A method and system for evaluating the narrow-band resonant stress response of thin-walled stator components

Abstract

This invention relates to the field of aero-engine technology and discloses a method and system for evaluating the narrow-band resonant stress response of thin-walled stator components. By filtering the spectral signal to eliminate the influence of background noise on the response evaluation, the vibration response is divided into several frequency bands with larger vibration responses in the frequency domain according to the magnitude of the vibration response. The vibration response of each frequency band is analyzed separately, and the responses of each frequency band are then superimposed based on mode shape superposition and probability distribution models to finally obtain the vibration response of the entire frequency band. This invention enables effective evaluation of the strain of thin-walled stator components when they vibrate simultaneously in multiple narrow bands, providing fundamental support for the high-cycle fatigue design of thin-walled stator components.

Description

Technical Field

[0001] This invention relates to the field of aero-engine technology and discloses a method and system for evaluating the narrow-band resonant stress response of thin-walled stator components. Background Technology

[0002] Thin-walled stator components are extensively used in the afterburner and nozzle parts of aero-engines. These components are subjected to extremely high temperature loads, as well as various vibration excitation loads including airflow pressure fluctuations, noise, and the fundamental frequency of the engine rotor. Due to the relatively weak structural rigidity of thin-walled stator components and their large-sized, continuous circular structure, they exhibit an extremely rich variety of modes. It is difficult to completely avoid these excitation frequencies during the design process, leading to resonance problems.

[0003] Unlike the single-frequency vibration response of engine blades, the vibration response of thin-walled stator structures exhibits a distinct narrow-band resonance characteristic and lacks a clear dominant frequency characteristic. Because blade vibration has a pronounced single-frequency characteristic, the maximum vibration response point is typically used directly for response evaluation. However, for the narrow-band resonance characteristics of thin-walled stator components, if the single-frequency response is used as a reference for blade vibration evaluation, the vibration response at other frequencies will be ignored, resulting in a significantly smaller vibration response than the actual result. Conversely, if the total vibration is used directly for evaluation, without effective filtering, the calculated vibration response will be far greater than the actual response, which is detrimental to structural design. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for evaluating the stress response of thin-walled stator components under narrow-band resonance, which can effectively evaluate the strain of thin-walled stator components when they vibrate simultaneously under multiple narrow bands, providing fundamental support for the high-cycle fatigue design of thin-walled stator components.

[0005] To achieve the above-mentioned technical effects, the technical solution adopted by the present invention is as follows:

[0006] A method for evaluating the narrow-band resonant stress response of a thin-walled stator includes:

[0007] Spectral analysis was performed on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data;

[0008] Background noise is eliminated from the vibration response data to construct a noise-reduced three-dimensional waterfall plot; a spectrum of the time point corresponding to the maximum strain amplitude is extracted from the three-dimensional waterfall plot, and local peak points are extracted from the spectrum plot.

[0009] The local peak point that is greater than a preset peak point and whose absolute value of the frequency difference between two adjacent local peak points is greater than the preset frequency difference is determined as the dominant frequency of the frequency band;

[0010] Centered on each main frequency, extending outwards on both sides of the spectrum, the boundary frequency values ​​at which the strain amplitude is less than a preset amplitude at n consecutive frequencies are determined as the frequency band boundary of the corresponding main frequency, forming the effective frequency band of the corresponding main frequency;

[0011] In the three-dimensional waterfall plot, an inverse Fourier transform is performed on the strain amplitude in each effective frequency band to obtain a time-domain function of the total strain amplitude of a single effective frequency band.

[0012] The strain amplitude function of all effective frequency bands is superimposed to obtain the strain amplitude function in the full frequency domain; the root mean square of the strain amplitude in the full frequency domain per unit time is calculated based on the strain amplitude function in the full frequency domain, the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time is statistically analyzed, and the +3σ value of the strain amplitude in the full frequency domain is determined to be the multi-narrowband resonant response value.

[0013] Furthermore, the method for background noise removal of the vibration response data includes:

[0014] Based on the vibration response data, non-vibration frequencies whose strain amplitude is less than the first preset strain threshold are selected throughout the entire working time.

[0015] Calculate the average value μ and the (μ+3σ) value of all non-vibration frequency strain amplitudes. If the average value μ is less than the second preset strain threshold, the average value μ is determined to be the background noise amplitude. If the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is determined to be the background noise amplitude.

[0016] Subtract the background noise amplitude from the strain amplitude at each frequency to obtain the vibration response signal after background noise elimination.

[0017] Furthermore, the first preset strain threshold is 0.1 times the maximum strain amplitude, and the second preset strain threshold is 0.05 times the maximum strain amplitude.

[0018] Furthermore, the preset peak value is 0.01 to 0.2 times the maximum value of the strain amplitude.

[0019] Furthermore, before calculating the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, the ratio of the total strain amplitude of each effective frequency band to the vibration response amplitude in the full frequency domain is calculated in advance. Effective frequency bands with a ratio less than a preset ratio value are deleted, and then the probability density function of the root mean square of the vibration response per unit time is statistically analyzed.

[0020] To achieve the above-mentioned technical effects, the present invention also provides a narrow-band resonant stress response evaluation system for thin-walled stators, used to implement the aforementioned narrow-band resonant stress response evaluation method for thin-walled stators, comprising:

[0021] The data acquisition module is used to perform spectral analysis on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data;

[0022] The preprocessing module is used to remove background noise from the vibration response data and construct a noise-reduced three-dimensional waterfall plot; to extract the spectrum of the time point corresponding to the maximum strain amplitude from the three-dimensional waterfall plot, and to extract local peak points from the spectrum plot.

[0023] The main frequency determination module is used to determine the main frequency of the frequency band as the local peak point where the local peak point is greater than a preset peak point and the absolute value of the frequency difference between two adjacent local peak points is greater than a preset frequency difference.

[0024] The effective frequency band determination module is used to determine the frequency band boundary of the corresponding main frequency by extending to both sides of the spectrum graph with each main frequency as the center, and determining the boundary frequency value when the strain amplitude at n consecutive frequencies is less than the preset amplitude, thus forming the effective frequency band of the corresponding main frequency.

[0025] The conversion module is used to perform inverse Fourier transform on the strain amplitude in each effective frequency band in the three-dimensional waterfall plot to obtain a time-domain function of the total strain amplitude of a single effective frequency band.

[0026] The response analysis module is used to superimpose the total strain amplitude function of all effective frequency bands to obtain the strain amplitude function in the full frequency domain; calculate the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, statistically analyze the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time, and determine the +3σ value of the strain amplitude in the full frequency domain as the multi-narrowband resonant response value.

[0027] Furthermore, the preprocessing module also includes:

[0028] The frequency filtering unit is used to filter out non-vibration frequencies whose strain amplitude is less than a first preset strain threshold throughout the entire working time based on the vibration response data.

[0029] The strain amplitude analysis unit is used to calculate the average value μ and the (μ+3σ) value of the strain amplitude of all non-vibration frequencies. If the average value μ is less than the second preset strain threshold, the average value μ is determined to be the background noise amplitude. If the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is determined to be the background noise amplitude.

[0030] The noise reduction unit is used to subtract the background noise amplitude from the strain amplitude at each frequency to obtain the vibration response signal after background noise is eliminated.

[0031] Furthermore, in the frequency filtering unit, the first preset strain threshold is 0.1 times the maximum strain amplitude; in the strain amplitude analysis unit, the second preset strain threshold is 0.05 times the maximum strain amplitude.

[0032] Furthermore, in the main frequency determination module, the preset peak value is 0.01 to 0.2 times the maximum value of the strain amplitude.

[0033] Furthermore, before calculating the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, the response analysis module pre-calculates the ratio of the total strain amplitude of each effective frequency band to the vibration response amplitude in the full frequency domain, deletes effective frequency bands with a ratio less than a preset ratio value, and then statistically analyzes the probability density function of the root mean square of the vibration response in each unit time.

[0034] Compared with the prior art, the beneficial effects of this invention are as follows: This invention filters the spectral signal to eliminate the influence of background noise on the response evaluation, and then divides the vibration response into several frequency bands with larger vibration responses in the frequency domain according to the magnitude of the vibration response. By analyzing the vibration response of each frequency band separately, and then superimposing the responses of each frequency band based on the mode shape superposition and probability distribution model, the vibration response of the entire frequency band is finally obtained. This enables effective evaluation of the strain of thin-walled stator components when they vibrate simultaneously in multiple narrow bands, providing basic support for the high-cycle fatigue design of thin-walled stator components. Attached Figure Description

[0035] Figure 1 This is a flowchart of the narrow-band resonant stress response evaluation method for thin-walled stator components in Example 1 or 2;

[0036] Figure 2 This is a block diagram of the narrow-band resonant stress response evaluation system for thin-walled stator components in Example 1;

[0037] Figure 3 This is the vibration response spectrum of the thin-walled stator before background noise elimination in Example 2;

[0038] Figure 4 This is the vibration response spectrum of the thin-walled stator after background noise elimination in Example 2;

[0039] Figure 5 This is a partial peak function graph of the spectrum of the thin-walled stator in Example 2;

[0040] Figure 6 This is the effective vibration frequency band diagram determined in Example 2;

[0041] Figure 7 This is the probability density distribution diagram of the vibration response in Example 2.

[0042] The module includes: 1. Data acquisition module; 2. Preprocessing module; 3. Main frequency determination module; 4. Effective frequency band determination module; 5. Conversion module; 6. Response analysis module; 201. Frequency screening unit; 202. Strain amplitude analysis unit; 203. Noise reduction unit. Detailed Implementation

[0043] The present invention will now be described in further detail with reference to the embodiments and accompanying drawings. However, this should not be construed as limiting the scope of the above-described subject matter of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0044] Example 1

[0045] See Figure 1 and Figure 2 A method for evaluating the narrow-band resonant stress response of a thin-walled stator, comprising:

[0046] Spectral analysis was performed on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data;

[0047] Background noise is eliminated from the vibration response data to construct a noise-reduced three-dimensional waterfall plot; a spectrum of the time point corresponding to the maximum strain amplitude is extracted from the three-dimensional waterfall plot, and local peak points are extracted from the spectrum plot.

[0048] The local peak point that is greater than a preset peak point and whose absolute value of the frequency difference between two adjacent local peak points is greater than the preset frequency difference is determined as the dominant frequency of the frequency band;

[0049] Centered on each main frequency, extending outwards on both sides of the spectrum, the boundary frequency values ​​at which the strain amplitude is less than a preset amplitude at n consecutive frequencies are determined as the frequency band boundary of the corresponding main frequency, forming the effective frequency band of the corresponding main frequency;

[0050] In the three-dimensional waterfall plot, an inverse Fourier transform is performed on the strain amplitude in each effective frequency band to obtain a time-domain function of the total strain amplitude of a single effective frequency band.

[0051] The strain amplitude function of all effective frequency bands is superimposed to obtain the strain amplitude function in the full frequency domain; the root mean square of the strain amplitude in the full frequency domain per unit time is calculated based on the strain amplitude function in the full frequency domain, the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time is statistically analyzed, and the +3σ value of the strain amplitude in the full frequency domain is determined to be the multi-narrowband resonant response value.

[0052] In this embodiment, the background noise signal is filtered to eliminate its influence on the response evaluation. Then, based on the magnitude of the vibration response, the vibration response is divided into several frequency bands with larger vibration responses in the frequency domain. The vibration response of each frequency band is analyzed separately, and the responses of each frequency band are superimposed based on the mode shape superposition and probability distribution model to finally obtain the vibration response of the entire frequency band. This enables effective evaluation of the strain of thin-walled stator components when they vibrate simultaneously in multiple narrow bands, providing fundamental support for the high-cycle fatigue design of thin-walled stator components.

[0053] Based on the same inventive concept, this embodiment also provides a narrow-band resonant stress response evaluation system for thin-walled stators, used to implement the aforementioned narrow-band resonant stress response evaluation method for thin-walled stators, including:

[0054] Data acquisition module 1 is used to perform spectral analysis on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data;

[0055] Preprocessing module 2 is used to eliminate background noise from the vibration response data and construct a noise-reduced three-dimensional waterfall plot; to extract the spectrum of the time point corresponding to the maximum strain amplitude from the three-dimensional waterfall plot, and to extract local peak points from the spectrum plot;

[0056] The main frequency determination module 3 is used to determine the main frequency of the frequency band as the local peak point where the local peak point is greater than the preset peak point and the absolute value of the frequency difference between two adjacent local peak points is greater than the preset frequency difference.

[0057] The effective frequency band determination module 4 is used to determine the frequency band boundary of the corresponding main frequency by extending to both sides of the spectrum graph with each main frequency as the center, and determining the boundary frequency value when the strain amplitude at n consecutive frequencies is less than the preset amplitude, thus forming the effective frequency band of the corresponding main frequency.

[0058] The conversion module 5 is used to perform inverse Fourier transform on the strain amplitude in each effective frequency band in the three-dimensional waterfall plot to obtain a time-domain function of the total strain amplitude of a single effective frequency band.

[0059] The response analysis module 6 is used to superimpose the total strain amplitude function of all effective frequency bands to obtain the strain amplitude function in the full frequency domain; calculate the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, statistically analyze the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time, and determine the +3σ value of the strain amplitude in the full frequency domain as the multi-narrowband resonant response value.

[0060] In this embodiment, the preprocessing module 2 further includes:

[0061] The frequency filtering unit 201 is used to filter out non-vibration frequencies whose strain amplitude is less than a first preset strain threshold throughout the entire working time based on the vibration response data.

[0062] The strain amplitude analysis unit 202 is used to calculate the average value μ and the (μ+3σ) value of the strain amplitude of all non-vibration frequencies. If the average value μ is less than the second preset strain threshold, the average value μ is determined to be the background noise amplitude. If the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is determined to be the background noise amplitude.

[0063] The noise reduction unit 203 is used to subtract the background noise amplitude from the strain amplitude at each frequency to obtain the vibration response signal after background noise is eliminated.

[0064] Example 2

[0065] See Figure 1 , Figures 3 to 7 A method for evaluating the narrow-band resonant stress response of a thin-walled stator, comprising:

[0066] Step 1: Perform spectral analysis on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data;

[0067] In this embodiment, the vibration original signal of the thin-walled stator (such as...) is analyzed. Figure 3 Spectral analysis was performed to determine the maximum value of the vibration stress response (i.e., strain) of the thin-walled stator across the entire frequency range under operating conditions. This provides a benchmark for subsequent vibration signal noise reduction and determination of effective bandwidth.

[0068] Step 2: Remove background noise from the vibration response data and construct a noise-reduced three-dimensional waterfall plot; extract the spectrum of the time point corresponding to the maximum strain amplitude from the three-dimensional waterfall plot, and extract local peak points from the spectrum plot;

[0069] In this embodiment, the frequency spectrum variation curves over the entire working time are first analyzed to identify frequencies where the strain remains essentially unchanged throughout the working time. These frequencies are derived from background noise and other signal interference. The calculation is first performed using the formula... Calculate the non-vibration frequency; where For frequency Maximum amplitude at time; For frequency Minimum amplitude at time; This is the threshold coefficient for non-vibration frequencies. The value of is positively correlated with the calculated result of the threshold. In this embodiment, the calculation... Take 0.1, if the frequency The strain change is less than the first preset strain threshold. Then the frequency This is denoted as the non-vibration frequency.

[0070] Background noise is a relatively uniform response signal over a wide frequency domain. For the selected non-vibration frequencies, the average value μ and the (μ+3σ) value of the amplitudes of all non-vibration frequencies are calculated. If the average value μ is less than a second preset strain threshold, the average value μ is selected as the background noise level; if the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is selected as the background noise amplitude. In this embodiment, the second preset strain threshold is 0.05. .

[0071] After determining the background noise level, the actual vibration response is separated from the background noise across the entire frequency domain. This involves subtracting the background noise amplitude from the vibration response amplitude at each frequency to obtain the vibration response signal after background noise removal. If the strain amplitude is negative during the calculation, the amplitude is set to zero. Figure 4 The vibration response spectrum of the thin-walled stator after background noise elimination is presented.

[0072] Step 3: Determine the local peak point where the local peak point is greater than the preset peak point and the absolute value of the frequency difference between two adjacent local peak points is greater than the preset frequency difference value as the dominant frequency of the frequency band;

[0073] This embodiment takes the moment corresponding to the maximum strain amplitude as the analysis object, and selects the amplitude at a certain frequency that satisfies the equation. The data points are taken as local peak points; in the formula For frequency The strain amplitude at that time To improve the frequency accuracy of spectrum analysis, and The frequencies are respectively and The strain amplitude at that time. Figure 5 The peak function graph of the spectrum of the thin-walled stator in this embodiment is given, and the analysis object is... Figure 4 The spectrum diagram in the image.

[0074] Further filtering is performed on the selected local peak points. When the amplitude of a local peak point is less than the preset peak value, the data point is deleted. At the same time, if the frequency difference between two consecutive peak points is less than the preset frequency difference, the point with the smaller value among the two peak points is deleted. The remaining points are the main frequencies of each frequency band.

[0075] In this embodiment, the preset peak value is 0.01 to 0.2 times the maximum value of the strain amplitude; the preset frequency difference is determined according to the requirements.

[0076] Step 4: Taking each main frequency as the center, extend to both sides on the spectrum diagram. The boundary frequency values ​​at which the strain amplitude is less than the preset amplitude at n consecutive frequencies are determined as the frequency band boundary of the corresponding main frequency, forming the effective frequency band of the corresponding main frequency.

[0077] In this embodiment, the search proceeds outwards in the spectrum. When the amplitude at n consecutive frequencies is less than a preset amplitude, the judgment is terminated. The frequency value at this point is the left or right boundary of the frequency band. During the analysis, there may be cases where frequency bands of different main frequencies are connected. In such cases, these frequency bands need to be merged and connected to form the effective frequency band of the corresponding main frequency. Figure 6 The effective vibration frequency band diagram determined by the analysis is presented.

[0078] Step 5: Perform an inverse Fourier transform on the strain amplitude in each effective frequency band in the three-dimensional waterfall plot to obtain the total strain amplitude function of a single effective frequency band based on the time domain.

[0079] In this embodiment, the response within each effective frequency band is determined according to... Performing an inverse Fourier transform yields the total vibration response of a single frequency band. ;in This corresponds to the left boundary frequency of the effective frequency band. This corresponds to the right boundary frequency of the effective frequency band. For frequency The strain amplitude at that time For frequency The vibration phase angle at that time.

[0080] Step 6: Superimpose the total strain amplitude functions of all effective frequency bands to obtain the strain amplitude function in the full frequency domain;

[0081] In this step, the calculation method for the strain amplitude function in the full frequency domain is the same as in step five, except that the calculation range is expanded from a single frequency band to the entire frequency domain, and the filtered signal is analyzed.

[0082] Step 7: Calculate the ratio of the total strain amplitude of each effective frequency band to the vibration response amplitude in the entire frequency domain, and delete effective frequency bands with a ratio less than the preset ratio value;

[0083] In this embodiment, in addition to calculating the total amount, it is also necessary to compare the proportion of vibration response under different frequency bands. For frequency bands with a small proportion, they need to be cropped in subsequent analysis to improve analysis efficiency.

[0084] Step 8: Calculate the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, statistically analyze the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time, and determine the +3σ value of the strain amplitude in the full frequency domain (i.e., the average value plus 3 times the standard deviation) as the multi-narrowband resonant response value.

[0085] In this embodiment, after obtaining the time-domain results of the vibration, calculations are performed with a period of Δt per unit time. root mean square Calculate the probability density function of the root mean square of the vibration response for each unit time interval, and determine the +3σ value of the vibration response. . This is the calculated narrowband resonant stress response value. Figure 7 The probability density distribution of the vibration response is given.

[0086] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the narrow-band resonant stress response of a thin-walled stator, characterized in that, include: Spectral analysis was performed on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data; Background noise is eliminated from the vibration response data to construct a noise-reduced three-dimensional waterfall plot; a spectrum of the time point corresponding to the maximum strain amplitude is extracted from the three-dimensional waterfall plot, and local peak points are extracted from the spectrum plot. The local peak point that is greater than a preset peak point and whose absolute value of the frequency difference between two adjacent local peak points is greater than the preset frequency difference is determined as the dominant frequency of the frequency band; Centered on each main frequency, extending outwards on both sides of the spectrum, the boundary frequency values ​​at which the strain amplitude is less than a preset amplitude at n consecutive frequencies are determined as the frequency band boundary of the corresponding main frequency, forming the effective frequency band of the corresponding main frequency; In the three-dimensional waterfall plot, an inverse Fourier transform is performed on the strain amplitude in each effective frequency band to obtain a time-domain function of the total strain amplitude of a single effective frequency band. The strain amplitude function of all effective frequency bands is superimposed to obtain the strain amplitude function in the full frequency domain; the root mean square of the strain amplitude in the full frequency domain per unit time is calculated based on the strain amplitude function in the full frequency domain, the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time is statistically analyzed, and the +3σ value of the strain amplitude in the full frequency domain is determined to be the multi-narrowband resonant response value.

2. The method for evaluating the narrow-band resonant stress response of a thin-walled stator according to claim 1, characterized in that, Methods for background noise removal of the vibration response data include: Based on the vibration response data, non-vibration frequencies whose strain amplitude is less than the first preset strain threshold are selected throughout the entire working time. Calculate the average value μ and the (μ+3σ) value of all non-vibration frequency strain amplitudes. If the average value μ is less than the second preset strain threshold, the average value μ is determined to be the background noise amplitude. If the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is determined to be the background noise amplitude. Subtract the background noise amplitude from the strain amplitude at each frequency to obtain the vibration response signal after background noise elimination.

3. The method for evaluating the narrow-band resonant stress response of a thin-walled stator according to claim 2, characterized in that, The first preset strain threshold is 0.1 times the maximum strain amplitude, and the second preset strain threshold is 0.05 times the maximum strain amplitude.

4. The method for evaluating the narrow-band resonant stress response of a thin-walled stator according to claim 1, characterized in that, The preset peak value is 0.01 to 0.2 times the maximum value of the strain amplitude.

5. The method for evaluating the narrow-band resonant stress response of a thin-walled stator according to claim 1, characterized in that, Before calculating the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, the ratio of the total strain amplitude of each effective frequency band to the vibration response amplitude in the full frequency domain is calculated in advance. Effective frequency bands with a ratio less than the preset ratio are deleted, and then the probability density function of the root mean square of the vibration response is statistically analyzed for each unit time.

6. A narrow-band resonant stress response evaluation system for thin-walled stators, used to implement the narrow-band resonant stress response evaluation method for thin-walled stators as described in claim 1, characterized in that, include: The data acquisition module is used to perform spectral analysis on the vibration response data of the thin-walled stator to obtain the maximum strain amplitude in the vibration response data; The preprocessing module is used to remove background noise from the vibration response data and construct a noise-reduced three-dimensional waterfall plot. A spectrum of the time point corresponding to the maximum strain amplitude is extracted from the three-dimensional waterfall plot, and local peak points are extracted from the spectrum plot. The main frequency determination module is used to determine the main frequency of the frequency band as the local peak point where the local peak point is greater than a preset peak point and the absolute value of the frequency difference between two adjacent local peak points is greater than a preset frequency difference. The effective frequency band determination module is used to determine the frequency band boundary of the corresponding main frequency by extending to both sides of the spectrum graph with each main frequency as the center, and determining the boundary frequency value when the strain amplitude at n consecutive frequencies is less than the preset amplitude, thus forming the effective frequency band of the corresponding main frequency. The conversion module is used to perform inverse Fourier transform on the strain amplitude in each effective frequency band in the three-dimensional waterfall plot to obtain a time-domain function of the total strain amplitude of a single effective frequency band. The response analysis module is used to superimpose the total strain amplitude function of all effective frequency bands to obtain the strain amplitude function in the full frequency domain; calculate the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, statistically analyze the probability density function of the root mean square of the strain amplitude in the full frequency domain per unit time, and determine the +3σ value of the strain amplitude in the full frequency domain as the multi-narrowband resonant response value.

7. The narrow-band resonant stress response evaluation system for thin-walled stator components according to claim 6, characterized in that, The preprocessing module further includes: The frequency filtering unit is used to filter out non-vibration frequencies whose strain amplitude is less than a first preset strain threshold throughout the entire working time based on the vibration response data. The strain amplitude analysis unit is used to calculate the average value μ and the (μ+3σ) value of the strain amplitude of all non-vibration frequencies. If the average value μ is less than the second preset strain threshold, the average value μ is determined to be the background noise amplitude. If the average value μ is greater than the second preset strain threshold, the (μ+3σ) value is determined to be the background noise amplitude. The noise reduction unit is used to subtract the background noise amplitude from the strain amplitude at each frequency to obtain the vibration response signal after background noise is eliminated.

8. The narrow-band resonant stress response evaluation system for thin-walled stator components according to claim 7, characterized in that, In the frequency filtering unit, the first preset strain threshold is 0.1 times the maximum strain amplitude; in the strain amplitude analysis unit, the second preset strain threshold is 0.05 times the maximum strain amplitude.

9. The narrow-band resonant stress response evaluation system for thin-walled stator components according to claim 6, characterized in that, In the main frequency determination module, the preset peak value is 0.01 to 0.2 times the maximum value of the strain amplitude.

10. The narrow-band resonant stress response evaluation system for thin-walled stator components according to claim 6, characterized in that, Before calculating the root mean square of the strain amplitude in the full frequency domain per unit time based on the strain amplitude function in the full frequency domain, the response analysis module pre-calculates the ratio of the total strain amplitude of each effective frequency band to the vibration response amplitude in the full frequency domain, deletes effective frequency bands with a ratio less than a preset ratio value, and then statistically analyzes the probability density function of the root mean square of the vibration response in each unit time.

Images