A method for extracting natural frequencies for timing signal processing at the tip of a single sensor leaf.

Abstract

This disclosure provides a method for extracting the natural frequency of a single-sensor blade tip timing signal. The method selects the blade resonance zone measurement signal based on single-sensor measurement data and prior information. Based on the blade resonance zone measurement signal, two sets of specific virtual signals are generated. Two sets of dot-product signals are generated using the frequency shift principle. The characteristic frequency is extracted by the slope of the spectral difference between the dot-product signals. De-aliasing is performed using the aliasing formula and prior information to extract the blade's natural frequency. This method solves the problem of undersampling of the blade tip timing measurement signal and can quickly and accurately extract the natural frequency of the rotating blade from the aliased spectrum.

Description

Technical Field

[0001] This disclosure pertains to the field of vibration testing of rotating machinery blades, and specifically relates to a method for extracting the natural frequency for timing signal processing at the tip of a single-sensor blade. Background Technology

[0002] As one of the core components of rotating machinery, rotor blades operate under harsh conditions and are subjected to high speeds and complex loads. The operating status of the blades directly affects the performance and safety of rotating machinery. Therefore, structural health monitoring of rotor blades is of great significance.

[0003] Blade tip timing measurement technology is a non-contact online monitoring method for rotor blades. In practical applications, the installation location and number of sensors are greatly limited by factors such as the structure of rotating machinery. Minimizing the number of sensors and adopting a layout with minimal requirements becomes crucial. However, the limited sensor sampling results in undersampled blade tip timing measurement signals, leading to frequency aliasing in the spectrum. Recovering the correct blade natural frequency from the aliased spectrum of the undersampled signal presents a significant challenge. While existing blade tip timing signal natural frequency extraction schemes can extract the blade's natural frequency, they rely solely on amplitude or white noise assumptions when extracting characteristic frequency components. This fails to overcome the limitations of limited samples under high undersampling conditions and is highly susceptible to interference when the noise is not white noise.

[0004] Therefore, the problem of undersampling and insufficient samples in the timing signal at the leaf tip, while meeting the limitations of actual installation location and limited sensors, and facilitating the construction of virtual signals to avoid noise interference, is a problem that needs to be solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this disclosure is to provide a method for extracting the inherent frequency of a single-sensor blade tip timing signal. This method uses a single sensor, which meets the requirements of actual installation location and fewer sensors. It extracts the characteristic frequency of the signal using the frequency shift principle and overcomes the undersampling problem by combining prior information, thereby accurately extracting the inherent frequency information of the blade from the blade tip timing signal.

[0006] To achieve the above objectives, the present disclosure adopts the following technical solution.

[0007] A method for extracting the intrinsic frequency for single-sensor leaf tip timing signal processing, the method comprising the following steps:

[0008] Step 1: Acquire blade vibration signals using a single sensor;

[0009] Step 2: Select the measurement signal for the blade resonance zone based on the blade vibration signal and prior information: X={x(t1),x(t2),…,x(t… N )};

[0010] Step 3: Determine the selection range f of the virtual frequency based on the measured signal of the blade resonance region and the prior information. virtual and the sweep gradient Δf; and in f virtual Select two virtual frequencies f 1m and f 1n ;

[0011] Step 4: Based on the virtual frequency f 1m and f 1n Generate two sets of virtual signals Y m and Y n According to the virtual signal Y m and Y n Two sets of time-domain signals S were obtained from the measurement signal X in the blade resonance region. xym and S xyn For signal S xym and S xyn The corresponding spectrum P is obtained by performing frequency domain processing respectively. xym and P xyn ;

[0012] Step 5: Based on spectrum P xym With P xyn The spectrum P is obtained by subtraction. mn Calculate the spectrum P mn The slope at the mid-frequency point is K = {k0, k2, …, k (N / 2)-2}, and extract the three frequencies of the top three absolute values ​​of the slope: max_slope = [s1, s2, s3];

[0013] Step 6: Verify and filter the components with frequencies max_slope = [s1, s2, s3] to obtain the frequency components p = [p1, p2]; where the required component is f 1m -f n and f 1m +f n The two frequency components extracted from p must be the two folding frequencies mentioned above.

[0014] Step 7: Calculate the preliminary natural frequency f based on the frequency component p and the measured signal X in the blade resonance region, combined with prior information. n extract ;

[0015] Step 8: Based on the sweep frequency gradient Δf, let f 1m =Δf+f 1m f 1n =2Δf+f 1m ,f 1m ,f 1nAs the value used to repeat steps 4-7, repeat steps 4-7 until f is traversed. virtual For all frequency points, obtain the corresponding natural frequency f. n extract , all f n extract Represented as f n temp ;

[0016] Step 9: Filter f based on clustering principles n temp To obtain the true natural frequency f n .

[0017] Optionally, the prior information mentioned in step 2 is the approximate natural frequency of the blade, denoted as f. prior The approximate natural frequency of the blade is obtained theoretically through finite element simulation during the blade design phase, or through the blade's Campbell diagram.

[0018] Optionally, step 3 includes:

[0019] Step 3.1: Obtain the rotational speed information f based on the measured signal X from the blade resonance zone. r Combined with prior information f prior The virtual frequency selection range is determined to be f. virtual = [f prior -f r / 2, f prior +f r / 2], determine the sweep frequency gradient as Δf;

[0020] Step 3.2: Select the virtual frequency range f virtual Select virtual frequency f 1m = f prior -f r / 2, determine the amplitude A of the natural frequency component. 1m and phase φ 1m The amplitude A of the frequency component rm and phase φ rm .

[0021] Optionally, step 4 includes:

[0022] Step 4.1 Based on the following formula and the time series of the measured signal in the blade resonance region, t={t1, t2, ..., t... n Generate a set of virtual signals Y m ;

[0023]

[0024] Among them, f rmThe frequency of the frequency component represents the rotational frequency, which is determined based on the time series t in the measured signal of the blade resonance zone.

[0025] Step 4.2: Based on the virtual signal Y m virtual frequency f 1m Choose another frequency component f 1n , where f 1n =f 1m +2*resolution, where resolution is the frequency resolution of the blade measurement signal, calculated using the following formula:

[0026]

[0027] Among them, f s The sampling frequency, f, is present in the timing measurement signal at the tip of a single sensor leaf. s =f r ;

[0028] Step 4.3: According to f 1n Generate another set of virtual signals Y n The calculation formula is as follows:

[0029] ;

[0030] Step 4.4: Convert the virtual signal Y m and Y n The dot product signal S is obtained by multiplying the measured signal in the blade resonance region. xym and S xyn ;

[0031] Step 4.5: Multiply the dot product of the signal S xym and S xyn Performing discrete Fourier transforms on each spectrum P yields the corresponding spectrum P. xym and P xyn .

[0032] Optionally, step 5 includes:

[0033] Step 5.1: Spectrum P xym With P xyn By subtracting the values, we obtain the spectrum P. mn =P xym -P xyn ={p mn0 , p mn1 , p mn2 , …,p mn(N / 2)}, where when N is odd, N = N - 1, and when N is even, N = N;

[0034] Step 5.2: Calculate the slope K for the first N / 2-1 frequency points using the following formula: K = {k0, k2, …, k(N / 2)-2}:

[0035]

[0036] Where, k i This represents the slope value at the i-th frequency point, where the value of i ranges from [0, N / 2-1].

[0037] Step 5.3: Extract the three slopes with the largest absolute values ​​from the slope K in the spectrum P. xym The corresponding frequency is max_slope = [s1, s2, s3]; by controlling the number of sampling points and the sampling frequency, at least five points on the spectrum can be used to calculate the slope.

[0038] Optionally, step 6 includes:

[0039] Step 6.1: Check if the frequency max_slope contains f. 1m folding frequency f 1m折叠 If it is included, continue to the next step; if it is not included, return to step 4 and select the next f. 1m Continue calculating;

[0040] The determination criteria include: the frequency max_slope and f. 1m折叠 The absolute value of the minimum difference is less than resolution / 2; f 1m折叠 Calculated using the following formula:

[0041]

[0042] in, It is an integer;

[0043] Step 6.2: Remove f from the frequency max_slope 1m折叠 If f is removed 1m折叠 If the two frequency components of the subsequent frequency max_slope are the same, then return to step 4 and select the next f. 1m Continue the calculation; if they are different, record it as p = [p1, p2];

[0044] The specific information contained in p is represented by the following formula:

[0045]

[0046] in, It is an integer, as can be known from the prior conditions. And there are .

[0047] Optionally, step 7 includes:

[0048] Step 7.1: Calculate the preliminary blade natural frequency f according to the following formula. n possible :

[0049] ;

[0050] Step 7.2: Calculate f n possible Substitute the values ​​into the corresponding formula below for verification, and select the one that passes the verification as the calculated f. n extract If all checks pass simultaneously, record the frequency of all successful checks as f. n extract ;

[0051] .

[0052] Optionally, the clustering principle described in step 9 is the K-means clustering method, with a number of categories of 3.

[0053] Optionally, select the category containing the most frequencies; the center point of this category is the natural frequency f. n If there are multiple categories with the highest frequency, then the average of the center points of the categories that meet the conditions is taken as the intrinsic frequency f. n .

[0054] Compared with the prior art, the beneficial effects of this disclosure are as follows:

[0055] This method uses a single blade-end timing sensor, overcoming the limitations of actual installation location and layout while achieving the minimum number of sensors required. It constructs different virtual signals using the frequency shift principle and extracts characteristic frequencies through frequency shifting, solving the problem of high undersampling of blade-end timing measurement signals. It can accurately extract the natural frequency of the rotating blade from the aliased spectrum. Attached Figure Description

[0056] Figure 1 This is a flowchart of an embodiment of the present disclosure of a method for extracting the intrinsic frequency of timing signals for single-sensor leaf tip processing;

[0057] Figure 2 This is a schematic diagram of a method for extracting the intrinsic frequency of a single-sensor blade tip timing signal, provided in one embodiment of the present disclosure, for selecting the measurement signal of the blade resonance region.

[0058] Figure 3 This is a schematic diagram of the blade resonance zone measurement signal, virtual signal, and dot product signal provided in an embodiment of the present disclosure for a method of extracting the inherent frequency for single-sensor blade tip timing signal processing;

[0059] Figure 4 This is a schematic diagram of the spectrum of two dot product signals and the corresponding difference frequency and slope of an intrinsic frequency extraction method for timing signal processing at the tip of a single sensor provided in one embodiment of this disclosure.

[0060] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0061] The following will refer to the appendix. Figures 1 to 4 Specific embodiments of this disclosure are described in detail. While specific embodiments of this disclosure are shown in the accompanying drawings, it should be understood that this disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of this disclosure to those skilled in the art.

[0062] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions of preferred embodiments of this disclosure are for the purpose of implementing the general principles of the specification and are not intended to limit the scope of this disclosure. The scope of protection of this disclosure is determined by the appended claims.

[0063] To facilitate understanding of the embodiments of this disclosure, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of this disclosure.

[0064] In one embodiment, this disclosure provides a method for extracting the intrinsic frequency for single-sensor leaf tip timing signal processing, the method comprising the following steps:

[0065] Step 1: Acquire blade vibration signals using a single sensor;

[0066] Step 2: Select the measurement signal for the blade resonance zone based on the blade vibration signal and prior information: X={x(t1),x(t2),…,x(t… N )};

[0067] Step 3: Determine the selection range f of the virtual frequency based on the measured signal of the blade resonance region and the prior information. virtual and the sweep gradient Δf; and in f virtual Select two virtual frequencies f1m and f 1n ;

[0068] Step 4: Based on the virtual frequency f 1m and f 1n Generate two sets of virtual signals Y m and Y n According to the virtual signal Y m and Y n Two sets of time-domain signals S were obtained from the measurement signal X in the blade resonance region. xym and S xyn For signal S xym and S xyn The corresponding spectrum P is obtained by performing frequency domain processing respectively. xym and P xyn ;

[0069] Step 5: Based on spectrum P xym With P xyn The spectrum P is obtained by subtraction. mn Calculate the spectrum P mn The slope at the mid-frequency point is K = {k0, k2, …, k (N / 2)-2}, and extract the three frequencies of the top three absolute values ​​of the slope: max_slope = [s1, s2, s3];

[0070] Step 6: Verify and filter the components with frequencies max_slope = [s1, s2, s3] to obtain the frequency components p = [p1, p2]; where the required component is f 1m -f n and f 1m +f n The two frequency components extracted from p must be the two folding frequencies mentioned above.

[0071] Step 7: Calculate the preliminary natural frequency f based on the frequency component p and the measured signal X in the blade resonance region, combined with prior information. n extract ;

[0072] Step 8: Based on the sweep frequency gradient Δf, let f 1m =Δf+f 1m f 1n =2Δf+f 1m ,f 1m ,f 1n As the value used to repeat steps 4-7, repeat steps 4-7 until f is traversed. virtual For all frequency points, obtain the corresponding natural frequency f. n extract , all f n extract Represented as fn temp ;

[0073] Step 9: Filter f based on clustering principles n temp To obtain the true natural frequency f n .

[0074] Optionally, the prior information mentioned in step 2 is the approximate natural frequency of the blade, denoted as f. prior The approximate natural frequency of the blade is obtained theoretically through finite element simulation during the blade design phase, or through the blade's Campbell diagram.

[0075] Optionally, step 3 includes:

[0076] Step 3.1: Obtain the rotational speed information f based on the measured signal X from the blade resonance zone. r Combined with prior information f prior The virtual frequency selection range is determined to be f. virtual = [f prior -f r / 2, f prior +f r / 2], determine the sweep frequency gradient as Δf;

[0077] Step 3.2: Select the virtual frequency range f virtual Select virtual frequency f 1m = f prior -f r / 2, determine the amplitude A of the natural frequency component. 1m and phase φ 1m The amplitude A of the frequency component rm and phase φ rm .

[0078] Optionally, step 4 includes:

[0079] Step 4.1 Based on the following formula and the time series of the measured signal in the blade resonance region, t={t1, t2, ..., t... n Generate a set of virtual signals Y m ;

[0080]

[0081] Among them, f rm The frequency of the frequency component represents the rotational frequency, which is determined based on the time series t in the measured signal of the blade resonance zone.

[0082] Step 4.2: Based on the virtual signal Y m virtual frequency f 1m Choose another frequency component f 1n, where f 1n =f 1m +2*resolution, where resolution is the frequency resolution of the blade measurement signal, calculated using the following formula:

[0083]

[0084] Among them, f s The sampling frequency, f, is present in the timing measurement signal at the tip of a single sensor leaf. s =f r ;

[0085] Step 4.3: According to f 1n Generate another set of virtual signals Y n The calculation formula is as follows:

[0086] ;

[0087] Step 4.4: Convert the virtual signal Y m and Y n The dot product signal S is obtained by multiplying the measured signal in the blade resonance region. xym and S xyn ;

[0088] Step 4.5: Multiply the dot product of the signal S xym and S xyn Performing discrete Fourier transforms on each spectrum P yields the corresponding spectrum P. xym and P xyn .

[0089] Optionally, step 5 includes:

[0090] Step 5.1: Spectrum P xym With P xyn By subtracting the values, we obtain the spectrum P. mn =P xym -P xyn ={p mn0 , p mn1 , p mn2 , …,p mn(N / 2)}, where when N is odd, N = N - 1, and when N is even, N = N;

[0091] Step 5.2: Calculate the slope K for the first N / 2-1 frequency points using the following formula: K = {k0, k2, …, k (N / 2)-2}:

[0092]

[0093] Where, k i This represents the slope value at the i-th frequency point, where the value of i ranges from [0, N / 2-1].

[0094] Step 5.3: Extract the three slopes with the largest absolute values ​​from the slope K in the spectrum P. xym The corresponding frequency is max_slope = [s1, s2, s3]; by controlling the number of sampling points and the sampling frequency, at least five points on the spectrum can be used to calculate the slope.

[0095] Optionally, step 6 includes:

[0096] Step 6.1: Check if the frequency max_slope contains f. 1m folding frequency f 1m折叠 If it is included, continue to the next step; if it is not included, return to step 4 and select the next f. 1m Continue calculating;

[0097] The determination criteria include: the frequency max_slope and f. 1m折叠 The absolute value of the minimum difference is less than resolution / 2; f 1m折叠 Calculated using the following formula:

[0098]

[0099] in, It is an integer;

[0100] Step 6.2: Remove f from the frequency max_slope 1m折叠 If f is removed 1m折叠 If the two frequency components of the subsequent frequency max_slope are the same, then return to step 4 and select the next f. 1m Continue the calculation; if they are different, record it as p = [p1, p2];

[0101] The specific information contained in p is represented by the following formula:

[0102]

[0103] in, It is an integer, as can be known from the prior conditions. And there are .

[0104] Optionally, step 7 includes:

[0105] Step 7.1: Calculate the preliminary blade natural frequency f according to the following formula. n possible :

[0106] ;

[0107] Step 7.2: Calculate f n possible Substitute the values ​​into the corresponding formula below for verification, and select the one that passes the verification as the calculated f. n extract If all checks pass simultaneously, record the frequency of all successful checks as f. n extract ;

[0108] .

[0109] Optionally, the clustering principle described in step 9 is the K-means clustering method, with a number of categories of 3.

[0110] Optionally, select the category containing the most frequencies; the center point of this category is the natural frequency f. n If there are multiple categories with the highest frequency, then the average of the center points of the categories that meet the conditions is taken as the intrinsic frequency f. n .

[0111] In one embodiment, this disclosure provides a method for extracting the intrinsic frequency for single-sensor leaf tip timing signal processing, the method comprising the following steps:

[0112] Step 1: Use a single sensor to acquire blade vibration signals; the sensor is arranged at an arbitrary angle on the casing;

[0113] Step 2: Select the blade resonance zone measurement signal X={x(t1), x(t2),…, x(t3)} based on the blade vibration signal and prior information. N )}, where the selected blade resonance region is as follows Figure 2 As shown, selection is based on the following principles: ① The vibration waveform undergoes a sudden change; ② The vibration amplitude suddenly increases.

[0114] Step 3: Based on the measurement signal of the blade resonance region, the number of sampling points is N=80. Given the prior information f... prior =320Hz, rotational speed frequency f r =79.3983Hz, then the virtual frequency selection range is f virtual = [280.3009, 359.6991]Hz, for convenience in taking f virtual = [281, 359]Hz, sweep gradient Δf = 1Hz. Frequency resolution resolution = 0.9925Hz, choose f 1m and f 1n The frequencies are 281Hz and 282.9855Hz, respectively.

[0115] Step 4: Generate virtual signals Y1 and Y2, where virtual signal Y1 is generated using the following formula:

[0116]

[0117] The virtual signal Y2 is generated using the following formula:

[0118]

[0119] Where f rm The Hz value of the measured signal X in the blade resonance region is 79.3983 Hz, and the sampling time t is consistent with that of the measured signal X. The generated virtual signal is as follows: Figure 3 As shown.

[0120] The dot product signal S is obtained by multiplying the virtual signals Y1 and Y2 with the measured signal X in the blade resonance region. xy1 and S xy2 ,like Figure 3 As shown; for the dot product signal S xy1 and S xy2 Perform discrete Fourier transforms on each to obtain the corresponding spectrum P. xy1 and P xy2 ,like Figure 4 As shown.

[0121] Step 5: Spectrum P xy1 and P xy2 By subtracting the frequencies, we obtain the difference frequency spectrum P. 12 ,like Figure 4 As shown. Calculate the slope of the first 39 frequency points, and extract the three frequencies with the largest absolute slope values: max_slope = [36.7251, 32.7548, 36.7251] Hz.

[0122] Step 6: Verify that max_slope contains f 1m Folded components f 1m折叠 :

[0123]

[0124] Then max_slope contains f 1m折叠 However, the remaining two frequency components still contain an f. 1m折叠 If the frequency components are the same, return to step 3 and let f 1m =f 1m +Δf=282Hz Continue the calculation until f 1m When the frequency is 312Hz, the remaining two frequency components are different, denoted as p=[4.9629, 5.9554]Hz.

[0125] Step 7: Calculate the possible blade natural frequencies f using the following formula. n possible :

[0126]

[0127] Verification f n possible :

[0128] At this point, record f n extract =[f n extract ,316.9629, 317.9554]Hz.

[0129] Step 8: Return to step 4 and let f 1m =f 1m +Δf=282Hz Continue calculating until f is traversed virtual Finally, all possible natural frequencies are obtained and denoted as f. n extract =[ 317.924786400757,

[0130] 316.932307760681, 276.157948558410, 280.127863118713,

[0131] 316.939829120606, 317.932307760681, 317.939829120606,

[0132] 316.947350480530, 316.954871840454, 317.947350480530,

[0133] 317.954871840454, 316.962393200379, 317.962393200379,

[0134] 316.969914560303, 316.984957280151, 317.977435920227,

[0135] 321.969914560303, 313.037606799621, 320.984957280151,

[0136] 314.037606799621, 317.022564079773, 318.015042719849,

[0137] 317.030085439697, 318.022564079773, 317.037606799621,

[0138] 318.030085439697, 317.045128159546, 318.037606799621,

[0139] 317.060170879394, 318.052649519470, 317.067692239319,

[0140] 318.060170879394, 317.075213599243, 318.067692239319,

[0141] 317.090256319091, 318.082734959167, 317.278290317198,

[0142] 318.270768957274,317.285811677123,318.278290317198]Hz.

[0143] Step 9: Set the number of categories to 3, and use the K-means clustering method. The center values ​​of the three categories are [318.395339603940, 278.142905838561, 16.671560913282]Hz, and the number of clusters is [19, 2, 19]. Then f n =(316.671560913282+318.395339603940) / 2 = 317.533450258611 Hz.

[0144] The above methods can be used to recover the true natural frequency components of aliasing. By analyzing the changes in natural frequencies, the working status of the blade can be determined, thereby realizing blade status monitoring.

[0145] This disclosure uses only one blade tip timing sensor, which overcomes the limitations of actual installation location and layout while achieving the minimum number of sensors used. It constructs different virtual signals using the frequency shift principle, extracts characteristic frequencies by frequency shifting, solves the frequency aliasing problem caused by undersampling, and extracts the blade's inherent frequency.

[0146] In one embodiment, this disclosure provides a computer-readable storage medium for storing a computer program configured to implement the method when invoked by a processor.

[0147] In one embodiment, this disclosure provides an electronic device, the electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor; wherein the processor implements the method when executing the program.

[0148] This invention is not limited to the embodiments described above. The above description of specific embodiments is intended to illustrate and explain the technical solutions of this invention. The specific embodiments described above are merely illustrative and not restrictive. Without departing from the spirit and scope of the claims, those skilled in the art can make many specific modifications based on the teachings of this invention, and these modifications all fall within the scope of protection of this invention.

Claims

1. A method for extracting the intrinsic frequency for timing signal processing at the tip of a single sensor leaf, wherein, The method includes the following steps: Step 1: Acquire blade vibration signals using a single sensor; Step 2: Select the measurement signal for the blade resonance zone based on the blade vibration signal and prior information: X={x(t1), x(t2),…, x(t… N )}; Step 3: Determine the selection range f of the virtual frequency based on the measured signal of the blade resonance region and the prior information. virtual and the sweep gradient Δf; and in f virtual Select two virtual frequencies f 1m and f 1n ; Step 4: Based on the virtual frequency f 1m and f 1n Generate two sets of virtual signals Y m and Y n According to the virtual signal Y m and Y n Two sets of time-domain signals S were obtained from the measurement signal X in the blade resonance region. xym and S xyn For signal S xym and S xyn The corresponding spectrum P is obtained by performing frequency domain processing respectively. xym and P xyn ; Step 5: Based on spectrum P xym With P xyn The spectrum P is obtained by subtraction. mn Calculate the spectrum P mn The slope at the mid-frequency point is K = {k0, k2, ..., k} (N / 2)-2 }, and extract the three frequencies of the top three absolute values ​​of the slope: max_slope = [s1, s2, s3]; Step 6: Verify and filter the components with frequencies max_slope = [s1, s2, s3] to obtain the frequency components p = [p1, p2]. Step 7: Calculate the preliminary natural frequency f based on the frequency component p and the measured signal X in the blade resonance region, combined with prior information. n extract ; Step 8: Based on the sweep frequency gradient Δf, let f 1m =Δf+f 1m f 1n =2Δf+f 1m ;f 1m ,f 1n As the value used to repeat steps 4-7, repeat steps 4-7 until f is traversed. virtual For all frequency points, obtain the corresponding natural frequency f. n extract , all f n extract Represented as f n temp ; Step 9: Filter f based on clustering principles n temp To obtain the true natural frequency f n .

2. The method according to claim 1, wherein, The prior information mentioned in step 2 is the approximate natural frequency of the blade, denoted as f. prior .

3. The method according to claim 2, wherein, Step 3 includes: Step 3.1: Obtain the rotational speed information f based on the measured signal X from the blade resonance zone. r Combined with prior information f prior The virtual frequency selection range is determined to be f. virtual = [f prior -f r / 2, f prior +f r / 2], determine the sweep frequency gradient as Δf; Step 3.2: Select the virtual frequency range f virtual Select virtual frequency f 1m = f prior -f r / 2, determine the amplitude A of the natural frequency component. 1m and phase The amplitude A of the frequency component rm and phase .

4. The method according to claim 3, wherein, Step 4 includes: Step 4.1 Based on the following formula and the time series of the measured signal in the blade resonance region, t={t1, t2, ..., t... n Generate a set of virtual signals Y m ; ； Among them, f rm The frequency of the frequency component represents the rotational frequency, which is determined based on the time series t in the measured signal of the blade resonance zone. Step 4.2: Based on the virtual signal Y m virtual frequency f 1m Choose another frequency component f 1n , where f 1n =f 1m +2*resolution, where resolution is the frequency resolution of the blade measurement signal, calculated using the following formula: ； Among them, f s f represents the sampling frequency, which is present in the timing measurement signal at the tip of a single sensor leaf. s =f r ; Step 4.3: According to f 1n Generate another set of virtual signals Y n The calculation formula is as follows: ； Step 4.4: Convert the virtual signal Y m and Y n The dot product signal S is obtained by multiplying the measured signal in the blade resonance region. xym and S xyn ; Step 4.5: Multiply the dot product of the signal S xym and S xyn Performing discrete Fourier transforms on each spectrum P yields the corresponding spectrum P. xym and P xyn .

5. The method according to claim 4, wherein, Step 5 includes: Step 5.1: Spectrum P xym With P xyn By subtracting the values, we obtain the spectrum P. mn =P xym -P xyn ={p mn0 , p mn1 , p mn2 , …,p mn(N / 2) }, where when N is odd, N = N - 1, and when N is even, N = N; Step 5.2: Calculate the slope K for the first N / 2-1 frequency points using the following formula: K = {k0, k2, …, k (N / 2)-2 }: ； Where, k i This represents the slope value at the i-th frequency point, where the value of i ranges from [0, N / 2-1]. Step 5.3: Extract the three slopes with the largest absolute values ​​from the slope K in the spectrum P. xym The corresponding frequencies are max_slope = [s1, s2, s3].

6. The method according to claim 5, wherein, Step 6 includes: Step 6.1: Check if the frequency max_slope contains f. 1m folding frequency f 1m折叠 If it is included, continue to the next step; if it is not included, return to step 4 and select the next f. 1m Continue calculating; The determination criteria include: the frequency max_slope and f. 1m折叠 The absolute value of the minimum difference is less than resolution / 2; f 1m折叠 Calculated using the following formula: ； in, It is an integer; Step 6.2: Remove f from the frequency max_slope 1m折叠 If f is removed 1m折叠 If the two frequency components of the subsequent frequency max_slope are the same, then return to step 4 and select the next f. 1m Continue the calculation; if they are different, record it as p = [p1, p2]; The specific information contained in p is represented by the following formula: ； in, It is an integer, as can be known from the prior conditions. And there are 。 7. The method according to claim 6, wherein, Step 7 includes: Step 7.1: Calculate the preliminary blade natural frequency f according to the following formula. n possible : ； Step 7.2: Calculate f n possible Substitute the values ​​into the corresponding formula below for verification, and select the one that passes the verification as the calculated f. n extract If all checks pass simultaneously, record the frequency of all successful checks as f. n extract ; 。 8. The method according to claim 1, wherein, The clustering principle described in step 9 is the K-means clustering method, with a number of 3 clusters.

9. The method according to claim 8, wherein, Select the category containing the most frequencies; the center point of this category is the natural frequency f. n If there are multiple categories with the highest frequency, then the average of the center points of the categories that meet the conditions is taken as the intrinsic frequency f. n .

Images