Fourier transform-based measured load and vibration correlation analysis method

Abstract

The invention belongs to the field of helicopter structure fatigue design, and relates to an actually measured load and vibration correlation analysis method based on Fourier transform. The method comprises the following steps: in different flight states, establishing a corresponding relation between a vibration frequency domain amplitude and a load time domain amplitude by adopting known load time-frequency domain data and vibration time-frequency domain data of a scientific research machine; according to the established corresponding relation, carrying out load inversion on the in-service aircraft in a corresponding flight state; the in-service aircraft and the scientific research aircraft are aircrafts of the same model.

Description

Technical Field

[0001] The invention belongs to the field of helicopter structure fatigue design and relates to a Fourier transform-based method for analyzing the correlation between measured load and vibration. Background Art

[0002] During the development of helicopters, the helicopters are usually tested and modified during the scientific research stage. Strain gauges and other testing equipment are placed at key parts and components of the helicopters to carry out load test flights. Through load test flights, the actual load conditions of the helicopter components under various flight conditions can be obtained, and further strength analysis and life assessment based on the measured loads can be carried out. When the helicopter enters the mass production and delivery stage, load tests are usually no longer carried out. Therefore, it is impossible to obtain the actual load conditions of the helicopter components in the delivery state, and it can only be inferred through the load data of each flight state during the scientific research test flight stage.

[0003] Traditionally, helicopters use load spectra obtained from load test flights to calculate their lifespan. However, in reality, the actual load spectra of each helicopter vary due to flexible usage, and there are certain problems with their economy and even safety.

[0004] Whether in the research stage or in the batch production and delivery stage, vibration sensors are placed in key parts of the helicopter to collect vibration data and monitor the working conditions of various systems of the helicopter. When the vibration data of a certain part is abnormal, timely and accurate investigation can be carried out to minimize or avoid the occurrence of flight accidents.

[0005] There is a certain correlation between helicopter vibration and load. In load test flights, it can be found that when the helicopter performs level flight, turns, etc., when the component load changes significantly, the vibration of the area near the component will also change. Therefore, establishing a relationship model between vibration and load, obtaining the actual load conditions of single-machine components through vibration data, and conducting timely and accurate single-machine use monitoring will greatly improve the use efficiency of helicopters, which is of great significance for improving the economy and safety of single aircraft. Summary of the invention

[0006] Purpose of the invention: To propose a Fourier transform-based method for analyzing the correlation between measured loads and vibrations. The obtained vibration-load relationship model can be used for structural load monitoring, damage analysis, and life calculation, providing data support for scientific research and flight safety assurance of mass-produced helicopters.

[0007] Technical solution:

[0008] A Fourier transform-based method for analyzing the correlation between measured load and vibration is provided, comprising:

[0009] Under different flight conditions, the known load time-frequency domain data and vibration time-frequency domain data of the scientific research aircraft are used to establish the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude;

[0010] According to the established corresponding relationship, the load inversion is carried out on the in-service aircraft under the corresponding flight state; the in-service aircraft and the scientific research aircraft are of the same model.

[0011] Furthermore, the known load time-frequency domain information and vibration time-frequency domain information of the scientific research machine are used to establish the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude, including:

[0012] Through the existing measured load time-frequency domain data, flight parameter data, and vibration time-frequency domain data, after flight state division, Fourier mathematical transform, frequency component analysis, filtering, and load contribution rate analysis, the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude is finally established.

[0013] Further, Fourier mathematical transform, including:

[0014] Divide the load time-frequency domain data and vibration time-frequency domain data in a period of stable flight into several data segments, and the number of points in each data segment is 2 to the kth power; the segment lengths of the load time-frequency domain data and the vibration time-frequency domain data should be consistent;

[0015] Perform Fourier transform on each data point to obtain the load amplitude table and vibration amplitude table at different frequency points.

[0016] Further, frequency component analysis includes:

[0017] Find the main load frequency component points and the corresponding load frequency domain amplitudes from the load amplitude table at different frequency points;

[0018] Find the main vibration frequency component points and the corresponding vibration frequency domain amplitudes from the vibration amplitude table at different frequency points;

[0019] The load frequency domain amplitude of each main load frequency component point and the vibration frequency domain amplitude of each main vibration frequency component point are fitted to obtain the correlation between load and vibration in the frequency domain.

[0020] Further, filtering and load contribution analysis include:

[0021] Filter the load time domain data, filter out the non-main load frequency components, and only retain the component data of each main load frequency component point and its adjacent domain;

[0022] The dynamic loads before and after each filtering are compared to obtain the contribution rate of the main load components to the load.

[0023] The contribution rates of all main load components to the load are averaged to obtain the load inversion coefficient from frequency domain to time domain.

[0024] Furthermore, the correspondence between the vibration frequency domain amplitude and the load time domain amplitude is obtained by dividing the load inversion coefficient on the basis of the correlation between the load and the vibration in the frequency domain.

[0025] Furthermore, before the Fourier mathematical transformation, the method further includes:

[0026] Clear jump points of load time-frequency domain data and vibration time-frequency domain data.

[0027] Furthermore, according to the established corresponding relationship, load inversion is performed on the in-service aircraft under the corresponding flight state, including:

[0028] Obtain the flight status and vibration frequency domain data of in-service aircraft;

[0029] Query the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude according to the flight status;

[0030] The load time domain amplitude is determined through the vibration frequency domain data of the in-service machine and the corresponding relationship.

[0031] Beneficial effects:

[0032] The present invention proposes a Fourier transform-based method for analyzing the correlation between measured load and vibration. The obtained vibration-load relationship model can be used for structural load monitoring, damage analysis, and life calculation, and has great engineering application prospects. For helicopters that cannot carry out load testing, this method predicts component dynamic loads through relatively easy-to-obtain vibration data, and further carries out load monitoring, damage, and life calculation, thereby maximizing the safety of helicopter flight. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic diagram of a channel before the data jump point is cleared.

[0034] Figure 2 This is a schematic diagram of a channel after the data jump point is cleared.

[0035] Figure 3a Frequency domain component analysis of tail rotor shaft load for a certain model Figure 1 .

[0036] Figure 3b Frequency domain component analysis of tail rotor shaft load for a certain model Figure 2 .

[0037] Figure 4a Frequency domain component analysis of X-axis vibration of a certain model of tail reduction Figure 1 .

[0038] Figure 4b Frequency domain component analysis of X-axis vibration of a certain model of tail reduction Figure 2 .

[0039] Figure 5 Schematic diagram of the correlation analysis between load frequency domain and vibration frequency domain.

[0040] Figure 6a This is the load time domain diagram before load filtering.

[0041] Figure 6b This is the load frequency domain diagram before load filtering.

[0042] Figure 6c This is the load time domain diagram after load filtering.

[0043] Figure 6d This is the load frequency domain diagram after load filtering.

[0044] Figure 7a Schematic diagram of the main component contribution rate analysis for filtering the front and tail rotor shaft bending moment.

[0045] Figure 7b Schematic diagram of the main component contribution rate analysis of the tail rotor shaft bending moment after filtering.

[0046] Figure 8a This is a schematic diagram of the heading angle in level flight.

[0047] Figure 8b Schematic diagram of the roll angle in level flight.

[0048] Figure 8c Schematic diagram of the pitch angle in level flight.

[0049] Figure 8d This is a schematic diagram of the speed in level flight.

[0050] Figure 8e Schematic diagram of pressure altitude in level flight.

[0051] Fig. 9 This is a schematic diagram of a channel before the data jump point is cleared.

[0052] Fig.10 This is a schematic diagram of a channel after the data jump point is cleared.

[0053] Fig.11a This is the time domain diagram of the tail rotor shaft load of a certain model.

[0054] Fig.11b This is the frequency domain diagram of the tail rotor shaft load of a certain model.

[0055] Fig.12a This is the time domain diagram of the X-axis vibration of a certain model of tail reduction.

[0056] Figure 12b This is the frequency domain diagram of the X-axis vibration of a certain model of tailgate.

[0057] Fig.13 This is a correlation analysis diagram of the X-axis vibration 2Ω of a certain model of tail reduction and the bending moment load 1Ω of the tail rotor shaft.

[0058] Fig.14a Load time domain diagram before load filtering

[0059] Fig.14b This is the load frequency domain diagram before load filtering.

[0060] Fig.14c This is the load time domain diagram after load filtering.

[0061] Fig.14d This is the load frequency domain diagram after load filtering.

[0062] Fig.15 The figure is a flow chart of a Fourier transform-based method for analyzing the correlation between measured load and vibration. DETAILED DESCRIPTION

[0063] A Fourier transform-based method for analyzing the correlation between measured load and vibration, through the existing measured load time-frequency domain data, flight parameter data, vibration time-frequency domain data, after flight state division, data processing, Fourier mathematical transformation, frequency component analysis, filtering and other steps, finally obtains the relationship between the measured load time-domain amplitude and the vibration frequency-domain amplitude. The specific steps are shown in Figure 1 , briefly described as follows:

[0064] [1] Flight status classification

[0065] According to the flight parameter data of the helicopter, the flight state is identified. Generally, the flight parameter data such as pitch angle, roll angle, speed, altitude, heading angle, etc. are used to identify the flight attitude of the helicopter. In terms of the use of helicopters, most of them are in stable states such as level flight, climbing, hovering, turning, and descending. It is more meaningful to analyze the load phase relationship in the stable state. Therefore, the flight state classification mainly identifies these stable states.

[0066] [2] Data jump point clearing

[0067] Jump points, also called outlier points or abnormal points, refer to some measurement points that are relatively rare in a test data and obviously have a large difference with nearby measurement values ​​and do not meet actual expectations. By removing jump points, the data curve is made smoother and closer to the true value. Figure 1 and Figure 2 .

[0068] [3] Data segmentation

[0069] In order to obtain more data points and more accurate relationship between load and vibration, a longer flight time period is usually divided into several data segments. The number of data points in each segment is usually 2 to the kth power (e.g. each data segment corresponds to 4 seconds in length). In order to obtain the relationship between load and vibration, the segmentation duration of load data and vibration data should be consistent.

[0070] [4] Fourier transform

[0071] Perform Fourier transform on the data points obtained through the jump point clearing in step [2] and the data segmentation in step [3] to obtain the amplitudes of the load and vibration data at different frequency points.

[0072] Since the measured load and vibration frequency domain data are discretized, the analysis accuracy depends on the data sampling frequency and the number of analysis data points involved in Fourier transform. The number of Fourier transform analysis data points can be taken as 2^k based on experience, and the sampling rate of the measured load data of a certain type of machine is f. Therefore, the frequency resolution = f / 2^k, and the frequency points and amplitudes that can be directly obtained are shown in Table 1.

[0073] Table 1 Frequency points and amplitudes that can be directly obtained

[0074]

[0075]

[0076] [5] Frequency component analysis

[0077] The periodic characteristics of helicopter structural vibration and load are very obvious, generally an integer multiple of the rotor speed. Figure 3a and Figure 3b As shown in the figure, the period of the tail rotor shaft load of a certain model is quite obvious. It can be seen from the frequency domain diagram that the main frequency component is the frequency corresponding to 1Ω, which is around 26.8Hz.

[0078] like Figure 4a and Figure 4b As shown in the figure, the X-axis vibration data period of a certain model of tail reduction is also obvious. It can be seen from the frequency domain diagram that the main frequency component of the X-axis vibration of the tail reduction is the frequency corresponding to 2Ω, which is around 53.6Hz.

[0079] [6] Extracting eigenvalues

[0080] Based on the component analysis results of step [5] and the frequency point data that can be accurately obtained by Fourier transform in step [4], the amplitude information of the load and vibration time-frequency domain channels is extracted. Due to the slight fluctuation of the rotor speed, the frequency corresponding to the maximum point of the frequency domain amplitude of each segment of data has a certain deviation. It is usually necessary to set a frequency bandwidth to extract the maximum amplitude within the frequency bandwidth. Table 2 shows the frequency domain amplitude of each data segment corresponding to the bending moment of the tail rotor shaft and the X-direction vibration of the tail reduction in the level flight state obtained by analysis.

[0081] Table 2 Frequency domain amplitude information

[0082]

[0083]

[0084] [7] Correlation analysis between load and vibration frequency domain

[0085] Fit the load frequency domain amplitude and the vibration frequency domain amplitude to obtain the correlation between the load and vibration in the frequency domain. The fitting method can be linear fitting, polynomial fitting, exponential fitting, etc. Taking linear fitting as an example, the following can be obtained: Figure 5 The illustrated correlation between the load frequency domain and the vibration frequency domain, the related equation is y=kx+b, where x is the vibration frequency domain amplitude; y is the load frequency domain amplitude; k is the slope; and b is the constant term.

[0086] [8] Load filtering

[0087] Typically, vibration data is usually represented by frequency domain amplitude, while load data is represented by time domain amplitude. Therefore, establishing the correlation between vibration frequency domain amplitude and load time domain amplitude has more practical application value.

[0088] The load channel data is filtered to filter out the non-main components and only retain the component information near the target frequency. The Fourier transform method can be used for filtering. The method is to set the amplitude of the component near the non-target frequency in step [4] to zero, and then perform an inverse Fourier transform to achieve the purpose of filtering, such as Figure 6a-6d This is a schematic diagram of the time domain and frequency domain before and after load filtering.

[0089] [9] Load contribution analysis

[0090] According to the flight state classification results [1], the dynamic loads before and after the load filtering in the flight state are compared to obtain the contribution rate of the main component to the load. Taking the bending moment of a certain type of tail rotor shaft as an example, the dynamic load before filtering is 393.89Nm, and the dynamic load after filtering is 245.63Nm. Therefore, the contribution rate of the main component to the load is 919.84 / 1071.77=0.86. The specific data is as follows: Figure 7a and Figure 7b shown.

[0091] According to the above method, the load contribution rate of the main component under each flight state can be obtained, and the average value of the load contribution rate can be taken as the load inversion coefficient k' from frequency domain to time domain.

[0092] Table 3 Load contribution analysis

[0093]

[0094]

[10] Load inversion

[0095] Based on the correlation equations between the load frequency domain and the vibration frequency domain obtained in step [7] and the load inversion coefficient k' obtained in step [9], a vibration frequency domain amplitude and load relationship model is established, ultimately realizing the function of predicting the load time domain amplitude from the vibration frequency domain amplitude.

[0096] For this example, the vibration frequency domain amplitude and the load time domain amplitude have the following relationship:

[0097] Where x is the vibration frequency domain amplitude, y is the load frequency domain amplitude, k is the slope, b is the constant term obtained in step [7], and k' is the load inversion coefficient obtained in step [9].

[0098] Through the above method, for helicopters that cannot carry out load testing, the dynamic load of components can be predicted through relatively easy-to-obtain vibration data, and further load monitoring, damage and life calculation can be carried out, thereby maximizing the protection of helicopter flight safety.

[0099] Example

[0100] The following is a detailed description of the invention using the bending moment of a certain type of aircraft tail rotor shaft and the X-direction vibration of the tail rotor as an example. A Fourier transform-based method for analyzing the correlation between measured load and vibration, the steps of which are as follows:

[0101] [1] Flight status classification

[0102] The flight state is identified based on the flight parameter data. For level flight, the speed change curve is mainly used. Combining the altitude, roll angle, pitch angle, and heading angle, it can be concluded that the 716s to 777s period of flight is the level flight state. Figure 8a-8e As shown, the area between the green and red cursors is level flight.

[0103] [2] Jump point clearing process

[0104] Before and after the data jump point of a channel is cleared Fig. 9 and Fig.10 .

[0105] [3] Data segmentation

[0106] The sampling rate of load data and vibration data is 2048. According to the data state division in step [1], the total time of the level flight state is 61s, the total number of data points = 61×2048 = 124928, and k = 13 in the data segmentation, so the number of data points in each segment is 8192. 124928 / 8192 = 15.25 is rounded up to 16, and the load data of the entire climb state can be divided into 16 data segments.

[0107] [4] Fourier transform

[0108] Perform Fourier transform on the data points that have been cleared in step [2] and segmented in step [3] to obtain the amplitude and phase information of the load data at different frequencies. The number of data points for Fourier transform analysis can be taken as 2^13=8192 based on experience. The sampling rate of the measured load data of a certain type of machine is 2048Hz. Therefore, the frequency resolution = 2048 / 8192 = 0.25. The frequency points and amplitudes that can be obtained in the first data segment are shown in Tables 4 and 5.

[0109] Table 4 Frequency points and corresponding amplitudes of the bending moment of the tail rotor shaft of a certain type of aircraft (first paragraph data)

[0110]

[0111]

[0112] Table 5 Frequency points and corresponding amplitudes of X-axis vibration of a certain type of tail (first paragraph data)

[0113]

[0114] [5] Load frequency component analysis

[0115] By analyzing the tail rotor shaft load and tail reduction X-axis vibration frequency domain of a certain model Fig.11a , 11b , 12a, 12b show that the main frequency component of the tail rotor shaft load of a certain model is 1Ω, that is, around 26.8Hz; the main frequency component of the X-axis vibration of a certain model of tail reduction is 2Ω, that is, around 53.6Hz. The frequency tolerance is set to ±1Hz.

[0116] [6] Extracting eigenvalues

[0117] According to the analysis results of step [5], the corresponding amplitude of each data segment for a certain model of tail rotor shaft load 1Ω and the corresponding amplitude of tail reduction 2Ω are extracted. The results are shown in Table 6.

[0118] Table 6 Frequency domain amplitude information

[0119]

[0120] [7] Correlation analysis between load frequency domain and vibration frequency domain

[0121] Fit the load frequency domain amplitude with the vibration frequency domain amplitude to obtain the correlation between the load and vibration in the frequency domain. In this example, linear fitting is used as an example to obtain the following Fig.13 The illustrated correlation between the load frequency domain and the vibration frequency domain, the correlation equation is y=573.14x-121.5.

[0122] [8] Load filtering

[0123] The load channel data is filtered to filter out the non-main components and retain only the component information near the target frequency. The Fourier transform method can be used for filtering. The method is to set the amplitude of the component near the non-target frequency in step [4] to zero, and then perform inverse Fourier transform to complete the load channel filtering. Figures 14a-14d .

[0124] [9] Load contribution analysis

[0125] According to the flight state division results [1], the dynamic loads before and after load filtering in the flight state are compared to obtain the contribution rate of the main components to the load, see Table 7. The average value of the load contribution rate is taken as the load inversion coefficient k'=0.76 from the frequency domain to the time domain.

[0126] Table 7 Contribution rate of tail rotor shaft bending moment 1Ω

[0127]

[0128]

[0129]

[10] Load inversion

[0130] Based on the correlation equations between the load frequency domain and the vibration frequency domain obtained in step [7] and the load inversion coefficient k' obtained in step [9], a vibration frequency domain amplitude and load relationship model is established, and finally the function of predicting the load time domain amplitude from the vibration frequency domain amplitude is realized.

[0131] For this example, the vibration frequency domain amplitude and the load time domain amplitude have the following relationship:

[0132] Through this formula, the tail rotor shaft bending moment can be calculated from the 2Ω amplitude of the tail subtraction x-direction vibration.

Claims

1. A Fourier transform-based method for analyzing the correlation between measured load and vibration, characterized in that: include: Under different flight conditions, the known load time-frequency domain data and vibration time-frequency domain data of the scientific research aircraft are used to establish the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude; According to the established corresponding relationship, the load inversion is carried out on the in-service aircraft under the corresponding flight state; the in-service aircraft and the scientific research aircraft are of the same model.

2. The method according to claim 1, characterized in that Using the known load time-frequency domain information and vibration time-frequency domain information of the scientific research machine, the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude is established, including: Through the existing measured load time-frequency domain data, flight parameter data, and vibration time-frequency domain data, after flight state division, Fourier mathematical transform, frequency component analysis, filtering, and load contribution rate analysis, the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude is finally established.

3. The method according to claim 1, characterized in that Fourier mathematical transforms, including: Divide the load time-frequency domain data and vibration time-frequency domain data in a period of stable flight into several data segments, and the number of points in each data segment is 2 to the kth power; the segment lengths of the load time-frequency domain data and the vibration time-frequency domain data should be consistent; Perform Fourier transform on each data point to obtain the load amplitude table and vibration amplitude table at different frequency points.

4. The method according to claim 3, characterized in that Frequency component analysis, including: Find the main load frequency component points and the corresponding load frequency domain amplitudes from the load amplitude table at different frequency points; Find the main vibration frequency component points and the corresponding vibration frequency domain amplitudes from the vibration amplitude table at different frequency points; The load frequency domain amplitude of each main load frequency component point and the vibration frequency domain amplitude of each main vibration frequency component point are fitted to obtain the correlation between load and vibration in the frequency domain.

5. The method according to claim 4, characterized in that Filtering and load contribution analysis, including: Filter the load time domain data, filter out the non-main load frequency components, and only retain the component data of each main load frequency component point and its adjacent domain; Compare the dynamic loads before and after each filtering to obtain the contribution rate of the main load components to the load; The contribution rates of all main load components to the load are averaged to obtain the load inversion coefficient from frequency domain to time domain.

6. The method according to claim 5, characterized in that The correspondence between the vibration frequency domain amplitude and the load time domain amplitude is obtained by dividing the load inversion coefficient based on the correlation between the load and vibration in the frequency domain.

7. The method according to claim 6, characterized in that Before the Fourier mathematical transformation, the method further comprises: Clear jump points of load time-frequency domain data and vibration time-frequency domain data.

8. The method according to claim 1, characterized in that: According to the established corresponding relationship, load inversion is performed on the in-service aircraft under the corresponding flight state, including: Obtain the flight status and vibration frequency domain data of in-service aircraft; Query the corresponding relationship between the vibration frequency domain amplitude and the load time domain amplitude according to the flight status; The load time domain amplitude is determined through the vibration frequency domain data of the in-service machine and the corresponding relationship.

Images