A method and system for mathematically cancelling propeller noise based on FFT

Abstract

This invention provides a method and system for mathematical noise cancellation of propellers based on FFT. It involves processing the time-domain signal data of propeller noise using FFT to obtain a frequency-domain signal. This frequency-domain signal is then plotted as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis. The spectrum is used to determine whether a singing tone exists within the frequency-domain signal of the propeller noise. If a singing tone is present, it is categorized into three types based on its influence range and position within the full frequency band of the spectrum: Type I, Type II, and Type III. Different judgment and calculation methods are then used to eliminate the singing tone in each of these three types, thus completing the noise cancellation of the propeller. This invention does not require time or financial costs to modify the propeller model, nor does it require changing the rated operating conditions in the original experimental plan.

Description

Technical Field

[0001] This invention relates to the field of acoustic control technology, specifically to a mathematical noise cancellation method and system for propellers based on FFT. Background Technology

[0002] Related research shows that when measuring the noise of a propeller model, resonance or other factors may cause a singing sound to appear on the propeller. Once this singing sound occurs, it will seriously hinder the acquisition of noise data from the model propeller, making it difficult to measure and evaluate the propeller's noise performance.

[0003] Therefore, it is necessary to eliminate propeller singing. Currently, the main methods are physical singing elimination and variable operating condition singing avoidance. Physical singing elimination is achieved by changing the shape of the propeller mold's trailing edge, attaching turbulence wires, or increasing the roughness of the guide edge to alter the frequency of the wake vortex emitted by the propeller blade's trailing edge or the natural frequency of the propeller blade. However, this method requires complex procedures and is cumbersome. Some methods can cause irreversible changes to the propeller mold, and it is difficult to verify its singing elimination effect before actual testing. Variable operating condition singing avoidance is achieved by changing parameters such as the propeller speed and the water velocity in the cavitation tank to avoid operating conditions where singing may occur, and continuing the experiment under operating conditions where singing does not occur. However, this method is not suitable for some experiments with strict operating conditions, and some propeller molds are very prone to singing, making it very difficult to find singing-free operating conditions through variable operating conditions. Summary of the Invention

[0004] To address a series of problems existing in current propeller noise cancellation processes, this invention provides a mathematical noise cancellation method for propellers based on FFT. Based on multiple sets of propeller noise data with singing under different operating conditions, FFT processing is used to convert the noise into frequency domain signals. Then, different singing phenomena are classified and cancelled using a mathematical noise cancellation method, without requiring changes to the rated operating conditions in the original test plan. This invention also relates to an FFT-based mathematical noise cancellation system for propellers.

[0005] The technical solution of the present invention is as follows:

[0006] A mathematical method for propeller noise cancellation based on FFT, characterized by the following steps:

[0007] Data acquisition and processing steps: acquire the time-domain signal data of propeller noise, perform FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plot the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0008] Singing sound classification steps: Determine whether singing sound appears in the frequency domain signal of propeller noise based on the spectrum diagram. If one or more sound pressure levels are abnormally increased in the narrow band of the frequency in the spectrum diagram, singing sound appears. According to the influence range of the singing sound and the position of the singing sound in the full frequency band of the spectrum diagram, the singing sound is divided into the first type of singing sound, the second type of singing sound, and the third type of singing sound.

[0009] The first type of singing tone elimination steps are as follows: obtain the center frequency, bandwidth, and peak sound pressure level of the first type of singing tone, and obtain the average sound pressure level of the frequency bands located on both sides of the first type of singing tone and having a certain multiple of bandwidth. Set a first threshold according to the peak and average sound pressure levels. Compare the sound pressure level at each frequency within the singing tone bandwidth with the first threshold. If the sound pressure level at each frequency is less than or equal to the first threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the first threshold, the sound pressure level is set to equal the first threshold to eliminate the first type of singing tone.

[0010] The second type of vocal tone elimination step: The second type of vocal tone includes a vocal peak and vocal bands located on the left and right sides of the vocal peak. The center frequency, bandwidth, and center frequency sound pressure level of the vocal peak and vocal band are obtained respectively. Multiple frequency points are obtained from adjacent frequency bands located to the left or right of the vocal band at intervals of a certain multiple of the bandwidth. A fitting equation for the vocal band is established based on the center frequency and each frequency point. The fitted sound pressure level at each frequency of the vocal band is calculated based on the fitting equation. The vocal tone in the vocal band is eliminated based on the fitted sound pressure level. The vocal tone in the vocal peak is eliminated based on the corresponding principle of the first type of vocal tone elimination step to eliminate the second type of vocal tone.

[0011] The third type of vocal tone elimination step is as follows: Obtain the center frequency, bandwidth, and peak sound pressure level of the third type of vocal tone. Based on the average sound pressure level of the frequency bands adjacent to the third type of vocal tone, determine whether the vocal tone is similar to the first or second type of vocal tone. If the vocal tone is similar to the second type of vocal tone, eliminate the vocal tone according to the corresponding principle of the second type of vocal tone elimination step. If the vocal tone is similar to the first type of vocal tone, calculate the first threshold according to the corresponding principle of the first type of vocal tone elimination step. Then, compensate the first threshold based on the peak sound pressure level of the third type of vocal tone to calculate the second threshold. Compare the sound pressure level at each frequency within the bandwidth of the third type of vocal tone with the second threshold. If the sound pressure level at each frequency is less than or equal to the second threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the second threshold, set the sound pressure level to the second threshold to eliminate the third type of vocal tone.

[0012] Preferably, in the data acquisition and processing steps, after obtaining the frequency domain signal, the amplitude unit of the frequency domain signal is converted from sound pressure to sound pressure level, and the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0013] Preferably, in the singing tone classification step, if the sound pressure level of a singing tone does not exceed the third threshold of the sound pressure level of the adjacent frequency band, then the singing tone is not processed; if the sound pressure level of a singing tone exceeds the third threshold of the sound pressure level of the adjacent frequency band and is at the highest point of the sound pressure level of the entire frequency band, or if the sound pressure level of the singing tone exceeds the fourth threshold of the sound pressure level of the adjacent frequency band, then the singing tone affects the noise data and needs to be processed.

[0014] Preferably, in the vocal classification step, the full frequency band includes the rising band, the falling band, and the middle band.

[0015] Preferably, in the singing tone classification step, classifying the singing tone into a first category, a second category, and a third category based on the influence range of the singing tone and its position in the full frequency band of the spectrogram includes:

[0016] Singing sounds that occur in the rising or falling segment and have little impact on the sound pressure level of adjacent frequency bands are called Class I singing sounds. Singing sounds that occur in the rising or falling segment and cause the sound pressure level of adjacent frequency bands to rise are called Class II singing sounds. Singing sounds that cannot be determined to occur in the rising or falling segment are called Class III singing sounds.

[0017] A propeller noise mathematical cancellation system based on FFT is characterized by comprising a data acquisition and processing module, a vocal classification module, a first type of vocal cancellation module, a second type of vocal cancellation module, and a third type of vocal cancellation module connected in sequence.

[0018] The data acquisition and processing module acquires the time-domain signal data of propeller noise, performs FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plots the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0019] The vocalization classification module determines whether vocalization occurs in the frequency domain signal of propeller noise based on the spectrum diagram. If one or more sound pressure levels are abnormally increased in the narrow band of the frequency in the spectrum diagram, vocalization occurs. Based on the influence range of the vocalization and its position in the full frequency band of the spectrum diagram, the vocalization is divided into Class I vocalization, Class II vocalization, and Class III vocalization.

[0020] The first type of singing tone elimination module obtains the center frequency, bandwidth, and peak sound pressure level of the first type of singing tone, and obtains the average sound pressure level of the frequency bands located on both sides of the first type of singing tone and having a certain multiple of bandwidth. Based on the peak and average sound pressure levels, a first threshold is set. The sound pressure level at each frequency within the singing tone bandwidth is compared with the first threshold. If the sound pressure level at each frequency is less than or equal to the first threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the first threshold, the sound pressure level is set to equal the first threshold to eliminate the first type of singing tone.

[0021] The second type of vocal cancellation module includes a vocal peak and vocal bands located on the left and right sides of the vocal peak. It acquires the center frequency, bandwidth, and center frequency sound pressure level of both the vocal peak and vocal bands. It obtains multiple frequency points from adjacent frequency bands located to the left or right of the vocal bands at intervals of a certain multiple of bandwidth. It establishes a fitting equation for the vocal bands based on the center frequency and each frequency point, calculates the fitted sound pressure level at each frequency of the vocal bands based on the fitting equation, and eliminates the vocals in the vocal bands based on the fitted sound pressure level. Furthermore, it eliminates the vocals in the vocal peaks based on the corresponding principle of the first type of vocal cancellation module to eliminate the second type of vocals.

[0022] The third type of vocal tone elimination module acquires the center frequency, bandwidth, and peak sound pressure level of the third type of vocal tone. Based on the average sound pressure level of the frequency bands adjacent to the third type of vocal tone, it determines whether the vocal tone is similar to the first or second type of vocal tone. If the vocal tone is similar to the second type of vocal tone, it eliminates the vocal tone according to the corresponding principle of the second type of vocal tone elimination module. If the vocal tone is similar to the first type of vocal tone, it calculates a first threshold according to the corresponding principle of the first type of vocal tone elimination module. Then, it compensates the first threshold based on the peak sound pressure level of the third type of vocal tone to calculate a second threshold. It compares the sound pressure level at each frequency within the bandwidth of the third type of vocal tone with the second threshold. If the sound pressure level at each frequency is less than or equal to the second threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the second threshold, the sound pressure level is set to equal the second threshold to eliminate the third type of vocal tone.

[0023] Preferably, in the data acquisition and processing module, after obtaining the frequency domain signal, the amplitude unit of the frequency domain signal is converted from sound pressure to sound pressure level, and the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0024] Preferably, in the vocal classification module, when the sound pressure level of a vocal note does not exceed the third threshold of the sound pressure level of the adjacent frequency band, the vocal note is not processed; when the sound pressure level of a vocal note exceeds the third threshold of the sound pressure level of the adjacent frequency band and is at the highest point of the sound pressure level of the entire frequency band, or when the sound pressure level of the vocal note exceeds the fourth threshold of the sound pressure level of the adjacent frequency band, the vocal note affects the noise data and needs to be processed.

[0025] Preferably, the full frequency band includes a rising band, a falling band, and a mid-band.

[0026] Preferably, in the singing tone classification module, the singing tone is classified into a first category, a second category, and a third category based on the influence range of the singing tone and its position in the full frequency band of the spectrogram, including:

[0027] Singing sounds that occur in the rising or falling segment and have little impact on the sound pressure level of adjacent frequency bands are called Class I singing sounds. Singing sounds that occur in the rising or falling segment and cause the sound pressure level of adjacent frequency bands to rise are called Class II singing sounds. Singing sounds that cannot be determined to occur in the rising or falling segment are called Class III singing sounds.

[0028] The beneficial effects of this invention are as follows:

[0029] This invention provides a mathematical method for eliminating propeller noise based on FFT. First, the time-domain signal data of the propeller noise is acquired, and then FFT processing is performed on the noise time-domain signal to obtain the frequency-domain signal. The frequency-domain signal is plotted as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis. Based on the spectrum, it is determined whether a singing sound appears in the frequency-domain signal of the propeller noise. If one or more sound pressure levels abnormally increase in a narrow frequency band in the spectrum, a singing sound is detected. Based on the influence range of the singing sound and its position in the full frequency band of the spectrum, the singing sound is classified into three categories: Category I, Category II, and Category III. Different judgment and calculation methods are then used to eliminate the singing sound in each category. Combined with fitting calculations and compensation calculations, the method determines whether secondary fitting is needed to set a threshold and whether compensation is required when setting the threshold, based on the influence range of the singing sound. This approach effectively eliminates the singing sound. Compared to existing technologies that use physical noise reduction and variable operating conditions to avoid noise, this invention uses mathematical noise reduction, which does not require time or money to modify the propeller model, nor does it require changing the rated operating conditions in the original test plan, and can achieve a more ideal noise reduction effect.

[0030] This invention also relates to an FFT-based propeller noise mathematical cancellation system, which corresponds to the aforementioned FFT-based propeller noise mathematical cancellation method. This system can be understood as a system that implements the aforementioned FFT-based propeller noise mathematical cancellation method. It includes a data acquisition and processing module, a noise classification module, a first-type noise cancellation module, a second-type noise cancellation module, and a third-type noise cancellation module connected in sequence. Each module works collaboratively. Based on multiple sets of noise data with noise from different propellers under different operating conditions, FFT processing is performed to obtain frequency domain signals, and different noise phenomena are classified. For different types of noise, different processing methods are used to set thresholds, achieving a relatively ideal cancellation effect without changing the rated operating conditions in the original experimental plan. Attached Figure Description

[0031] Figure 1 This is a flowchart of the propeller noise mathematical cancellation method based on FFT of the present invention.

[0032] Figure 2 This is a schematic diagram of the propeller noise test spectrum of the present invention.

[0033] Figure 3 This is a mathematical cancellation effect diagram of the propeller-shaped audio spectrum of the present invention.

[0034] Figure 4 This is a diagram showing the physical cancellation effect of the existing propeller-driven audio spectrum. Detailed Implementation

[0035] The present invention will now be described with reference to the accompanying drawings.

[0036] This invention relates to a mathematical method for propeller noise cancellation based on FFT, the flowchart of which is shown below. Figure 1 As shown, the steps are as follows:

[0037] I. Data Acquisition and Processing Steps: Acquire the time-domain signal data of propeller noise, perform FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plot the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0038] Specifically, the propeller noise time-domain signal data is first obtained using conventional noise measurement methods based on the noise measurement system. Then, FFT processing is used to convert the propeller noise time-domain signal into a frequency-domain signal. Due to limitations of laboratory hardware, only the frequency band of 1000Hz to 80000Hz is considered reliable. Through mathematical conversion, the amplitude of the frequency-domain signal is converted from sound pressure (unit: Pa) to sound pressure level (unit: dB). The specific formula is as follows:

[0039]

[0040] In the above formula, L is the sound pressure level (dB) at a single frequency, p is the sound pressure (Pa) at that frequency, and the reference sound pressure p0 = 1 × 10 -6 Pa.

[0041] Finally, the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0042] II. Singing Tone Classification Steps: Determine whether singing tone appears in the frequency domain signal of the propeller noise based on the spectrum diagram. If one or more sound pressure levels abnormally increase in a narrow frequency band in the spectrum diagram, singing tone is present. Based on the range of influence of the singing tone and its position in the full frequency band of the spectrum diagram, the singing tone is classified into three categories: Category I, Category II, and Category III. That is, as follows... Figure 1 The diagram shows the classification of singing sounds appearing in the frequency domain of noise signals.

[0043] Specifically, the presence of singing sounds in the frequency domain signal of propeller noise is determined by analyzing the spectrum. If one or more sound pressure levels abnormally increase in a narrow frequency band in the spectrum, singing sounds are considered to be present. In the laboratory, if the sound pressure level of a singing sound does not exceed the sound pressure level of the adjacent frequency band by 5 dB (preferably the third threshold), the singing sound phenomenon is considered insignificant and no action is required. If the sound pressure level of a singing sound exceeds the sound pressure level of the adjacent frequency band by more than 5 dB (preferably the third threshold) and is at the highest sound pressure level across the entire frequency band, or if the sound pressure level of a singing sound exceeds the sound pressure level of the adjacent frequency band by more than 10 dB (preferably the fourth threshold), the singing sound is considered to have a significant impact on the noise data and requires processing.

[0044] Among them, propeller noise signals without cavitation (where the singing disappears naturally after cavitation occurs) exhibit certain patterns. Specifically, across the entire frequency band (1000Hz–80000Hz), the sound pressure level gradually increases from 1000Hz to XHz (hereinafter referred to as the rising segment), gradually decreases from YHz to 80000Hz (hereinafter referred to as the falling segment), and exhibits relatively stable or irregularly fluctuating noise sound pressure levels from XHz to YHz (hereinafter referred to as the middle segment). For certain propeller operating conditions, X = Y may occur, in which case the middle segment does not exist. After the singing occurs, some singing frequencies are extremely narrow, having minimal impact on the sound pressure level of adjacent frequency bands; in this case, the singing can be approximated as a singing peak with an extremely narrow frequency band. Other singing frequencies are not limited to an extremely narrow frequency band but also cause a certain increase in the sound pressure level of adjacent frequency bands; in this case, the singing can be considered as a singing band with a certain bandwidth. Therefore, once the occurrence of a singing tone is determined, it can be classified according to the range of its influence and its position in the entire frequency band. Specifically, singing tones that are determined to occur in the rising or falling segment and have minimal impact on the sound pressure level of adjacent frequency bands are called Class I singing tones; singing tones that are determined to occur in the rising or falling segment and cause an increase in the sound pressure level of adjacent frequency bands are called Class II singing tones; and singing tones that are difficult to determine directly as occurring in the rising or falling segment are called Class III singing tones.

[0045] For example, taking two different propeller models, A and B, as examples, each propeller model is subjected to two operating conditions that produce sound (condition 1 and condition 2), and its noise spectrum diagram is as follows. Figure 2 As shown, a series of vocal tones appear in the four groups of spectrograms, all of which are numbered. Based on the classification method described above, the vocal tones in the diagrams can be classified as follows:

[0046] The first type of singing tone: E peak in A-paddle condition 1, H peak in A-paddle condition 2, I peak in B-paddle condition 1, and L peak in B-paddle condition 2.

[0047] The second type of singing tone: A and D peaks in A-paddle condition 1, and G peak in A-paddle condition 2;

[0048] The third type of vocalization: B and C peaks in A-paddle condition 1, F peak in A-paddle condition 2, J and K peaks in B-paddle condition 1, and M and N peaks in B-paddle condition 2.

[0049] Furthermore, before performing sing-off processing on the aforementioned types of vocalizations, it is necessary to understand that the essence of vocalization is an energy concentration phenomenon caused by resonance. Even if the vocalization phenomenon at a certain frequency is eliminated, the energy at that frequency itself should be slightly higher than that of the adjacent frequency bands. Therefore, completely eliminating the vocalization phenomenon would cause the noise data to lose its spectral characteristics, which is undesirable.

[0050] III. First-Type Singing Tone Elimination Steps: Obtain the center frequency, bandwidth, and peak sound pressure level (SPL) of the first-type singing tone. Also, obtain the average SPL of the frequency bands located on either side of the first-type singing tone, each with a certain multiple of bandwidth. Set a first threshold based on the peak and average SPL. Compare the SPL at each frequency within the singing tone bandwidth with the first threshold. If the SPL at each frequency is less than or equal to the first threshold, the SPL remains unchanged. If the SPL at each frequency is greater than the first threshold, set the SPL equal to the first threshold to eliminate the first-type singing tone. Based on the range of the singing tone's influence, this first-type singing tone does not require secondary fitting to set the first threshold, nor does it require compensation when setting the first threshold.

[0051] Specifically, the center frequency a, bandwidth b, and peak sound pressure level L1 of the first type of singing tone are first obtained.

[0052] Because the characteristics of the first type of singing tone are highly concentrated energy and a very small bandwidth b, the processing method for this type of singing tone is to remove it by setting a threshold (first threshold). Specifically, the threshold is set by obtaining the average sound pressure level L2 of 5 times the bandwidth on both the left and right sides of the singing tone's narrow band.

[0053]

[0054] Then, based on the peak sound pressure level L1 of the singing tone and the average sound pressure level L2 of the adjacent frequency band, a first threshold D was set. Based on the results of multiple experiments and the comparison of the effect after physical cancellation, a more applicable threshold calculation formula was summarized:

[0055]

[0056] Finally, using D as the first threshold, for the first type of singing tone, the sound pressure level L at each frequency within the bandwidth b of that type of singing tone is calculated. x (a+0.5b≥x≥a-0.5b) is processed as follows: When L x When L ≤ D, x Remain unchanged; when L x When >D, let L x =D, thus completing the elimination of the first type of singing tone.

[0057] IV. Second Type of Vocal Removal Steps: The second type of vocal resonance includes a vocal peak and vocal bands located to the left and right of the vocal peak. The center frequency, bandwidth, and center frequency sound pressure level of both the vocal peak and vocal bands are obtained. Multiple frequency points are obtained from adjacent frequency bands located to the left or right of the vocal bands at intervals. A fitting equation for the vocal band is established based on its center frequency and each frequency point. The fitted sound pressure level at each frequency of the vocal band is calculated based on the fitting equation. Vocal resonance in the vocal band is removed based on the fitted sound pressure level. Furthermore, the vocal resonance in the vocal peak is removed according to the corresponding principle of the first type of vocal resonance removal steps to eliminate the second type of vocal resonance. That is, as follows... Figure 1 The method shown determines whether a second fitting is needed to set a threshold based on the range of influence of the singing tone. This second type of singing tone uses a fitting calculation technique to perform a second fitting before calculating and setting the first threshold.

[0058] Specifically, the second type of vocal tone consists of a vocal peak similar to that of the first type of vocal tone, and an adjacent frequency band (i.e., the vocal band) whose sound pressure level is increased due to the influence of the vocal peak. When processing the second type of vocal tone, the vocal band needs to be processed first, followed by the vocal peak. The specific processing flow is as follows:

[0059] s1: Obtain the center frequency a, bandwidth b, and peak sound pressure level L1 of the vocal peak, and obtain the center frequency m, bandwidth n, and center frequency sound pressure level L of the vocal band. m .

[0060] s2: To facilitate the calculation of the fitting formula, the center frequency m of the vocal band is used as the dividing line. The left vocal band (m-0.5n) Hz to m Hz and the right vocal band (m Hz to (m+0.5n) Hz) are processed separately. For the left vocal band, noise data from the unaffected bandwidth of 5 times the vocal band (m-5.5n) Hz to (m-0.5n) Hz is used, with 50 frequency points n1 to nn at 0.1n Hz intervals. 50 and their respective sound pressure levels L n1 ~L n50 Using these fifty points, a quadratic fit was performed, ultimately yielding the quadratic equation (fitting equation) for the left side of the vocal track:

[0061]

[0062] s3: Let the original sound pressure level at each frequency of the vocal cord be L. n1 When performing a vocal cancellation operation, the center frequency of the vocal peak is used as the dividing point. The frequencies (m-0.5n) Hz to a Hz to the left of this point are substituted into the quadratic equation obtained above to calculate the fitted sound pressure level L at each frequency. n2 In order to preserve the original spectral characteristics, the corrected left vocal band sound pressure level L nxfor:

[0063]

[0064] s4: Repeat step s3 for the frequency band aHz~(m+0.5n)Hz to the right of the center frequency of the singing tone peak, thereby completing the singing tone band cancellation work for the second type of singing tone.

[0065] After completing step s4, following the principle of the first type of vocal cancellation step, D is calculated as the first threshold. For the second type of vocalization, the sound pressure level L at each frequency within the bandwidth b of that type of vocalization is then calculated. x (a+0.5b≥x≥a-0.5b) is processed as follows: When L x When L ≤ D, x Remain unchanged; when L x When >D, let L x =D, thus completing the elimination of the second type of singing tone.

[0066] V. Third-Category Singing Tone Elimination Steps: Obtain the center frequency, bandwidth, and peak sound pressure level (SPL) of the third-category singing tone. Based on the average SPL of the frequency bands adjacent to the third-category singing tone, determine if it is similar to the first or second-category singing tone. If similar to the second-category singing tone, eliminate it according to the principle of the second-category singing tone elimination step. If similar to the first-category singing tone, calculate a first threshold according to the principle of the first-category singing tone elimination step. Then, compensate the first threshold based on the peak SPL of the third-category singing tone to calculate a second threshold. Compare the SPL at each frequency within the bandwidth of the third-category singing tone with the second threshold. If the SPL at each frequency is less than or equal to the second threshold, keep the SPL unchanged. If the SPL at each frequency is greater than the second threshold, set the SPL equal to the second threshold to eliminate the third-category singing tone. That is, as follows... Figure 1 The method shown determines whether compensation is needed when setting a threshold based on the range of influence of the singing tone. This third type of singing tone uses compensation calculation technology to calculate the second threshold after compensation by compensating the first threshold.

[0067] Specifically, the third type of vocalization is no different in form from the first or second type. The only difference lies in the fact that the location of the third type of vocalization cannot be directly determined as the rising or falling segment of the entire frequency band. This means that the location of this type of vocalization is actually at the peak of the sound pressure level in the absence of vocalization. In this case, referring to the processing methods for the first or second type of vocalization and directly using the sound pressure level of the adjacent frequency band for vocalization removal will result in a loss of its spectral characteristics, causing the removed vocalization sound pressure level to be lower than the true sound pressure level of the absence of vocalization, making the vocalization removal result unreliable. To ensure the spectral characteristics of the third type of vocalization after vocalization removal, its threshold (first threshold D) needs to be compensated. The specific processing flow is as follows:

[0068] 1) Obtain the center frequency a, bandwidth b, and peak sound pressure level L1 of the third type of singing tone, and determine whether the singing tone is similar to the first type of singing tone or the second type of singing tone based on its performance.

[0069] 2) If the singing tone is similar to the second type of singing tone and both drive the surrounding frequency bands to increase, then according to the second type of singing tone elimination steps s1 to s4, the singing tone band elimination work is completed.

[0070] 3) If the singing tone is similar to the first type of singing tone, the first threshold D is calculated according to the first type of singing tone elimination steps. At this time, the third type of singing tone will not be processed.

[0071] 4) Compensate for the first threshold D. Since the singing tone is in the middle of the full frequency band, its spectral characteristics will be preserved to a greater extent. After calculation, the following formula can properly preserve the spectral characteristics of the third type of singing tone after the singing tone is canceled.

[0072] D2 = 0.15 × L1 + 0.85 × D (6)

[0073] Finally, D2 is used as the compensated threshold, i.e., the second threshold, to calculate the sound pressure level L at each frequency within the bandwidth of the third type of vocal tone. x (a+0.5b≥x≥a-0.5b) is processed as follows: When L x When L ≤ D2, x Remain unchanged; when L x When >D2, let L x =D2, thus completing the cancellation of the third type of singing tone.

[0074] Example:

[0075] by Figure 2 Taking the musical spectrogram shown in the figure as an example, Figure 2 All vocalizations under four operating conditions (Oarsor A Condition 1, Oarsor A Condition 2, Oarsor B Condition 1, Oarsor B Condition 2) are mathematically canceled according to the method described above in this invention, thereby obtaining the following... Figure 3 The four sets of mathematical cancellation spectrum diagrams are shown. Meanwhile, to enhance the comparison between this invention and existing technologies, and by altering the flow regime on the blade surface of propellers A and B by attaching copper wires at 0.5R to 0.9R along the propeller mold edge, the existing technology performs physical cancellation and repeats the above four sets of operating conditions to obtain the following results. Figure 4 The four sets of physical cancellation spectrum diagrams are shown. That is, the mathematical cancellation implemented in this invention and the physical cancellation spectrum diagrams implemented in the prior art are respectively as follows: Figure 3 and Figure 4 As shown.

[0076] Figure 3 Represents the result of mathematical elimination. Figure 4 The results of physical cancellation show that both have achieved a relatively ideal cancellation effect, and the final spectrograms are extremely similar, thus proving the feasibility of the method of the present invention.

[0077] This invention also relates to an FFT-based propeller noise mathematical cancellation system, which corresponds to the aforementioned FFT-based propeller noise mathematical cancellation method and can be understood as a system for implementing the above method. The system includes a data acquisition and processing module, a vocal classification module, a first-type vocal cancellation module, a second-type vocal cancellation module, and a third-type vocal cancellation module connected in sequence. Specifically,

[0078] The data acquisition and processing module acquires the time-domain signal data of propeller noise, performs FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plots the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0079] The vocalization classification module determines whether vocalization occurs in the frequency domain signal of propeller noise based on the spectrum diagram. If one or more sound pressure levels are abnormally increased in the narrow band of the frequency in the spectrum diagram, vocalization occurs. Based on the influence range of the vocalization and its position in the full frequency band of the spectrum diagram, the vocalization is divided into Class I vocalization, Class II vocalization, and Class III vocalization.

[0080] The first type of singing tone elimination module obtains the center frequency, bandwidth, and peak sound pressure level of the first type of singing tone, and obtains the average sound pressure level of the frequency bands located on both sides of the first type of singing tone and having a certain multiple of bandwidth. Based on the peak and average sound pressure levels, a first threshold is set. The sound pressure level at each frequency within the singing tone bandwidth is compared with the first threshold. If the sound pressure level at each frequency is less than or equal to the first threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the first threshold, the sound pressure level is set to equal the first threshold to eliminate the first type of singing tone.

[0081] The second type of vocal cancellation module includes a vocal peak and vocal bands located on the left and right sides of the vocal peak. It acquires the center frequency, bandwidth, and center frequency sound pressure level of both the vocal peak and vocal bands. It obtains multiple frequency points from adjacent frequency bands located to the left or right of the vocal bands at intervals of a certain multiple of bandwidth. It establishes a fitting equation for the vocal bands based on the center frequency and each frequency point, calculates the fitted sound pressure level at each frequency of the vocal bands based on the fitting equation, and eliminates the vocals in the vocal bands based on the fitted sound pressure level. Furthermore, it eliminates the vocals in the vocal peaks based on the corresponding principle of the first type of vocal cancellation module to eliminate the second type of vocals.

[0082] The third type of vocal tone elimination module acquires the center frequency, bandwidth, and peak sound pressure level of the third type of vocal tone. Based on the average sound pressure level of the frequency bands adjacent to the third type of vocal tone, it determines whether the vocal tone is similar to the first or second type of vocal tone. If the vocal tone is similar to the second type of vocal tone, it eliminates the vocal tone according to the corresponding principle of the second type of vocal tone elimination module. If the vocal tone is similar to the first type of vocal tone, it calculates a first threshold according to the corresponding principle of the first type of vocal tone elimination module. Then, it compensates the first threshold based on the peak sound pressure level of the third type of vocal tone to calculate a second threshold. It compares the sound pressure level at each frequency within the bandwidth of the third type of vocal tone with the second threshold. If the sound pressure level at each frequency is less than or equal to the second threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the second threshold, the sound pressure level is set to equal the second threshold to eliminate the third type of vocal tone.

[0083] Preferably, in the data acquisition and processing module, after obtaining the frequency domain signal, the amplitude unit of the frequency domain signal is converted from sound pressure to sound pressure level, and the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

[0084] Preferably, in the singing sound classification module, when the sound pressure level of a singing sound does not exceed the third threshold of the sound pressure level of the adjacent frequency band, the singing sound phenomenon is not obvious and no processing is required; when the sound pressure level of a singing sound exceeds the third threshold of the sound pressure level of the adjacent frequency band and is at the highest point of the sound pressure level of the entire frequency band, or when the sound pressure level of the singing sound exceeds the fourth threshold of the sound pressure level of the adjacent frequency band, the singing sound affects the noise data and needs to be processed.

[0085] Preferably, the full frequency band includes a rising band, a falling band, and a mid-band.

[0086] Preferably, in the vocal tone classification module, vocal tones are classified into three categories—a first category, a second category, and a third category—based on their influence range and position within the full frequency band of the spectrogram:

[0087] Singing sounds that occur in the rising or falling segment and have little impact on the sound pressure level of adjacent frequency bands are called Class I singing sounds. Singing sounds that occur in the rising or falling segment and cause the sound pressure level of adjacent frequency bands to rise are called Class II singing sounds. Singing sounds that cannot be determined to occur in the rising or falling segment are called Class III singing sounds.

[0088] This invention provides an objective and scientific FFT-based mathematical noise cancellation method and system for propellers. Based on multiple sets of propeller noise data with singing under different operating conditions, the method uses FFT processing to convert the noise into a frequency domain signal, then classifies the different singing phenomena, and sets thresholds for different types of singing using different processing methods. The mathematical noise cancellation scheme achieves the ideal noise cancellation effect without changing the rated operating conditions in the original test plan.

[0089] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail with reference to the accompanying drawings and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention. In short, all technical solutions and improvements that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention patent.

Claims

1. A mathematical method for propeller noise cancellation based on FFT, characterized in that, Includes the following steps: Data acquisition and processing steps: acquire the time-domain signal data of propeller noise, perform FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plot the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis. Singing sound classification steps: Determine whether singing sound appears in the frequency domain signal of propeller noise based on the spectrum diagram. If one or more sound pressure levels are abnormally increased in the narrow band of the frequency in the spectrum diagram, singing sound appears. According to the influence range of the singing sound and the position of the singing sound in the full frequency band of the spectrum diagram, the singing sound is divided into the first type of singing sound, the second type of singing sound, and the third type of singing sound. The first type of singing tone elimination steps are as follows: obtain the center frequency, bandwidth, and peak sound pressure level of the first type of singing tone, and obtain the average sound pressure level of the frequency bands located on both sides of the first type of singing tone and having a certain multiple of bandwidth. Set a first threshold according to the peak and average sound pressure levels. Compare the sound pressure level at each frequency within the singing tone bandwidth with the first threshold. If the sound pressure level at each frequency is less than or equal to the first threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the first threshold, the sound pressure level is set to equal the first threshold to eliminate the first type of singing tone. The second type of vocal tone elimination step: The second type of vocal tone includes a vocal peak and vocal bands located on the left and right sides of the vocal peak. The center frequency, bandwidth, and center frequency sound pressure level of the vocal peak and vocal band are obtained respectively. Multiple frequency points are obtained from adjacent frequency bands located to the left or right of the vocal band at intervals of a certain multiple of the bandwidth. A fitting equation for the vocal band is established based on the center frequency and each frequency point. The fitted sound pressure level at each frequency of the vocal band is calculated based on the fitting equation. The vocal tone in the vocal band is eliminated based on the fitted sound pressure level. The vocal tone in the vocal peak is eliminated based on the corresponding principle of the first type of vocal tone elimination step to eliminate the second type of vocal tone. The third type of vocal tone elimination step is as follows: Obtain the center frequency, bandwidth, and peak sound pressure level of the third type of vocal tone. Based on the average sound pressure level of the frequency bands adjacent to the third type of vocal tone, determine whether the vocal tone is similar to the first or second type of vocal tone. If the vocal tone is similar to the second type of vocal tone, eliminate the vocal tone according to the corresponding principle of the second type of vocal tone elimination step. If the vocal tone is similar to the first type of vocal tone, calculate the first threshold according to the corresponding principle of the first type of vocal tone elimination step. Then, compensate the first threshold based on the peak sound pressure level of the third type of vocal tone to calculate the second threshold. Compare the sound pressure level at each frequency within the bandwidth of the third type of vocal tone with the second threshold. If the sound pressure level at each frequency is less than or equal to the second threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the second threshold, set the sound pressure level to the second threshold to eliminate the third type of vocal tone.

2. The propeller noise mathematical cancellation method based on FFT according to claim 1, characterized in that, In the data acquisition and processing steps, after obtaining the frequency domain signal, the amplitude unit of the frequency domain signal is converted from sound pressure to sound pressure level, and the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

3. The propeller noise mathematical cancellation method based on FFT according to claim 1, characterized in that, In the singing tone classification step, if the sound pressure level of a singing tone does not exceed the third threshold of the sound pressure level of the adjacent frequency band, then the singing tone will not be processed; if the sound pressure level of a singing tone exceeds the third threshold of the sound pressure level of the adjacent frequency band and is at the highest point of the sound pressure level of the entire frequency band, or if the sound pressure level of the singing tone exceeds the fourth threshold of the sound pressure level of the adjacent frequency band, then the singing tone will affect the noise data and needs to be processed.

4. The FFT-based mathematical cancellation method for propeller noise according to claim 1 or 3, characterized in that, In the vocal classification step, the full frequency band includes the rising band, the falling band, and the middle band.

5. The propeller noise mathematical cancellation method based on FFT according to claim 4, characterized in that, In the singing tone classification step, the singing tone is divided into three categories—a first category, a second category, and a third category—based on its influence range and its position in the full frequency band of the spectrogram: Singing sounds that occur in the rising or falling segment and have little impact on the sound pressure level of adjacent frequency bands are called Class I singing sounds. Singing sounds that occur in the rising or falling segment and cause the sound pressure level of adjacent frequency bands to rise are called Class II singing sounds. Singing sounds that cannot be determined to occur in the rising or falling segment are called Class III singing sounds.

6. A propeller noise mathematical cancellation system based on FFT, characterized in that, It includes a data acquisition and processing module, a vocal classification module, a first-type vocal elimination module, a second-type vocal elimination module, and a third-type vocal elimination module, which are connected in sequence. The data acquisition and processing module acquires the time-domain signal data of propeller noise, performs FFT processing on the noise time-domain signal to obtain the frequency-domain signal, and plots the frequency-domain signal as a spectrum with frequency on the horizontal axis and sound pressure level on the vertical axis. The vocalization classification module determines whether vocalization occurs in the frequency domain signal of propeller noise based on the spectrum diagram. If one or more sound pressure levels are abnormally increased in the narrow band of the frequency in the spectrum diagram, vocalization occurs. Based on the influence range of the vocalization and its position in the full frequency band of the spectrum diagram, the vocalization is divided into Class I vocalization, Class II vocalization, and Class III vocalization. The first type of singing tone elimination module obtains the center frequency, bandwidth, and peak sound pressure level of the first type of singing tone, and obtains the average sound pressure level of the frequency bands located on both sides of the first type of singing tone and having a certain multiple of bandwidth. Based on the peak and average sound pressure levels, a first threshold is set. The sound pressure level at each frequency within the singing tone bandwidth is compared with the first threshold. If the sound pressure level at each frequency is less than or equal to the first threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the first threshold, the sound pressure level is set to equal the first threshold to eliminate the first type of singing tone. The second type of vocal cancellation module includes a vocal peak and vocal bands located on the left and right sides of the vocal peak. It acquires the center frequency, bandwidth, and center frequency sound pressure level of both the vocal peak and vocal bands. It obtains multiple frequency points from adjacent frequency bands located to the left or right of the vocal bands at intervals of a certain multiple of bandwidth. It establishes a fitting equation for the vocal bands based on the center frequency and each frequency point, calculates the fitted sound pressure level at each frequency of the vocal bands based on the fitting equation, and eliminates the vocals in the vocal bands based on the fitted sound pressure level. Furthermore, it eliminates the vocals in the vocal peaks based on the corresponding principle of the first type of vocal cancellation module to eliminate the second type of vocals. The third type of vocal tone elimination module acquires the center frequency, bandwidth, and peak sound pressure level of the third type of vocal tone. Based on the average sound pressure level of the frequency bands adjacent to the third type of vocal tone, it determines whether the vocal tone is similar to the first or second type of vocal tone. If the vocal tone is similar to the second type of vocal tone, it eliminates the vocal tone according to the corresponding principle of the second type of vocal tone elimination module. If the vocal tone is similar to the first type of vocal tone, it calculates a first threshold according to the corresponding principle of the first type of vocal tone elimination module. Then, it compensates the first threshold based on the peak sound pressure level of the third type of vocal tone to calculate a second threshold. It compares the sound pressure level at each frequency within the bandwidth of the third type of vocal tone with the second threshold. If the sound pressure level at each frequency is less than or equal to the second threshold, the sound pressure level remains unchanged. If the sound pressure level at each frequency is greater than the second threshold, the sound pressure level is set to equal the second threshold to eliminate the third type of vocal tone.

7. The propeller noise mathematical cancellation system based on FFT according to claim 6, characterized in that, In the data acquisition and processing module, after obtaining the frequency domain signal, the amplitude unit of the frequency domain signal is converted from sound pressure to sound pressure level, and the frequency domain signal after unit conversion is plotted as a spectrum diagram with frequency on the horizontal axis and sound pressure level on the vertical axis.

8. The propeller noise mathematical cancellation system based on FFT according to claim 6, characterized in that, In the vocal classification module, if the sound pressure level of a vocal note does not exceed the third threshold of the sound pressure level of the adjacent frequency band, then the vocal note will not be processed. If the sound pressure level of a vocal note exceeds the third threshold of the sound pressure level of the adjacent frequency band and is at the highest point of the sound pressure level of the entire frequency band, or if the sound pressure level of the vocal note exceeds the fourth threshold of the sound pressure level of the adjacent frequency band, then the vocal note will affect the noise data and needs to be processed.

9. The propeller noise mathematical cancellation system based on FFT according to claim 6 or 8, characterized in that, The full-band frequency band includes the rising band, the falling band, and the middle band.

10. The propeller noise mathematical cancellation system based on FFT according to claim 9, characterized in that, In the singing tone classification module, singing tones are divided into three categories—a first category, a second category, and a third category—based on the influence range of the singing tone and its position in the full frequency band of the spectrogram: Singing sounds that occur in the rising or falling segment and have little impact on the sound pressure level of adjacent frequency bands are called Class I singing sounds. Singing sounds that occur in the rising or falling segment and cause the sound pressure level of adjacent frequency bands to rise are called Class II singing sounds. Singing sounds that cannot be determined to occur in the rising or falling segment are called Class III singing sounds.

Images