Method for guiding automobile noise reduction optimization based on automobile road noise sample preparation

Abstract

The invention provides a method for guiding automobile noise reduction optimization based on automobile road noise sample preparation, and belongs to the field of automobile NVH performance development. The method comprises the following steps: acquiring in-vehicle road noise data when an automobile to be subjected to noise reduction optimization runs on different road surfaces at various constant speeds respectively; determining a time-invariant FIR (Finite Impulse Response) filter according to a to-be-edited frequency band of the in-vehicle road noise data and a sound pressure level change demand of the frequency band, and performing frequency spectrum editing on the in-vehicle road noise data by utilizing the time-invariant FIR filter; performing multi-loudness-level equal loudness equalization processing on the in-vehicle road noise data and the in-vehicle road noise data after spectrum editing to obtain an in-vehicle road noise data sample; performing high-fidelity playback on the in-vehicle road noise data sample and the in-vehicle road noise data, and then performing evaluation to obtain an evaluation score; and optimizing the noise reduction system of the automobile based on the evaluation score. According to the invention, abundant road noise sound samples are prepared, and the workload of sound sample preparation is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of automotive NVH performance development, and particularly relates to a method for preparing automotive road noise samples to guide automotive noise reduction and optimization. Background Art

[0002] With the increasing requirements of the automotive market for comfort and noise control, the road noise quality of automobiles has become an important criterion for users to evaluate vehicles. Road noise not only affects driving comfort, but may also have a negative impact on the driver's attention and the passengers' auditory experience. Therefore, it is particularly important to evaluate the road noise quality of automobiles during the early prototype testing stage. Through scientific evaluation methods, the noise performance of vehicles under different road conditions can be comprehensively understood, providing a basis for subsequent improvements.

[0003] The evaluation of road noise quality involves multiple aspects, including psychoacoustic parameters such as the loudness, sharpness, and roughness of the noise. These parameters can quantify the evaluation of the noise, helping engineers more accurately identify the noise sources and formulate improvement measures. For example, loudness reflects the intensity of the sound, sharpness describes the harshness of the high-frequency components, and roughness is related to low-frequency modulated noise. By measuring and analyzing these parameters, the road noise quality can be evaluated more comprehensively.

[0004] Based on the evaluation results, engineers can take various measures to improve the road noise quality. For example, optimizing the tire design to reduce pattern noise and cavity noise, adjusting the modal frequency of the suspension system to avoid resonance with the body structure, and adding sound insulation materials at key parts of the body to reduce noise propagation. In addition, through modal optimization and acoustic package design, the generation and propagation of road noise can be further reduced.

[0005] Through these improvement measures, the road noise quality of vehicles can be effectively improved, thus obtaining automobiles with a better user experience. This not only helps to improve the competitiveness of vehicles, but also provides a more comfortable and quiet driving environment for users.

[0006] However, in the existing automotive road noise sample preparation technologies, the commonly used method is to collect road noise samples of a vehicle under different driving speed conditions and on different test road surfaces. The vehicle speed conditions can be divided into constant speed driving at different speeds, full-throttle acceleration driving, and engine idle speed conditions when the vehicle is stationary. Professional automotive test sites include various test road surfaces such as rough asphalt, smooth asphalt, large-grained asphalt, grooved cement, and stone roads. As a result, a large number of road noise samples are obtained. Combining different speeds and different test road surfaces into a large number of test conditions and collecting road noise samples will require a large amount of work, and the usage cost of the test site is also quite expensive.

[0007] Moreover, during the acquisition process of road noise sound samples, generally based on different vehicle speeds or different road surfaces, simply through the acquisition of in-vehicle road noise sound samples, it may be somewhat challenging to make the noise samples have a strong mutual correlation with the evaluation indicators or exhibit good distribution characteristics. SUMMARY OF THE INVENTION

[0008] To solve the problems in the prior art, the present invention provides a method for guiding vehicle noise reduction optimization based on automotive road noise samples.

[0009] The technical solution of the present invention is as follows:

[0010] The present invention discloses a method for guiding vehicle noise reduction optimization based on automotive road noise samples, comprising the following steps:

[0011] 1) Collect in-vehicle road noise data when the vehicle to be optimized for noise reduction is driving at a constant speed at multiple speeds on different road surfaces;

[0012] 2) Determine a time-invariant FIR filter according to the frequency band that needs to be edited in the in-vehicle road noise data and the sound pressure level change requirement of this frequency band, and then use the time-invariant FIR filter to perform spectral editing on the in-vehicle road noise data to obtain the in-vehicle road noise data after spectral editing;

[0013] 3) Perform equal-loudness equalization processing on the in-vehicle road noise data and the in-vehicle road noise data after spectral editing at multiple loudness levels to obtain in-vehicle road noise data samples;

[0014] 4) Perform high-fidelity playback on the in-vehicle road noise data samples obtained in step 3) and the in-vehicle road noise data collected in step 1), and use comfort, single-frequency feeling, jitter feeling, and ear-pressure feeling as evaluation indicators. Based on the paired comparison method and the graded scoring method with semantic assistance, evaluate the in-vehicle road noise data samples and the in-vehicle road noise data to obtain evaluation scores;

[0015] 5) Optimize the noise reduction system of the vehicle based on the evaluation scores.

[0016] Compared with the prior art, the specific beneficial effects of the present invention are:

[0017] (1) The present invention aims at a method for preparing road noise samples of a vehicle to guide vehicle noise reduction optimization, and proposes a method for preparing road noise samples. By means of spectral editing and equal loudness equalization, rich road noise samples can be prepared, the workload of sound sample preparation can be reduced, and the noise samples can have a strong correlation with the evaluation index or exhibit good distribution characteristics, solving the problems existing in the prior art and also solving the problems of large workload and high usage cost of the test site in the prior art. The sample preparation is not only obtained by collecting on a real vehicle, but also can generate road noise samples with certain characteristics through spectral editing.

[0018] (2) The present invention conducts an evaluation on the original samples and edited samples through specific spectral editing in combination with the paired comparison method, and deeply analyzes the influence mechanism of different time-frequency characteristics of the sound samples on the road noise evaluation index; explores the sound quality differences at different loudness levels through equalization of different loudness levels, and explores the sound quality differences of the spectral characteristics of different road noise samples under the equalization of the same loudness level.

[0019] (3) The present invention proposes four road noise quality evaluation indexes, which can comprehensively evaluate the road noise samples, analyze the relationship between different characteristics of the road noise samples and the road noise evaluation index, and provide data support for the subsequent optimization of the road noise quality through different evaluation indexes.

[0020] (4) The method for preparing road noise samples of the present invention is not only applicable to vehicle noise reduction optimization, but also applicable to the preparation of samples in the noise optimization of the sound quality of other sound samples. Description of the Drawings

[0021] Figure 1 is a schematic flow chart of the method for preparing road noise samples of a vehicle to guide vehicle noise reduction optimization described in the embodiment of the present invention;

[0022] Figure 2 is a comparison of the time-frequency diagrams of road noise samples before and after spectral editing in the embodiment of the present invention;

[0023] Figure 3 is the spectral characteristics of sound samples at different loudness levels in the embodiment of the present invention;

[0024] Figure 4 is a radar chart of the evaluation index scores of different edited samples in the embodiment of the present invention. Detailed Embodiments

[0025] The following further elaborates and explains the present invention in combination with specific embodiments. The described embodiments are only demonstrations of the present disclosure content and do not delimit the scope of limitation. The technical features of each embodiment of the present invention can be combined correspondingly without conflict.

[0026] In the process of evaluating road noise quality, the preparation of noise samples is very important. How to obtain high-quality sound samples that can truly reflect the acoustic attributes of products and are minimally affected by the external environment is an important factor in the success of evaluation tests. The collection of noise samples and the selection of evaluation indicators should also be interrelated. For specific noise samples, different selected evaluation indicators should be able to reflect the quality characteristics that are most concerned about the noise source in engineering applications; similarly, for determined evaluation indicators, the selection of noise samples should also have corresponding distribution characteristics.

[0027] However, in the existing technologies for preparing evaluation samples of automotive road noise quality, the commonly used method is to collect road noise samples of a vehicle under different driving speed conditions and on different test road surfaces. The vehicle speed conditions can be divided into constant-speed driving at different speeds, full-throttle acceleration driving, and engine idle speed conditions when the vehicle is stationary. Professional automotive test sites include various test road surfaces such as rough asphalt, smooth asphalt, large-grained asphalt, grooved cement, and stone roads. A large number of road noise sound samples are thus obtained. Combining different speeds and different test road surfaces into a large number of test conditions, collecting road noise sound samples will require a large amount of work, and the usage cost of the test site is also quite expensive.

[0028] Moreover, during the collection process of road noise sound samples, generally based on different vehicle speeds or different road surfaces, just by collecting real-vehicle road noise sound samples, it may be somewhat challenging to make the noise samples and evaluation indicators have a strong mutual correlation or present good distribution characteristics.

[0029] In summary, in the research on improving vehicle road noise quality, it is necessary to develop a method for guiding vehicle noise reduction optimization based on the preparation of automotive road noise samples, including the method for preparing automotive road noise quality samples and their related evaluation indicators, thereby reducing the workload of evaluation, more intuitively obtaining the relationship between road noise spectrum and sound quality, and providing data support for the optimization research of in-vehicle road noise quality.

[0030] To achieve the above object, as Figure 1 shown, the present invention provides a method for guiding vehicle noise reduction optimization based on the preparation of automotive road noise samples, including the following steps:

[0031] 1) Collect in-vehicle road noise data of the vehicle to be optimized for noise reduction when driving at a constant speed at various speeds on different road surfaces;

[0032] 2) Determine a time-invariant FIR filter according to the frequency band that the in-vehicle road noise data needs to be edited and the sound pressure level change requirement of this frequency band, and then use the time-invariant FIR filter to perform spectrum editing on the in-vehicle road noise data to obtain the in-vehicle road noise data after spectrum editing;

[0033] 3) Perform equal-loudness equalization processing on the in-vehicle road noise data and the edited in-vehicle road noise data at multiple loudness levels to obtain rich in-vehicle road noise data samples;

[0034] 4) Perform high-fidelity playback on the in-vehicle road noise data samples obtained in step 3) and the in-vehicle road noise data collected in step 1). Taking comfort, single-frequency sense, jitter sense, and ear-pressure sense as evaluation indicators, evaluate different in-vehicle road noise data samples and in-vehicle road noise data based on the paired comparison method and the ranking scoring method with semantic assistance to obtain evaluation scores;

[0035] 5) Optimize the noise reduction system of the vehicle based on the evaluation scores to improve the road noise quality of the vehicle.

[0036] In a specific embodiment of the present invention, step 1) is to perform multiple vehicle speed steady-state condition tests on rough asphalt roads, smooth asphalt roads, grooved cement roads, etc. in a professional test field, and use a professional artificial head to collect in-vehicle road noise data. The professional artificial heads are uniformly arranged at the right rear seat of the vehicle.

[0037] Furthermore, filtering by a time-invariant FIR filter can filter out sound components, add constant frequency components to the sound, attenuate or amplify the sound over time, and add order noise and other functions, and can freely perform editing and design on the sound samples in the time-frequency domain.

[0038] Therefore, in step 2), first determine the frequency-domain response of the time-invariant ideal filter according to the frequency band to be edited of the in-vehicle road noise data and the sound pressure level change requirement of this frequency band. The frequency-domain response is:

[0039]

[0040] In the formula, H d (e jω ) is the frequency-domain response of the time-invariant ideal filter; ΔL p is the sound pressure level change; f is the frequency within the frequency band to be edited of the in-vehicle road noise data, f s is the sampling frequency of the in-vehicle road noise data; ω1 is the normalized digital angular frequency corresponding to the lower cut-off frequency of the passband of the ideal filter, f1 is the lower cut-off frequency of the passband of the ideal filter, that is, the minimum frequency within the frequency band to be edited of the in-vehicle road noise data; ω2 is the normalized digital angular frequency corresponding to the upper cut-off frequency of the passband of the ideal filter, f2 is the upper cut-off frequency of the ideal filter, which is the maximum frequency within the frequency band where the in-vehicle road noise data needs to be edited; τ is the group delay, representing the time delay of the time-invariant ideal filter. When τ = (N - 1) / 2, the filter has a linear phase characteristic, which means that the filter has the same time delay for all frequency components, helping to avoid signal distortion; N is the number of points after truncating the infinite length h d (n) of the ideal filter, which is the order of the filter.

[0041] Then, perform the inverse discrete-time Fourier transform on the frequency response H d (e jω ) of the time-invariant ideal filter to obtain the unit impulse response. The formula for performing the inverse discrete-time Fourier transform to obtain the unit impulse response is:

[0042]

[0043] In the formula, h d (n) is the unit impulse response of the time-invariant ideal filter; n is the discrete-time variable, representing the time sequence number of the time-invariant ideal filter, and n takes integer values.

[0044] Therefore, its unit impulse response can be calculated as:

[0045]

[0046] Furthermore, since h d (n) is an infinite-length even-symmetric sequence, it is necessary to select an even-symmetric finite-length (N-point length) window function ω(n) to truncate h d (n) to obtain the required linear-phase FIR filter:

[0047] h(n) = h d (n)ω(n)

[0048] In the formula, h(n) is the time-invariant FIR filter; ω(n) is the window function, where 0 ≤ n ≤ N - 1, which is determined by the given change ΔL p of the sound pressure level in this frequency band. The window function is selected according to the change of the sound pressure level and the minimum attenuation of the stopband of different window functions. The types of window functions include rectangular window, triangular window, Hanning window, Hamming window, Blackman window, Kaiser window, etc.; N is inversely proportional to the transition bandwidth Δω, and the proportionality coefficients of different-shaped windows are different. The transition bandwidth refers to the frequency range where the signal transitions from the passband to the stopband or from the stopband to the passband in the frequency response of the filter. When the transition bandwidth is smaller, the order of the filter is higher, and the transition between the passband and the stopband can be controlled more precisely. The value of the transition bandwidth is mainly determined according to the performance requirements of the filter.

[0049] Furthermore, ΔLp represents the change in sound pressure level of this frequency band, and by changing the magnitude of ΔL p the magnitude of can increase or decrease the sound pressure level within each frequency band, and then design a finite-length time-invariant FIR filter h(n) according to the above steps. Finally, perform convolution calculation on the time-domain sampling data of the road noise sound sample to be designed and the FIR filter h(n), and the spectrum design of the road noise sound sample can be completed.

[0050] In step 3, by setting different target loudness levels, equal-loudness equalization processing can scale different samples proportionally up or down to the target loudness level, making them have the same loudness level.

[0051] Equal-loudness equalization is an audio processing technology aimed at keeping audio at a consistent loudness level under different loudness levels. The calculation method of equal-loudness equalization is to calculate the loudness level of the input sound sample according to EBU R.128, specify the target loudness level of equal-loudness equalization, and then calculate the difference between the target loudness level and the input loudness level as the gain to make the input sound sample reach the target loudness level, obtaining the time-domain data of the sound sample corresponding to the target loudness level. The specific equal-loudness equalization calculation method is as follows:

[0052]

[0053] In the formula, y(n) target is the in-vehicle road noise data sample corresponding to the target loudness level; y(n) is the in-vehicle road noise data or the in-vehicle road noise data after spectrum editing; L target is the target loudness level, and L is the loudness level corresponding to the in-vehicle road noise data or the loudness level corresponding to the in-vehicle road noise data after spectrum editing.

[0054] Specify the same target loudness level, and convert the road noise sound samples to be evaluated into the same target loudness level through the method of equal-loudness equalization, ensuring that the difference in loudness level in the listening sense is minimized between different audio samples, and avoiding the influence of too large or too small loudness level differences of different road noise sound samples on the evaluation results. At the same time, some representative working condition road noise sound samples can be equal-loudness equalized at different loudness levels to obtain several groups of road noise sound sample data at different loudness levels, and then explore the influence of the loudness level on the sound quality evaluation of the road noise sound samples.

[0055] In step 4, in this experiment, a professional Head LabP2 equalizer is used in combination with Sennheiser HD 600 type headphones for high-fidelity playback of the in-vehicle road noise data samples.

[0056] The evaluation indicators are divided into comfort, single-frequency perception, jitter perception, and ear-pressure perception. Among them, comfort refers to the comprehensive comfort evaluation of road noise data samples; single-frequency perception refers to the perception degree of a prominent single frequency in road noise sound samples; jitter perception refers to the perception degree of the fluctuation intensity of the low-frequency components in road noise data samples; and ear-pressure perception refers to the pressure sensation caused by the low-frequency components in road noise data samples.

[0057] Example 1:

[0058] Please refer to Figures 1 to 4 , the present invention provides a method for guiding the preparation of automotive road noise samples to optimize automotive noise reduction, including the following steps:

[0059] S1 Collection of in-vehicle road noise data: Collect in-vehicle road noise data when the vehicle is driving at a constant speed on different road surfaces at different speeds.

[0060] In this experiment, an artificial head of the HMS IV model from Head Acoustics company is used as the noise collection device. It is made of special materials with acoustic characteristics similar to those of the human ear, and can approximately simulate the reflection and scattering processes of sound waves by structures such as the human head and auricle. The test microphones are installed at the entrances of the left and right ear canals of the artificial head, and can accurately obtain the binaural sound signals that reflect the true hearing of the human ear. In order to obtain the true noise signal at the human ear, the vertical coordinates of the binaural microphones of the artificial head should be located at 0.75 ± 0.05 m above the connection line between the seat surface and the backrest surface, and the horizontal coordinates should be located at the center of the seat.

[0061] The data collection site for this study is a professional automotive test field, which includes various experimental road surfaces and meets the test environment specified in GB / T18697-2002 "Acoustics - Measurement Method for In-vehicle Noise of Motor Vehicles". The test road surfaces include rough asphalt, smooth asphalt, large-grained asphalt, grooved cement, stone roads, and other test road surfaces. During the experiment, the vehicle drives at a constant speed, and the vehicle speed varies from 40 km / h to 90 km / h. The HEAD data collection device is used to collect in-vehicle vibration and noise data, and the collection duration is not less than 10 seconds.

[0062] S2 Spectrum editing: Edit the spectrum of the original signal through a time-invariant FIR filter, filter different frequency bands, set the noise amplification or attenuation level in a specific area on the time-frequency diagram to achieve the editing design of the sound sample, and simulate the change of the spectral characteristics of the sound sample in noise optimization, so as to obtain the in-vehicle road noise data after spectrum editing.

[0063] To explore the influence of different time-frequency characteristics of in-vehicle road noise on the sound quality perception, this paper conducts spectral editing on the sound samples obtained from the real vehicle artificial head test. By filtering with a time-invariant FIR filter, functions such as filtering out sound components, adding constant frequency components to the sound, attenuating or amplifying the sound over time, and adding order noise can be achieved, enabling free editing and design of the sound samples in the time-frequency domain.

[0064] In this experiment, the 50 - 500 Hz frequency band of the sound samples is mainly edited. By increasing or decreasing the sound pressure level in each frequency band, the change in the spectral characteristics of the sound samples during active noise control is simulated. Taking the sound sample under the condition of a 70 km / h vehicle speed on a stone road surface as an example, the sound in different frequency bands can be amplified or attenuated. The spectral editing rules for each edited sample are as follows: Edited sample V1, reduce the peaks at 120 Hz and 220 Hz by 10 dB; Edited sample V2, increase the noise in the 120 - 200 Hz frequency band; Edited sample V3, reduce the noise near 120 Hz and 220 Hz to highlight the single-frequency peak; Edited sample V4, reduce the noise in the 50 - 250 Hz frequency band except for the single-frequency peak at 120 Hz; Edited sample V5, reduce the noise in the 50 - 200 Hz frequency band.

[0065] Taking the edited sample V5 under the condition of a 70 km / h vehicle speed on a stone road surface as an example, the sound pressure level in the range of 50 Hz and 200 Hz can be reduced by 25 dB to highlight the single-frequency peak. The detailed filter design process is described as follows:

[0066] According to the requirements of the filter, the stopband is 50 Hz ≤ f ≤ 200 Hz, and the sound pressure level is -25 dB. According to the normalized digital angular frequency calculation formula:

[0067]

[0068] At this time, ω1 = 50 Hz, ω1 = 200 Hz. Since the sound pressure level of the entire frequency band is reduced by 25 dB, so ΔL p Take -25 dB, and then substitute the above into the frequency domain response expression:

[0069]

[0070] After calculating the ideal frequency domain response H d (e jω ), the inverse Fourier transform of the frequency domain response is taken to obtain its unit sample response:

[0071]

[0072] This filter requires a 25 dB reduction in sound pressure level in the range of 50 Hz to 200 Hz. By querying the window spectrum performance indicators of different window functions and the filter performance indicators after windowing, it is found that the minimum stopband attenuation of the triangular window is 25 dB, which meets the design requirements of this filter. Therefore, the triangular window function is selected to truncate the infinite-length unit impulse response h d (n), and this triangular window function ω(n) is:

[0073]

[0074] After the triangular window is determined, query the relationship between the number of points N of the triangular window length and the transition bandwidth Δω:

[0075]

[0076] Since the value of the transition bandwidth Δω is mainly determined according to the filter performance requirements, the maximum allowable transition bandwidth in the preparation of this road noise sound sample is 0.01π. Considering that the increase in the filter order will lead to an increase in computing power, the maximum value 0.01π is selected as the transition bandwidth in the design of this FIR filter. Then, the number of points N of the filter can be determined according to the relationship between the number of points N of the triangular window length and the transition bandwidth Δω, that is, the order of the filter is determined.

[0077] After the window function type and N are determined, the required triangular window function ω(n) is obtained, and the required FIR filter h(n) can be calculated:

[0078] h(n) = h d (n)ω(n)

[0079] Finally, by performing convolution calculation on the FIR filter h(n) and the in-vehicle road noise data x(n) that needs to be spectrally edited, the in-vehicle road noise data after spectral editing can be obtained.

[0080] Reference Figure 2 , the red box is the frequency band range from 50 Hz to 200 Hz. The left figure is the original road noise sound sample, and the right figure is the road noise sound sample after spectral editing with a 25 dB reduction in this frequency band range by the FIR filter. The color represents the sound pressure level in the time-frequency diagram. It can be clearly seen that the color in the red box changes from the higher sound pressure level red to the lower sound pressure level blue-green.

[0081] According to the given spectral editing requirements, a corresponding FIR filter can be designed to achieve different forms of spectral editing, thereby generating a large number of sound samples to explore the relationship between road noise with different spectral characteristics and sound quality, and more intuitively obtaining the rules therein.

[0082] S3 Equal-loudness Equalization: Perform equal-loudness equalization on sound samples at multiple loudness levels to explore the changes in sound quality of different spectral characteristics at different loudness levels. Equal-loudness equalization is an audio processing technique aimed at keeping the audio at a consistent loudness level across different loudness levels.

[0083] To ensure a more detailed resolution of the evaluation of different time-frequency characteristics in noise, rather than being solely affected by the loudness level, and to explore the differences in sound quality perception of different spectral characteristics at different loudness levels, equal-loudness equalization was performed on the sound samples of some working conditions at multiple loudness levels.

[0084] Taking the sound sample of the rough asphalt pavement at a speed of 40 km / h as an example, for the calculation method of equal-loudness equalization, the loudness level of the input sound sample is calculated according to EBUR.128, the target loudness level of equal-loudness equalization is specified, and then the difference between the target loudness level and the input loudness level is calculated as the gain to make the input sound sample reach the target loudness level, obtaining the time-domain data of the sound sample corresponding to the target loudness level.

[0085] Reference Figure 3 , the red line segment is the spectrogram of the original road noise sound sample. By performing equal-loudness equalization on it, three road noise sound samples with loudness levels L1, L2, and L3 can be generated. Analyzing the spectral characteristics of the sound samples at different loudness levels, it can be seen that the overall spectral characteristics of the road noise sound samples after equal-loudness equalization are roughly the same.

[0086] At the same time, some representative working condition road noise sound samples can be selected for equal-loudness equalization at different loudness levels to obtain several groups of road noise sound sample data at different loudness levels, and then explore the influence of the loudness level on the sound quality evaluation of the road noise sound samples.

[0087] S4 Playback Evaluation: Use an equalizer to perform high-fidelity playback of the road noise sound samples with a headphone and evaluate different samples to obtain evaluation scores.

[0088] When conducting acoustic evaluation, to ensure the authenticity and reliability of the evaluation results, it should be ensured that the played sound is consistent with the sound in the actual test environment. In this study, a professional artificial head was used for sound sample collection and an equalizer and headphones were used for sound playback. During the artificial head test process, acquisition equalization is required. To eliminate distortion, the data collected by the artificial head needs playback equalization during playback to compensate for the effects of acquisition equalization, the secondary filtering effect of the concha cavity, and the pressure of the headphone on the auricle. Precise playback is achieved by defining the corresponding headphone compensation curve in the equalizer. This study ensured the authenticity of the noise playback.

[0089] In this experiment, a professional Head LabP2 equalizer was used in conjunction with Sennheiser HD 600 type headphones for high-fidelity playback of sound samples.

[0090] Taking comfort, single - frequency sense, jitter sense, and ear - pressing sense as evaluation indicators, the evaluation is carried out based on the paired comparison method and the rank - scoring method with semantic assistance.

[0091] Comfort is used to reflect the quality of the sound sample, single - frequency sense is used to reflect the obviousness of the single - frequency peak in the sound, ear - pressing sense is used to reflect the degree of ear - pressing feeling caused by the low - frequency rumbling of road noise, and jitter sense is used to reflect the time - frequency fluctuation degree of the sound sample.

[0092] The rank - scoring method with semantic assistance is to first divide into multiple evaluation levels, provide clear semantic descriptions for each evaluation level to help understand the scoring criteria and reduce scoring deviation. Among the evaluation levels of the four evaluation indicators, the higher the comfort score, the more comfortable it is. For the other three indicators, the higher the score, the stronger the feeling. Among them, the evaluation level of comfort uses 11 - level semantic subdivision for scoring, and the evaluation levels of the other three indicators use 7 - level semantic subdivision for scoring. The evaluation levels of the four evaluation indicators are shown in Table 1. The higher the comfort score, the more comfortable it is, and for the other three indicators, the higher the score, the stronger the feeling.

[0093] Table 1 Rating Table for Each Indicator

[0094]

[0095]

[0096] The paired comparison method is also called the A / B method. It is necessary to play the sound samples grouped in pairs and make paired comparisons in the playing order. That is, the sound sample played first is recorded as A, and the sound sample played later is recorded as B. Then, make evaluations of "A is better than B", "A and B are about the same", or "B is better than A". If "A is better than B", then A gets 3 points. If "A and B are about the same", then each adds 1 point. If "B is better than A", then B gets 3 points. Finally, sum up the scores of each sound sample. The higher the total score of the sound sample, the better the performance of the sound sample in this acoustic property. The paired comparison method is relatively simple to operate and has low requirements for evaluation ability.

[0097] S5 Optimization Design: Optimize the noise reduction system of the car based on the evaluation scores to improve the road noise quality of the car.

[0098] In a specific embodiment of the present invention, in order to deeply analyze the influence mechanism of different time - frequency characteristics of the sound sample on the road noise evaluation indicators and demonstrate the benefits of preparing samples through spectrum editing in the present invention, this experiment performs time - frequency feature editing on the sound sample data under the condition of a stone road surface at 60 km / h, obtains the edited samples V1 - V5, and uses the paired comparison method to conduct four - indicator evaluations on the original samples and the edited samples.

[0099] Reference Figure 4, which is a radar chart of the average evaluation scores of the evaluation indicators for different edited samples. It can be seen from the figure that there are obvious differences in the indicator scores among different edited samples. Adjusting the amplitude of a certain frequency band of the adjusted sound sample will have a greater impact on the sound evaluation. Edited sample V1 reduces the peaks of 120 Hz and 220 Hz by 10 dB, resulting in a reduction in the overall energy of the noise, a reduction in the ear-pressure feeling, and a corresponding increase in comfort; Edited sample V2 adds noise to the 120 - 200 Hz frequency band, increasing the overall energy of the noise, and increasing the ear-pressure feeling, single-frequency feeling, and jitter feeling, while reducing comfort; Edited sample V3 reduces the noise near 120 Hz and 220 Hz, highlighting the single-frequency peak, and increasing both the single-frequency feeling and the jitter feeling, while reducing comfort; Edited sample V4 reduces the noise in the 50 - 250 Hz frequency band except for the single-frequency peak of 120 Hz, increasing both the single-frequency feeling and the jitter feeling, reducing the ear-pressure feeling, and increasing comfort; Edited sample V5 reduces the noise in the 50 - 200 Hz frequency band, reducing both the ear-pressure feeling and the jitter feeling, and increasing comfort, but due to the prominence of the 220 Hz single-frequency sound, the single-frequency feeling increases.

[0100] The paired comparison method for different edited samples shows that there is an obvious negative correlation between comfort and ear-pressure feeling and jitter feeling in the indicator score results. The correlation between the single-frequency feeling and other indicators is not obvious. In the perception of road noise samples, the ear-pressure feeling and jitter feeling have a greater impact on comfort.

[0101] Finally, based on the evaluation scores, engineers can improve the road noise quality by optimizing the tire design to reduce the pattern noise and cavity noise; they can also reduce the noise transmission by adjusting the modal frequency of the suspension system to avoid resonance with the body structure, and use high-damping materials or improve the body sheet metal process to reduce vibration and noise radiation.

[0102] The above-described embodiments merely represent several implementation manners of the present invention. Their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. For those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A method for preparing a guide for optimizing automotive noise reduction based on automotive road noise samples, characterized in that, It includes the following steps: 1) Collect the in-vehicle road noise data when the vehicle to be optimized for noise reduction is driving at a constant speed on different road surfaces at various speeds. 2) Determine a time-invariant FIR filter according to the frequency band to be edited in the in-vehicle road noise data and the sound pressure level change requirement of this frequency band, and then use the time-invariant FIR filter to perform spectral editing on the in-vehicle road noise data to obtain the in-vehicle road noise data after spectral editing. 3) Perform equal-loudness equalization processing at multiple loudness levels on both the in-vehicle road noise data and the in-vehicle road noise data after spectral editing to obtain in-vehicle road noise data samples. 4) Perform high-fidelity playback on the in-vehicle road noise data samples obtained in step 3) and the in-vehicle road noise data collected in step 1). Using comfort, single-frequency sense, jitter sense, and ear-pressure sense as evaluation indicators, evaluate the in-vehicle road noise data samples and the in-vehicle road noise data based on the paired comparison method and the rank scoring method with semantic assistance to obtain evaluation scores. 5) Optimize the noise reduction system of the vehicle based on the evaluation scores.

2. The method for guiding the optimization of vehicle noise reduction based on vehicle road noise samples according to claim 1, characterized in that, In step 1), the road surfaces include rough asphalt road surfaces, smooth asphalt road surfaces, and grooved cement road surfaces. When collecting the in-vehicle road noise data, the professional artificial head is uniformly arranged at the right rear seat of the vehicle to collect the in-vehicle road noise data.

3. The method for guiding the optimization of vehicle noise reduction based on vehicle road noise samples according to claim 1, wherein In step 2), determining the time-invariant FIR filter according to the frequency band to be edited in the in-vehicle road noise data and the sound pressure level change requirement of this frequency band includes: First, determine the frequency-domain response of the time-invariant ideal filter according to the frequency band to be edited in the in-vehicle road noise data and the sound pressure level change requirement of this frequency band, and then perform inverse discrete-time Fourier transform on the frequency-domain response to obtain the unit sample response of the filter. Select a window function according to the sound pressure level change and the minimum stopband attenuation of different window functions to truncate the infinite-length filter into a finite-length FIR filter. Finally, convolve the unit sample response with the window function to obtain the time-invariant FIR filter.

4. The method for preparing a guide for optimizing vehicle noise reduction based on vehicle road noise samples according to claim 3, wherein The frequency-domain response of the time-invariant ideal filter is: Among them, H d (e jω ) is the frequency-domain response of an ideal time-invariant filter; ΔL p is the sound pressure level change; f is the frequency within the frequency band where the in-vehicle road noise data needs to be edited, f s is the sampling frequency of the in-vehicle road noise data; ω1 is the normalized digital angular frequency corresponding to the lower cut-off frequency of the passband of the ideal filter, f1 is the lower cut-off frequency of the passband of the ideal filter, that is, the minimum frequency within the frequency band where the in-vehicle road noise data needs to be edited; ω2 is the normalized digital angular frequency corresponding to the upper cut-off frequency of the passband of the ideal filter, f2 is the upper cut-off frequency of the passband of the ideal filter, that is, the maximum frequency within the frequency band where the in-vehicle road noise data needs to be edited; τ is the group delay, representing the time delay amount of the ideal time-invariant filter.

5. The method for guiding the optimization of vehicle noise reduction based on vehicle road noise samples according to claim 3, wherein The unit sample response is: where h d (n) is the unit sample response of a time-invariant ideal filter; n is the discrete-time variable.

6. The method for guiding the optimization of vehicle noise reduction based on vehicle road noise samples according to claim 3, characterized in that, The types of the window function include rectangular window, triangular window, Hanning window, Hamming window, Blackman window, and Kaiser window.

7. The method for preparing a guide for optimizing vehicle noise reduction based on vehicle road noise samples according to claim 6, characterized in that, The time-invariant FIR filter is h(n) = h d (n)ω(n) where h(n) is the time-invariant FIR filter; ω(n) is the window function.

8. The method for guiding the optimization of vehicle noise reduction based on vehicle road noise samples according to claim 1, wherein The calculation method of the equal-loudness equalization processing is: Among them, y(n) target is the in-vehicle road noise data sample corresponding to the target loudness level; y(n) is the in-vehicle road noise data or the in-vehicle road noise data after spectral editing; L target is the target loudness level, and L is the loudness level corresponding to the in-vehicle road noise data or the loudness level corresponding to the in-vehicle road noise data after spectral editing.

9. The method for preparing a guidance for optimizing vehicle noise reduction based on vehicle road noise samples according to claim 1, wherein, In step 4), the rank scoring method with semantic assistance is to score the evaluation level of comfort using 11-level semantic subdivision, and score the evaluation levels of single-frequency sense, jitter sense, and ear-pressure sense using 7-level semantic subdivision.

Images