondary path. Specifically, in the experimental phase, such as... Figure 10 As shown, four virtual microphones 140, each installed on the ears of four passengers in the vehicle, acquire the signal of the vehicle engine noise after it travels from the microphone to the ear. Based on the engine noise and the signal after it travels from the microphone to the ear, the transfer function of the physical path between the microphone and the ear is determined. In practice, without the virtual microphones 140, the noise at the target noise reduction location is determined based on a pre-simulated noise variation relationship between the target sound receiving device and the target noise reduction location. This application does not limit the number, type, or placement of the virtual microphones 140.

[0126] Optionally, when determining the transfer function of the physical path between the microphone and the ear, white noise, a frequency-sweeping radiation signal from an engine ranging from low to high speed, or a single-frequency sinusoidal signal can be used as the excitation source. For example: Figure 11 As shown, during the training phase, the results are first obtained through global modeling and updating. Then The input data post-processing system uses a single-frequency signal at the target frequency. The local modeling update is performed using the excitation signal to obtain... The formula is expressed as:

[0127]

[0128]

[0129]

[0130] in, Indicates the global modeling order. Indicates the updated , Indicates a single-frequency signal , Indicates the first Local modeling step size for each frequency, express In the Modeling error at each frequency component This indicates the order of the local modeling. For a single target frequency, the above modeling and fitting process can be achieved using a second-order FIR filter (a third virtual second-order filter).

[0131] In some embodiments, the parameters of the third virtual second-order filter are determined using a single-frequency sinusoidal signal as the excitation source.

[0132] For example, to avoid the increased cost and complexity caused by full-frequency point-by-point testing, multiple fixed-frequency single-frequency sinusoidal signals are selected as excitation sources for testing. The physical path between the microphone and the ear position is determined for the response of the signal at a fixed frequency. Then, the physical path between the microphone and the ear position is determined for the response of the signal in the target frequency range through interpolation calculation. The coefficients of the higher-order filter (reference filter) corresponding to the third virtual second-order filter are then determined, and finally the parameters of the third virtual second-order filter are obtained.

[0133] In some embodiments, when reducing engine noise, since the engine speed varies under different operating conditions, the simulation of the first, second, and third virtual second-order filters needs to cover the frequency range included in the engine noise. For example, with 1Hz intervals, the frequency range of 20-200Hz requires 180 second-order filters. Therefore, 180 first, second, and third virtual second-order filters are needed to perform equivalent simulations of the secondary path, the physical path from the speaker to the microphone, and the physical path from the microphone to the ear position, respectively.

[0134] Step 3: Optimize the theoretical noise reduction signal based on the noise reduction error signal and the corrected narrowband noise signal to obtain the target noise reduction signal.

[0135] For example, the weight coefficients of the adaptive filter are optimized based on the noise reduction error signal and the corrected narrowband noise signal, and the optimized adaptive filter is used to generate the target noise reduction signal.

[0136] For example, such as Figure 12 As shown, the signal generator acquires the engine speed via the CAN bus, determines the noise frequency based on the mapping relationship between engine speed and engine noise frequency, and then generates a sinusoidal signal with the same frequency as the noise. Sum and cosine signals Then, through the coefficients of the controller (adaptive filter) and For sinusoidal signals Sum and cosine signals Adjustments were made to obtain the noise-reduced signal. Noise reduction signal The formula is expressed as:

[0137] Among them, calculation The required number of multiplications is 2IJ. This represents the j-th speaker, where J represents the number of speakers. Let I represent the i-th frequency component, and let I represent the number of frequencies.

[0138] Speaker output noise reduction signal Then, the noise-reduced signal passes through the physical acoustic path The noise reduction signal travels through the physical acoustic path (between the loudspeaker and the microphone) to reach the microphone. The signal is then converted into an intermediate noise-reduced signal. The sum of the intermediate noise-reduced signal and the noise at the microphone location (estimated desired noise) is the noise reduction error signal detected by the microphone. Therefore, the noise at the microphone location can be determined based on the noise reduction error signal detected by the microphone and the intermediate noise-reduced signal. Noise at the microphone location This can be expressed as the following formula:

[0139] in, This indicates the noise level at the microphone's location. This represents the noise reduction error signal detected by the microphone, where k represents the k-th microphone. Indicates the noise reduction signal Through physical acoustic pathways The intermediate noise-reduced signal after processing. Calculation. The required number of multiplications is IJK K represents the number of microphones.

[0140] The noise at the microphone location was calculated. Then, based on the noise at the microphone location and virtual acoustic path (The physical path between the microphone and the ear position) determines the noise at the ear position (target noise reduction position). Noise at the ear position This can be expressed as the following formula:

[0141] Among them, calculation The required number of multiplications is IK. , Indicates the first V represents the number of target noise reduction locations (the locations of virtual microphones).

[0142] Next, based on the noise-reduced signal passing through the secondary path The signal that reaches the ear (the physical path between the speaker and the ear) and the noise at the ear's location. Determine the noise reduction error signal at the location of the human ear. Noise reduction error signal at the human ear position This can be expressed as the following formula:

[0143] in, This represents the noise reduction error signal at the location of the human ear. This indicates the noise level at the location of the human ear. Indicates the noise reduction signal via secondary path The signal after. Calculation. The number of multiplications required is IJ .

[0144] Simultaneously, through the first virtual second-order filter (simulating the secondary path) For sinusoidal signals Sum and cosine signals Filtering is performed based on the filtered signal and the noise reduction error signal at the ear position. For the coefficients of the adaptive filter and Adjustments (using the FxLMS algorithm) are made to modify the noise-reduced signal generated by the adaptive filter, thereby improving the noise reduction effect.

[0145] For example, the first virtual second-order filter for sinusoidal signals Sum and cosine signals The formula for filtering is expressed as:

[0146]

[0147] in, This represents the filtered cosine signal. This represents the filtered sine wave signal. This represents the cosine signal before filtering. This represents the sinusoidal signal before filtering. This represents the secondary path. The number of multiplications required for the filtering process is 2IJV. .

[0148] The update formula for the adaptive filter is expressed as:

[0149]

[0150] in, Indicates the updated , Indicates the updated , This represents the noise reduction error signal at the location of the human ear. This represents the cosine signal after being filtered by the first virtual second-order filter. This represents the sinusoidal signal after being filtered by the first virtual second-order filter. The number of multiplications required for the update process is 2IJ(V+1).

[0151] Step S204: The noise reduction device uses the target noise reduction signal to perform noise reduction.

[0152] For example, a target noise reduction signal is emitted through a speaker so that the target noise reduction signal cancels out the noise at the target noise reduction location.

[0153] This application also provides a noise reduction device, such as Figure 13 As shown in the figure, a noise reduction device provided in this application embodiment may include a processing module 1301 and a noise reduction module 1302. The processing module 1301 is used to perform... Figure 2 In the illustrated method, steps S201, S202, and S203 are performed by the noise reduction module 1302. Figure 2 The illustrated method includes step S204.

[0154] The technical solutions provided in this application, as shown in Table 1, reduce the number of multiplications compared to conventional active noise reduction algorithms. This is because the noise signal at frequency f of a conventional active noise reduction algorithm is compared with... The convolution operation of a second-order broadband secondary path is equivalent to the noise signal at that frequency and the second-order ( The convolution operation is performed using a multiplication operator (=2). This application's technical solution reduces the number of multiplications when calculating the convolution of physical acoustic paths. Conventional active noise reduction algorithms perform JK times the noise reduction signal. The convolution operation of the physical acoustic path is different from that of the previous method, which only requires JK second-order (JK-th order) convolution operations on the noise-reduced signals of I frequencies. =2) Convolution operation of physical acoustic paths. Similarly, the technical solution of this application reduces the number of multiplications when calculating the convolution of secondary paths. The number of multiplications is reduced by a factor of two when calculating the convolution of virtual acoustic paths. times.

[0155] Taking a single-frequency, single-channel broadband acoustic path system of order 32 as an example, the conventional active noise cancellation algorithm requires 166 multiplications per update, while the proposed solution requires only 16 multiplications, approximately 10 times less. Taking a dual-frequency, four-channel broadband acoustic path system of order 128 as an example, the proposed solution requires 416 multiplications per update, approximately 35 times less than the conventional active noise cancellation algorithm's 14432 multiplications. Therefore, the proposed solution significantly reduces computational complexity compared to conventional active noise cancellation algorithms when dealing with high-order broadband acoustic paths and a large number of system channels.

[0156] Table 1

[0157] Among them, in this application , and Both are 2. Conventional active noise reduction algorithms use filters with an order greater than two (high-order filters). , and All values ​​are higher than 2. The technical solution proposed in this application has lower computational complexity compared to conventional active noise reduction algorithms.

[0158] For example, such as Figure 14 As shown, the number of multiplication operations is illustrated for conventional active noise reduction algorithms and the technical solution of this application when the number of frequencies I ranges from 1 to 10. Figure 14 The horizontal axis represents the frequency quantity I, and the vertical axis represents the number of multiplication operations (unit: ×10⁴). Figure 14 It is known that the number of multiplication operations in conventional active noise reduction algorithms increases approximately linearly with the increase of I, rising from approximately 1 × 10⁴ times when I = 1 to approximately 4.7 × 10⁴ times when I = 10. However, the number of multiplication operations in the technical solution of this application remains at an extremely low level close to 0, with only a very slight increase as I increases. For example, as... Figure 15 The diagram illustrates the number of multiplication operations required for conventional active noise cancellation algorithms and the present application's technical solution when the secondary path order, the physical path order between the speaker and microphone, and the physical path order between the microphone and the ear position change (e.g., the order of the reference filter). The horizontal axis represents the filter order L, and the vertical axis represents the number of multiplication operations (unit: ×10⁴). Figure 15 It is known that the number of multiplication operations in conventional active noise reduction algorithms increases approximately linearly with the increase of L, rising from nearly 0 when L=0 to approximately 2.1×10⁴ when L=250; while the number of multiplication operations in the technical solution of this application remains at an extremely low level close to 0, with only a very slight increase as L increases. Therefore, the technical solution of this application has a significant performance advantage in reducing the number of multiplication operations and controlling computational complexity.

[0159] In some embodiments, to verify the effectiveness of the technical solution of this application in actual application scenarios, a secondary path of length 128, the physical path between the speaker and the microphone, and the physical path between the microphone and the ear position in a real vehicle test are used to simulate and analyze the noise reduction performance of the established conventional active algorithm and the technical solution of this application in the target noise under the low-end working condition of vehicle fuel replenishment.

[0160] For example, such as Figure 16 As shown, during the observation period of 1-10 seconds, under the low-gear condition of vehicle fuel replenishment, the engine speed generally fluctuated within the range of approximately 1230-1248 r / min, with an average speed of approximately 1240 r / min. No obvious continuous upward or downward trend was observed, indicating that the engine speed was relatively stable under this condition. This data segment can be considered as a process of small fluctuations in engine speed superimposed near the stable operating condition, providing a stable speed reference for subsequent line spectrum frequency extraction and noise reduction control. Figure 17 As shown, the signal amplitude is mainly distributed between approximately ±0.03, with no obvious trend or abrupt change, indicating that the signal has good stability during this time period, and the overall time-domain waveform of the signal exhibits stable periodic oscillations. Correspondingly, as... Figure 18 As shown, the frequency domain amplitude spectrum exhibits a significant single main peak in the 20-200Hz range, with the main frequency concentrated around approximately 41Hz and 82Hz. The amplitudes of other frequency components are low, with only extremely weak background spectral lines present. Therefore, the signal energy is highly concentrated at multiple frequency points, exhibiting typical line spectrum (narrowband) characteristics. In the time domain, it shows an approximate single-frequency sinusoidal oscillation, while in the frequency domain, the energy is dominated by the main frequency and harmonic frequencies. Figure 18 It can be seen that the noise signal emitted by the engine under the low-gear condition of fuel charging is mainly the 2nd harmonic and the 4th harmonic. The key to noise reduction is whether the frequency components of the corresponding harmonics can be effectively reduced.

[0161] After noise reduction, such as Figure 19 As shown, under low-power fuel-charging conditions, the spectrum before noise reduction exhibits significant line peaks. After noise reduction using both conventional active noise reduction algorithms and the proposed solution, the line peaks are significantly suppressed. Furthermore, the curves corresponding to the two noise reduction methods largely overlap in terms of peak position, suppression depth, and noise floor variation trend, indicating that both methods achieve consistent noise reduction and frequency domain suppression effects under low-power fuel-charging conditions. Further, as... Figure 20 and Figure 21 As can be seen from the mean squared error (MSE) curves of the error signals corresponding to the two noise reduction methods, in the initial stage, the MSE curves of the two noise reduction methods both drop rapidly and enter the stable fluctuation range. The steady-state levels are similar and there is no divergence, indicating that the convergence speed and steady-state performance of the two are basically equivalent.

[0162] In summary, the technical solution of this application significantly reduces the number of multiplication operations through local second-order equivalent modeling, but the denoising effect and convergence characteristics are not reduced as a result. It can effectively reduce the computational burden while maintaining the same control and denoising performance as conventional active denoising algorithms, making it more suitable for controller platforms with high real-time requirements or limited computing power.

[0163] The foregoing mainly describes the solutions provided by the embodiments of this application from the perspective of methods and apparatus. To achieve the above functions, the noise reduction device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0164] This application embodiment can, based on the above-described noise reduction method, exemplarily divide the noise reduction device into functional modules. For example, the noise reduction device may include functional modules corresponding to each functional division, or two or more functions may be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods.

[0165] This application also provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the noise reduction method provided in the above-described method embodiments.

[0166] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a read-only memory (ROM), random access memory (RAM), magnetic tape, floppy disk, and optical data storage device.

[0167] This application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, they implement the noise reduction method provided in the above-described method embodiments.

[0168] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the processor of a computing device, they implement the various processes of the above-described method embodiments and achieve the same technical effects as the above-described methods. To avoid repetition, they will not be described again here.

[0169] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0170] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0171] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0172] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0173] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Those skilled in the art can understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this application still fall within the scope of this application.

Claims

1. A method of noise reduction, characterized by, The method includes: Determine the theoretical noise reduction signal corresponding to the narrowband noise signal; Based on the noise frequency and target mapping relationship of the narrowband noise signal, a first virtual second-order filter is determined; wherein, the target mapping relationship is used to characterize the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the simulated performance of the first virtual second-order filter and the reference filter is consistent with respect to the signal change of the theoretical noise-reduced signal on the secondary path; the secondary path is the path from the transmitting end of the theoretical noise-reduced signal to the target noise-reducing position; the order of the reference filter is greater than 2; Using the first virtual second-order filter, a target noise reduction signal for the narrowband noise signal is determined; wherein, the target noise reduction signal is used to eliminate noise at the target noise reduction location; Noise reduction is performed using the target noise reduction signal.

2. The noise reduction method of claim 1, wherein, The step of determining the target noise-reduced signal of the narrowband noise signal using the first virtual second-order filter includes: The narrowband noise signal is corrected using the first virtual second-order filter to obtain a corrected narrowband noise signal; wherein the correction operation is used to eliminate the influence of the secondary path on noise reduction. Determine the noise reduction error signal at the target noise reduction location; wherein, the noise reduction error signal is the residual sound pressure signal after the theoretical noise reduction signal cancels out the narrowband noise signal; Based on the noise reduction error signal and the corrected narrowband noise signal, the theoretical noise reduction signal is optimized to obtain the target noise reduction signal.

3. The noise reduction method according to claim 2, characterized in that, The target noise reduction location is the position of the human ear; determining the noise reduction error signal at the target noise reduction location includes: Determine the noise level at the target sound receiving device; wherein the distance between the location of the target sound receiving device and the target noise reduction location is less than a first preset distance; Based on the noise at the target sound receiving device, the noise reduction error signal at the target noise reduction location is determined.

4. The noise reduction method according to claim 3, characterized in that, Determining the noise at the target receiving device includes: Determine the intermediate noise reduction signal from the theoretical noise reduction signal to the target sound receiving device; The noise at the target sound receiving device is determined based on the noise reduction error signal measured at the target sound receiving device and the intermediate noise reduction signal.

5. The noise reduction method according to claim 3, characterized in that, The step of determining the noise reduction error signal at the target noise reduction location based on the noise at the target sound receiving device location includes: Based on the noise at the location of the target sound receiving device, determine the noise at the target noise reduction location; Based on the noise at the target noise reduction location and the signal change of the theoretical noise reduction signal simulated by the first virtual second-order filter on the secondary path, the noise reduction error signal at the target noise reduction location is determined.

6. The noise reduction method according to claim 5, characterized in that, Determining the noise at the target noise reduction location based on the noise at the target sound receiving device location includes: Obtain the pre-simulated noise variation relationship between the target sound receiving device and the target noise reduction location; Based on the noise at the target sound receiving device and the relationship between the noise changes, the noise at the target noise reduction location is determined.

7. The noise reduction method according to any one of claims 1-6, characterized in that, The process of determining the target mapping relationship includes: The filtered output signal of the reference filter on the input signal is represented as a sum of a first sine signal and a first cosine signal; wherein the frequency of the input signal is the noise frequency; and both the first sine signal and the first cosine signal are related to the order coefficients of the reference filter. The filtered output signal of the first virtual second-order filter on the input signal is represented as a sum of a second sine signal and a second cosine signal; both the second sine signal and the second cosine signal are related to the order coefficients of the first virtual second-order filter. Solve the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the preset conditions are met; the preset conditions are: the first sine signal is equal to the second sine signal and the first cosine signal is equal to the second cosine signal. The target mapping relationship is determined using the obtained correspondence.

8. The noise reduction method according to any one of claims 1-6, characterized in that, The process of determining the reference filter includes: Using multiple single-frequency sinusoidal signals of fixed frequencies as excitation sources, the phase response and amplitude response of the reference filter at the fixed frequencies are determined respectively; wherein, the fixed frequencies are multiple frequencies determined at fixed frequency intervals within a target frequency range, and the target frequency range is the frequency range to which the narrowband noise signal belongs; The phase response and amplitude response of the reference filter in the target frequency range are calculated based on the phase response and amplitude response of the reference filter at the fixed frequency. The coefficients of the reference filter are fitted based on the phase response and amplitude response of the reference filter within the target frequency range.

9. The noise reduction method according to claim 2, characterized in that, The determination of the theoretical noise reduction signal corresponding to the narrowband noise signal includes: The narrowband noise signal is input into an adaptive filter to obtain the theoretical noise reduction signal; wherein, the theoretical noise reduction signal is a signal with opposite phase and equal amplitude to the narrowband noise signal.

10. The noise reduction method according to claim 9, characterized in that, The step of optimizing the theoretical denoising signal based on the denoising error signal and the corrected narrowband noise signal to obtain the target denoising signal includes: Based on the noise reduction error signal and the corrected narrowband noise signal, the weight coefficients of the adaptive filter are optimized; The target noise-reduced signal is generated using an optimized adaptive filter.

11. A noise reduction device, characterized in that, include: The processing module is used to determine the theoretical noise reduction signal corresponding to the narrowband noise signal; The processing module is further configured to determine a first virtual second-order filter based on the noise frequency and target mapping relationship of the narrowband noise signal; The processing module is further configured to use the first virtual second-order filter to determine the target noise reduction signal of the narrowband noise signal; A noise reduction module is used to perform noise reduction using the target noise reduction signal; Wherein, the target mapping relationship is used to characterize the correspondence between the order coefficients of the first virtual second-order filter and the order coefficients of the reference filter when the simulation performance of the first virtual second-order filter and the reference filter is consistent with respect to the signal change of the theoretical noise-reduced signal on the secondary path; the secondary path is the path between the transmitting end of the theoretical noise-reduced signal and the target noise-reducing position; the order of the reference filter is greater than 2; the target noise-reducing signal is used to eliminate noise at the target noise-reducing position.

12. A vehicle, characterized in that, The vehicle includes the noise reduction device as described in claim 11.

Images