Main control time domain reproduction and spectrum average random vibration test control method based on actual measurement

Abstract

The invention discloses a main control time domain reproduction and spectrum averaging random vibration test control method based on actual measurement, and belongs to the technical field of vibration control. A system identification transfer function library mode is adopted, a framing technology and a frame overlapping technology are utilized, driving signals are calculated and output in real time, and a PSD spectrum weighted average method of all control points is utilized, so that real-time control is realized. According to the method, a PSD spectrum error and PSD spectrum root-mean-square error data of a current frame are obtained, and a driving frequency spectrum is corrected in a frequency domain by introducing the spectrum error and root-mean-square error data and combining a non-linear correction method, so that a satisfactory effect of overall PSD spectrum control consistency of random vibration is obtained. On one hand, reproduction control of the time history vibration response signal of the main control point can be obtained, on the other hand, consistency of overall PSD spectrum control of the multiple control points can be guaranteed, namely, the phase of the random vibration signal is controlled, and the random vibration test control effect of simulating the actual vibration environment in a better and more vivid mode can be achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of vibration test control, and in particular relates to a main control time domain reproduction and spectrum average random vibration test control method based on actual measurement. Background Art

[0002] Random vibration testing is a major measure used to simulate various random vibration environments that products are subjected to, test product reliability, expose structural material and process defects, and analyze mechanical vibration characteristics. It is also an important means to test and evaluate product functional reliability and solve various mechanical and structural vibration problems. At the same time, random vibration testing is also a necessary means to simulate various vibration working conditions, such as vehicle, locomotive and high-speed rail transportation vibration environment simulation, airborne and shipborne vibration environment simulation, spacecraft and missile weapon flight vibration environment simulation, and earthquake vibration simulation. The actual random vibration environment is usually non-stationary random vibration, but in the past, due to the limitations of the test technology level, traditional random vibration test simulations all used spectral envelopes and stationary random vibration methods to approximate the real vibration environment.

[0003] Traditional random vibration tests usually involve vibration durability tests, structural assessment tests, and environmental adaptability assessment tests that simulate vibration environments. These tests mainly use a vibration test controller + vibration table system to conduct a single test on a single table. Traditional practices use a stationary random vibration test method using PSD envelopes in the frequency domain to approximately simulate the vibration dynamics environment in order to conduct assessment tests and tests on specimens or products.

[0004] With the continuous development of science and technology, people have higher and higher design requirements for various spacecraft systems and large weapon systems. The traditional spectral envelope tightening assessment method approximates all kinds of random vibration environments to stable random vibrations. At the same time, the PSD envelope can only approximately simulate the PSD of the real vibration environment, and cannot realistically simulate the actual vibration environment. The traditional random vibration test method has been increasingly unable to meet the testing and assessment requirements of high-end aircraft, spacecraft systems and weapon systems. At the same time, the spectral envelope tightening method may cause over-assessment problems for the test piece, which will have a certain degree of impact on further optimization design. If the vibration environment simulation can be carried out in a way that accurately reproduces the real vibration response, it will be more scientific, reasonable and effective.

[0005] On the other hand, when conducting overall vibration test for high-end weapons or spacecraft systems, some components or subsystems cannot touch each other, such as detonator components and explosive components. At this time, it is necessary to develop a way to separate the overall level of the host system from the corresponding subsystems in the host system. The usual practice is to replace a certain type of component with a simulated part, and then conduct a decomposition test on a certain type of component separately. However, the test conditions used for the decomposition test of the subsystem at this time cannot truly simulate the actual vibration response. The random vibration test control method will be able to solve this problem well. By obtaining the measured data of the vibration response in real time, and then using the random vibration test control method based on the measured data combined with the time domain waveform reproduction of the main control point and the multi-point spectrum average combination, the test condition problem of the decomposition test of the subsystem can be well solved, and the work efficiency of the test is improved.

[0006] In recent years, some research has been carried out in China on vibration waveform reproduction control technology. For example, the invention patent "A general waveform reproduction control method and control device" (patent authorization announcement number: CN104407547B) proposes a general waveform reproduction control method and control device with vibration waveform as the target wave. This method is only applicable to the reproduction of short-term vibration waveforms with known target waves, such as seismic waves, elastic resistance waves, etc. The invention patents "Continuous vibration signal time history reproduction control method" (patent authorization announcement number: CN105353789B) and "A vibration load rapid reproduction control method" (patent authorization announcement number: CN114923651B) propose a method based on time history target signal processing, system self-checking, initial transfer function estimation, transfer function update, next frame drive signal calculation and other steps, which can only realize the continuous vibration signal reproduction control of a single point time history, but cannot realize multi-point PSD average control. However, these inventions cannot realize frequency domain random vibration test control based on PSD under multiple control points.

[0007] In summary, this patent proposes a random vibration test control method based on the measured master control time domain reproduction and spectral averaging combination. It provides a new control method for vibration test technology, which is different from the traditional random vibration test method and other vibration waveform reproduction test control methods. This method can achieve a relatively accurate reproduction of the vibration response of key position points (master control points) in the time domain, and can also ensure the consistency of the frequency domain PSD control of the overall random vibration. It can better achieve accurate reproduction of the vibration environment, not only for conventional random vibration tests, but also for complex random vibration tests at the overall level + component level. It can achieve the effect of realistically simulating real random vibrations, which has very important engineering significance and can also promote the technological development of my country in the field of vibration testing and vibration control technology.

[0008] After searching patent literature, no other publicly available literature on the random vibration test control method based on actual measurement of master control time domain reproduction and spectral averaging combination was found, and no related products were used in applications. Summary of the invention

[0009] In order to solve the problems raised in the above background technology, the present invention provides a main control time domain reproduction and spectral average random vibration test control method based on actual measurement, so as to solve the problem that the existing method cannot realistically simulate the actual vibration environment, may cause over-testing of the test piece, and cannot realize the frequency domain random vibration test control based on PSD spectrum under multiple control points.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] The master control time domain reproduction and spectrum average random vibration test control method based on actual measurement includes three stages: actual vibration data, pre-test preparation and test execution:

[0012] The first stage is the actual measurement or collection stage of vibration data. The measured vibration data mainly comes from the vibration response data measured in the vibration environment actually experienced by the product. First, according to the vibration environment actually experienced by the product and the actual installation status, the vibration response data measurement points are selected. The measurement point location is selected from the rigid hard connection part of the product or the product installation. The measured data is usually multiple vibration measurement points. One point of interest is selected as the main control point, and the other points are used as secondary control points.

[0013] The measured vibration data and the control response points in the later vibration test form a one-to-one correspondence, that is, the measured data position points (including the main control points and the secondary control points) and the test control response data position points need to be strictly one-to-one corresponding;

[0014] There are two ways to measure the actual vibration signal data. One is to use the data acquisition system to obtain the measured vibration response data for later offline processing. The other is online real-time acquisition, that is, using real-time acquisition + variable reflection memory technology to obtain the actual vibration response data online in real time. Since the random vibration test control method based on the measured main control time domain reproduction and spectrum average combination adopts the frame cycle control method, the actual measured data length is not limited by the length of time.

[0015] The vibration response data obtained can be vibration data that is measured and then processed offline, or it can be real-time vibration data measured online. If the vibration data is measured online in real time, the vibration data will be directly used in the test stage using variable reflection memory technology and frame processing technology.

[0016] The second stage is the test preparation stage, which includes three steps: system parameter setting, reference vibration data processing, and transfer function library construction:

[0017] Step 1: Test parameter setting; the test parameters include the setting of basic parameters, control parameters, correction parameters, channel parameters, transfer function library identification parameters and safety parameters, among which the basic parameters include basic test information and working path; the control parameters include sampling period, single frame points, frame overlap points, number of spectrum lines, upper and lower frequency limits; the correction parameters include frequency domain correction coefficient and root mean square correction coefficient; the channel parameters include control channel selection, main control point selection, acceleration signal sensitivity coefficient and output channel selection; the transfer function library identification and update parameters include identification drive signal length, average number of times and transfer function update weight; the safety parameters include maximum acceleration and maximum drive voltage;

[0018] Step 2: Processing of the reference time-domain vibration data of the master control point; The reference time-domain vibration data of the master control point is divided into two categories. One is offline reference vibration data, that is, the vibration response data measured by the data acquisition system, and then resampled according to the sampling period set by the system through file import. After resampling, the offline reference vibration data file is obtained, and later in the test phase, it enters the controller in the form of frame reading as offline master control point reference time-domain vibration data; the other is called online reference time-domain vibration data, that is, using real-time acquisition + variable reflection memory technology, according to the set sampling period, single frame point number and other parameters, according to the frame acquisition processing method, the online master control point reference time-domain vibration data is obtained in real time;

[0019] Step 3: Constructing the transfer function library; First, by analyzing the measured vibration data or predicting before the test, the maximum target magnitude (maximum target root mean square) and the rough PSD target spectrum of the random vibration test can be preliminarily obtained. Based on this maximum target magnitude and this target spectrum, random pink noise driving signals of different magnitudes consistent with the target spectrum are sent continuously in advance, and the acceleration response signal data of the main control point is collected. According to the transfer function estimation method and theory, the transfer function samples of the system at different test magnitudes are obtained. The specific steps are as follows:

[0020] According to the maximum target level and target spectrum shape as the benchmark, the transfer function is estimated according to the levels of -6dB, -5dB, -4dB, -3dB, -2dB, -1dB, and 0dB, that is, random pink noise driving signals of different levels consistent with the reference spectrum shape are sent in advance, and the root mean square value of the acceleration response signal of the control point is collected and calculated in real time, and then judged and processed. Finally, 7 sets of corresponding driving signal data can be obtained. and response signal data And the RMS value of the response signal , where the root mean square value Ry i is the maximum root mean square R that meets the test target ref The one-to-one correspondence between 0db and -6db is then calculated according to the driving signal data x.i (t) and response signal data y i (t), according to the transfer function identification method and theory, the transfer function is estimated, and the transfer function library data of 7 groups of systems are obtained:

[0021] ;

[0022] Among them, i corresponds to the test level. When i=0, the test level is 0dB, when i=1, the test level is -1dB, when i=2, the test level is -2dB, when i=3, the test level is -3dB, when i=4, the test level is -4dB, when i=5, the test level is -5dB, and when i=6, the test level is -6dB.

[0023] The third stage is the test stage, which mainly includes six steps: dynamic transfer function calculation and transfer function update, receiving the reference time domain vibration data of the master control point, time-varying target PSD and control response PSD calculation, drive signal calculation under frame overlap technology, drive signal frequency domain windowing, frequency domain correction window function and RMS compensation coefficient generation:

[0024] Step 1: Dynamic transfer function calculation and transfer function update: Introduce the nonlinear compensation coefficient group of the transfer function, perform weighted processing on the transfer functions at different magnitudes, and obtain the system dynamic transfer function at the corresponding magnitude. In the vibration test control process, according to the root mean square value R of the reference vibration data of the current frame ref (k) and the maximum target RMS value of the test ref Compare and calculate the corresponding magnitudes, then find a group of transfer functions with the same or closest magnitude in the transfer function library as the maximum weight transfer function, and the weights of other transfer functions with similar magnitudes are second, and so on.

[0025] The dynamic transfer function calculation formula is as follows:

[0026] ;

[0027] Among them, the root mean square of the reference vibration data of the current frame is R ref (k) and the maximum target RMS R ref The relationship is:

[0028] when When i=0, the corresponding test level is 0dB, and the sovereign coefficient γ 0 (1) Take 0.6-0.8, the secondary weight coefficient γ 0 (2) Take , the sub-weight coefficient γ 0 (3) Take , again the weight coefficient γ 0 (4) Take ;

[0029] when When i=1, the corresponding test level is -1dB, the sovereign coefficient γ 1 (1) Take 0.6-0.8, the secondary weight coefficient γ 1 (2) Take , the sub-weight coefficient γ 1 (3) Take , again the weight coefficient γ 1 (4) Take ;

[0030] when When i=2, the corresponding test level is -2dB, the sovereign coefficient γ 2 (1) Take 0.6-0.8, the secondary weight coefficient γ 2 (2) Take , the sub-weight coefficient γ 2 (3) Take , again the weight coefficient γ 2 (4) Take ;

[0031] when When i=3, the corresponding test level is -3dB, the sovereign coefficient γ 3 (1) Take 0.6-0.8, the secondary weight coefficient γ 3 (2) Take , the sub-weight coefficient γ 3 (3) Take , again the weight coefficient γ 3 (4) Take ;

[0032] when When i=4, the corresponding test level is -4dB, the sovereign coefficient γ 4 (1) Take 0.6-0.8, the secondary weight coefficient γ 4 (2) Take , the sub-weight coefficient γ 4 (3) Take , again the weight coefficient γ 4 (4) Take ;

[0033] when When i=5, the corresponding test level is -5dB, the sovereign coefficient γ 5 (1) Take 0.6-0.8, the secondary weight coefficient γ 5 (2) Take , the sub-weight coefficient γ 5 (3) Take , again the weight coefficient γ 5 (4) Take ;

[0034] when When i=6, the corresponding test level is -6dB, the sovereign coefficient γ 6 (1) Take 0.6-0.8, the secondary weight coefficient γ 6 (2) Take , the sub-weight coefficient γ 6 (3) Take , again the weight coefficient γ 6 (4) Take ;

[0035] Considering the time-varying problem of the system, the transfer function online update method is used to improve the impact of the system's resistance to time-varying and enhance the system control stability. In the vibration test control process, according to the previous frame of drive signal data D k-1 (t) and response signal data Y k-1 (t), according to the transfer function identification method and theory, the latest transfer function H of the system can be estimated k-1 (ω);

[0036] ;

[0037] Where λ is the transfer function update coefficient, is the transfer function after online update, H Dik (ω) is the dynamic transfer function.

[0038] Step 2: Receive the reference time domain vibration data of the master control point; if the vibration data is offline reference vibration data, the file frame reading method is used to receive the data; if the vibration data is online reference vibration data, the real-time acquisition + variable reflection memory technology is used to receive the data in the frame acquisition method, and then the received reference vibration data is subjected to FFT window filtering, that is, the reference data is subjected to rectangular window filtering according to the following formula, which is described as follows:

[0039] ;

[0040] Among them, Rec k (ω) is the frequency domain rectangular window function, S k (t) is the original reference time domain vibration data, is the reference time domain vibration data after filtering.

[0041] Step 3: Calculate the time-varying target PSD and the time-varying control response PSD. The time-varying target PSD uses the measured data as the target data for calculation, and the time-varying control response PSD uses the control response data in the test for calculation.

[0042] The calculation formula of time-varying PSD is as follows:

[0043] ;

[0044] Among them, m is the update weight coefficient, , DOF is the statistical degree of freedom, ranging from 50 to 80, PSD k (ω) is the updated PSD, PSD 0 (ω) is the PSD after arithmetic mean of the current frame;

[0045] The above method is used to calculate the time-varying target PSD and control response PSD of all the main control points and secondary control points.

[0046] The time-varying target PSD and the time-varying control response PSD are calculated by weighted average of the time-varying target PSD and the time-varying control response PSD of the primary control point and the secondary control point, where the weight of the primary control point is 0.5 and that of the secondary control point is (1-0.5) / n, where n is the number of secondary control points.

[0047] ;

[0048] Among them, Y 1 (ω) is the time-varying PSD of the main control point, Y 2 (ω)~Y n+1 (ω) is the time-varying PSD of n secondary control points.

[0049] Step 4: Calculation of reference time domain vibration data under frame overlap technology; If the drive signal is directly processed as a single frame, there will be a frame truncation error problem. In order to eliminate the truncation error, the frame overlap processing technology must be introduced to ensure the smooth transition of the drive signal time history frame. The transition section processing method is as follows:

[0050] If it is offline reference vibration data, get the current frame ref(k), the previous frame ref(k-1) and the next frame ref(k+1); if it is online reference vibration data, get the current frame ref(k), the previous frame ref(k-1) and the previous frame ref(k-2), set the number of points per frame to L, the number of points calculated each time to 2L, that is, overlapping transitions, set the number of forward transition points to 0.5L, set the number of backward transition points to 0.5L, and process according to the data length of 2 frames ref k (t).

[0051] Step 5: Calculation of the drive frame signal and double correction of the frequency domain window function and RMS; The calculation method of the drive signal in vibration control is as follows, and the calculation is performed according to the data length of 2 frames:

[0052] ;

[0053] in, is the inverse of the online update transfer function of the system, win k (ω) is the frequency domain correction window function, is the reference time domain vibration data after filtering, D k (t) is the output drive time history signal, △R is the RMS compensation coefficient;

[0054] in, and win k The number of single-frame points of (ω) is L. The direct interpolation method is used to change the number of single-frame points to 2L to match the calculation;

[0055] The driving frame signal D is calculated k After (t), the data of the first 0.5L frame and the second 0.5L frame are removed, and the data of the middle segment with a length of 1 frame is taken as the final driving single-frame time history signal.

[0056] Step 6: Generate frequency domain correction window function and RMS compensation coefficient; calculate the vibration test control spectrum PSDY through random vibration control theory k-1 (ω), at the same time, the frequency domain correction window function win k (ω) is calculated as follows:

[0057] ;

[0058] Among them, PSDR k-1 (ω) is the target spectrum of the k-1th frame, PSDY k-1 (ω) is the control response spectrum of the k-1th frame, The correction factor is between 1 and 2;

[0059] The RMS compensation factor is calculated as follows:

[0060] ;

[0061] Among them, RR k-1 The PSDR of the target spectrum of the k-1th frame k-1 (ω) Calculated frequency domain RMS value, RY k-1 is the frequency domain RMS value calculated from the control response spectrum of the k-1th frame, n is the correction factor between 1 and 2, △R k is the RMS compensation coefficient.

[0062] In the present application, step 2 of the second stage and steps 1, 2, and 4 of the third stage are time domain waveform reproduction methods, step 3 of the third stage is a multi-point spectral averaging test control method, and steps 1 and 3 of the first stage, and steps 5 and 6 of the third stage are methods shared by time domain waveform reproduction and multi-point spectral averaging.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] The test control method in the present application is based on the measured vibration data as the test reference data, and proposes a better and more realistic random vibration test control method for simulating the actual vibration environment. First, the measured vibration data mainly comes from the vibration response data measured in the vibration environment actually experienced by the product. The measuring point position must be installed in the rigid hard connection position. With these vibration measuring point data as reference conditions, the installation state of the product must be kept consistent during the test implementation. The vibration measuring point position corresponds to each control point of the random vibration test of the product one by one, and then one of the points of interest is selected as the main control point, and the other points are used as secondary control points. The dynamic transfer function calculation and update method is used for the main control point, and then the frame division technology and frame overlap technology are used to calculate the output drive signal in real time. At the same time, the time-varying PSD weighted average method of all control points can be used to obtain the time-varying control response PSD of the current frame, and then the time-varying target PSD and the time-varying control response PSD are used to obtain PSD spectrum error and PSD root mean square error data. By introducing the spectrum error and root mean square error data, combined with the nonlinear correction method, the drive spectrum is further corrected in the frequency domain, so as to obtain a satisfactory effect of the overall PSD control consistency of the random vibration.

[0065] On the one hand, this control method can obtain a relatively accurate reproduction control of the time history vibration response signal of the main control point, and on the other hand, it can also ensure the consistency of the overall PSD control of multiple control points, and can better achieve accurate reproduction of the vibration environment. Compared with traditional random vibration control, not only can satisfactory PSD control consistency be obtained, but also the time domain time history vibration response signal of the main control point can achieve a relatively accurate reproduction control effect, that is, the random vibration signal is not only controlled in amplitude but also in phase. This method can achieve a better and more realistic random vibration test control effect that simulates the actual vibration environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a flowchart of the present application. DETAILED DESCRIPTION

[0067] In order to facilitate those skilled in the art to understand the technical content of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific examples. It should be understood that the specific examples described herein are only used to explain the present invention and are not used to limit the present invention.

[0068] like Figure 1 As shown, the workflow of the combined random vibration test control method of the present invention mainly includes three stages: vibration data measurement, pre-test preparation and test execution:

[0069] The first stage is the actual measurement or collection stage of vibration data. The data comes from the vibration response data measured in the vibration environment actually experienced by the product. The measurement point position must be installed in the rigid hard connection position. With these vibration measurement point data as reference conditions, the installation state of the product must be kept consistent during the test implementation, and the vibration measurement point position corresponds to each control point of the random vibration test of the product. Select one of the points of interest as the main control point, and the other points as secondary control points.

[0070] According to the vibration environment and actual installation status of the product, select the vibration response data measurement points. Then you can choose two ways to obtain the data. First, use a general dynamic data acquisition system to directly measure the vibration data and do offline processing; second, use real-time acquisition + variable reflection memory technology to obtain real-time vibration response data, which is directly entered into the test stage using frame processing.

[0071] The second stage is the test preparation stage, which includes three steps: system parameter setting, reference vibration data processing, and transfer function library construction:

[0072] Step 1: Test parameter setting, namely, basic parameters include basic test information, working path, etc.; control parameter settings include sampling period of 0.1ms, single frame points of 6000, frame overlap points of 3000, number of spectrum lines of 800, frequency lower limit of 10Hz, frequency upper limit of 2000Hz, etc.; correction parameters include frequency domain window correction coefficient of 1.5, root mean square correction coefficient of 1.3; channel parameters include control channel selection, main control point selection, acceleration signal sensitivity coefficient, output channel selection, etc.; transfer function library identification and update parameters include identification drive signal length of 5s, average number of times of 100, transfer function update weight of 0.08, etc.; safety parameters include maximum acceleration and maximum drive voltage.

[0073] Step 2: Processing of the reference time-domain vibration data of the master control point. According to the above description, the reference time-domain vibration data of the master control point is divided into two categories. One is called offline reference vibration data, that is, the vibration response data measured by the data acquisition system, and then imported through file reading and writing. According to the sampling period of 0.1ms set by the system, resampling processing is performed to obtain the offline reference time-domain vibration data file after resampling processing; the other is called online reference time-domain vibration data, that is, using real-time acquisition + variable reflection memory technology, according to the set sampling period of 0.1ms, single-frame point number of 6000 points and other parameters, according to the frame acquisition processing method, the online master control point reference time-domain vibration data is obtained in real time.

[0074] Step 3: Transfer function library construction. First, through the analysis of measured vibration data or pre-test prediction, the maximum target level (maximum target root mean square) and PSD target rough spectrum of the random vibration test can be preliminarily obtained. Based on this maximum target level of 20g and the target rough spectrum (trapezoidal spectrum, with slopes of +3dB / oct, +0dB / oct, -6dB / oct, and inflection point coordinates of 10Hz, 100Hz, 1000Hz, 2000Hz) as the benchmark, random pink noise driving signals of different levels consistent with the target spectrum can be sent in advance and continuously, and the acceleration response signal data of the main control point can be collected. According to the transfer function estimation method and theory, the transfer function samples of the system at different test levels can be obtained. The specific approach is as follows:

[0075] Based on the maximum target level and target spectrum, the transfer function is estimated according to the levels of -6dB, -5dB, -4dB, -3dB, -2dB, -1dB, and 0dB. That is, random pink noise driving signals of different levels consistent with the reference spectrum are sent in advance, and the root mean square value of the acceleration response signal of the control point is collected and calculated in real time, and then judged and processed. Finally, 7 sets of corresponding driving signal data can be obtained. and response signal data And the RMS value of the response signal The root mean square value Ry i is the maximum target root mean square R ref Then, according to the driving signal data x i (t) and response signal data y i (t), according to the transfer function identification method and theory, the transfer function is estimated, and the transfer function library data of 7 groups of systems are obtained:

[0076] ;

[0077] Among them, i corresponds to the test level. When i=0, the test level is 0dB, when i=1, the test level is -1dB, when i=2, the test level is -2dB, when i=3, the test level is -3dB, when i=4, the test level is -4dB, when i=5, the test level is -5dB, and when i=6, the test level is -6dB.

[0078] The third stage is the test stage, which mainly includes six steps: dynamic transfer function calculation and transfer function update, receiving single frame reference vibration data, total time-varying target PSD and total time-varying control response PSD calculation, time-varying target PSD and control response PSD calculation, reference vibration data calculation under frame overlap technology, drive frame signal calculation and frequency domain windowing correction, frequency domain correction window function and RMS compensation coefficient generation:

[0079] Step 1: Dynamic transfer function calculation and transfer function update. Considering the system nonlinearity caused by the difference in vibration response magnitude, a transfer function nonlinear compensation coefficient group is introduced to perform weighted processing on the transfer functions at different magnitudes to obtain the system dynamic transfer function at the corresponding magnitude. In the vibration test control process, according to the root mean square value R of the reference vibration data of the current frame, ref (k) and the maximum target RMS value of the test ref Compare and calculate the corresponding magnitude. Then, find a group of transfer functions with the same or closest magnitude in the transfer function library as the maximum weight transfer function, and the weights of other transfer functions with similar magnitudes are second, and so on.

[0080] The dynamic transfer function calculation formula is as follows:

[0081] ;

[0082] At the same time, during the vibration test control process, according to the previous frame of driving signal data D k-1 (t) and response signal data Y k-1 (t), according to the transfer function identification method and theory, the latest transfer function H of the system can be estimated k-1 (ω):

[0083] ;

[0084] Where λ is the transfer function update coefficient, λ=0.08, is the transfer function after online update, H Dik (ω) is the dynamic transfer function.

[0085] Step 2: Receive the reference time-domain vibration data of the master control point. If the vibration data is offline reference vibration data, the file frame reading method is used to receive the data, with 6000 points per frame; if the vibration data is online reference vibration data, real-time acquisition + variable reflection memory technology is used to receive the data in a frame (6000 points per frame) acquisition method. Then perform FFT window filtering on the received reference time-domain vibration data, that is, perform rectangular window filtering on the reference data according to the following formula. Description is as follows:

[0086] ;

[0087] Among them, Rec k (ω) is the frequency domain rectangular window function, S k (t) is the original reference time domain vibration data, is the reference time domain vibration data after filtering.

[0088] Step 3: Calculate the total time-varying target PSD and the total time-varying control response PSD. Since the usual random vibration environment simulation can be approximated as a stationary random process, its PSD calculation method can still be obtained using the traditional periodogram method, but the statistical degree of freedom parameter can generally be selected as a smaller parameter to obtain a suitable time-varying update effect of PSD. The calculation formula for time-varying PSD is as follows:

[0089] ;

[0090] Among them, m is the update weight coefficient, , DOF is the statistical degree of freedom, which is 60, PSD k (ω) is the updated PSD, PSD 0 (ω) is the PSD after arithmetic mean of the current frame.

[0091] The above method is used to calculate the time-varying target PSD and control response PSD of all the main control points and secondary control points.

[0092] In addition, the time-varying target PSD and the time-varying control response PSD will adopt the weighted average method of the time-varying target PSD and the time-varying control response PSD of the main control point and the secondary control point, where the weight of the main control point is 0.5, and the weight of the secondary control point is (1-0.5) / n, where n is the number of secondary control points.

[0093] ;

[0094] Among them, Y 1 (ω) is the time-varying PSD of the main control point, Y 2 (ω)~Y n+1 (ω) is the time-varying PSD of n secondary control points.

[0095] Specifically, the time-varying target PSD and the time-varying control response PSD will adopt the weighted average method of the time-varying target PSD and the control response PSD, where the weight of the main control point is 0.5, the weight of the secondary control point is (1-0.5) / 2, and the number of secondary control points is 2.

[0096] ;

[0097] Among them, Y2(ω) and Y3(ω) are the PSDs of the two secondary control points.

[0098] Step 4: Calculation of reference vibration data under frame overlap technology. The drive signal is directly processed as a single frame, which will result in frame truncation error. In order to eliminate the truncation error, the frame overlap processing technology must be introduced to ensure the smooth transition of the drive signal time history frame. The transition processing method is as follows: If it is offline reference vibration data, the current frame ref (k), the previous frame ref (k-1) and the next frame ref (k+1) can be obtained; if it is online reference vibration data, the current frame ref (k), the previous frame ref (k-1) and the previous frame ref (k-2) can be obtained. Set the number of single frame points to L=6000, and the number of calculation points each time to 2L=12000, that is, overlapping transition before and after, set the forward transition points to 0.5L=3000, set the backward transition points to 0.5L=3000, and process according to the data length of 2 frames (12000 points) ref k (t).

[0099] Step 5: Calculation of drive frame signal and frequency domain windowing correction. The calculation method of the drive signal in vibration control is as follows, calculated according to the data length of 2 frames:

[0100] ;

[0101] in, is the inverse of the system's dynamic transfer function, is the frequency domain correction window function, is the output drive time history signal, is the RMS compensation coefficient;

[0102] in, and win k The number of single-frame points of (ω) is L=6000. It is necessary to use direct interpolation method to change the number of single-frame points to 2L=12000 to match the above calculation.

[0103] The driving frame signal D is calculated k (t), the data of the first 0.5L frame and the second 0.5L frame are removed, and the data of the middle segment of 1 frame length is taken as the final driving single frame time history signal.

[0104] Step 6: Generate frequency domain correction window function and RMS compensation coefficient. The frequency domain correction window function win can be calculated by the total time-varying target PSD and total time-varying control response PSD obtained in step (3). k (ω) is calculated as follows:

[0105] ;

[0106] Among them, PSDR k-1 (ω) is the total time-varying target PSD of the k-1th frame, PSDYk-1 (ω) is the total time-varying control response PSD of the k-1th frame, and φ is the correction factor, which is 1.5.

[0107] The RMS compensation factor is calculated as follows:

[0108] ;

[0109] Among them, RR k-1 is the frequency domain RMS value of the total time-varying target PSD of the k-1th frame, RY k-1 is the frequency domain root mean square value of the total time-varying control response PSD of the k-1th frame, φ is the correction factor and takes 1.3, △R k is the RMS compensation coefficient.

[0110] In this embodiment, the present application adopts a system identification transfer function library method for the main control point, and utilizes the frame division technology and the frame overlapping technology to calculate the output drive signal in real time. At the same time, the PSD spectrum error and the PSD spectrum root mean square error data of the current frame can be obtained by utilizing the PSD spectrum weighted average method of all control points. By introducing the spectrum error and the root mean square error data, combined with the nonlinear correction method, the drive spectrum is further corrected in the frequency domain, so as to obtain a satisfactory effect of the consistency of the overall PSD spectrum control of the random vibration. On the one hand, the present application can obtain a relatively accurate reproduction control of the time history vibration response signal of the main control point, and on the other hand, it can also ensure the consistency of the overall PSD spectrum control of multiple control points. Compared with the traditional random vibration control, not only can a satisfactory PSD spectrum control consistency be obtained, but also the time domain time history vibration response signal of the main control point can achieve a relatively accurate reproduction control effect, that is, the phase of the random vibration signal is also controlled. The present application can achieve a better and more realistic random vibration test control effect that simulates the actual vibration environment.

Claims

1. A master control time domain reproduction and spectrum average random vibration test control method based on actual measurement, characterized in that: The method comprises the following steps: setting a plurality of vibration response data measurement points on the product, selecting one of the measurement points as the main control point and the other measurement points as the secondary control points based on the measured vibration time domain data, and then adopting the dynamic transfer function calculation and update method for the main control point, and utilizing the frame segmentation technology and the frame overlap technology to calculate the output drive signal in real time; at the same time, all the target PSDs and control response PSDs of the main control point and the secondary control point are obtained by adopting the time-varying PSD calculation method to obtain the time-varying target PSD and the time-varying control response PSD of the current frame; then utilizing the time-varying target PSD and the time-varying control response PSD to obtain the PSD spectrum error and the PSD root mean square error data; and by introducing the PSD spectrum error and the PSD root mean square error data and combining the nonlinear correction method, the drive spectrum is corrected in the frequency domain.

2. The master control time domain reproduction and spectrum average random vibration test control method based on actual measurement according to claim 1 is characterized in that: The measured vibration time domain data is derived from the vibration response data measured in the vibration environment actually experienced by the product. The measured vibration measurement point is installed at the rigid hard connection part of the product. The vibration measurement point data is used as a reference condition. During the test implementation, the installation status of the product remains consistent. The vibration measurement point position of the measured vibration data corresponds one by one to each control measurement point of the product random vibration test. Any one of the measurement points is selected as the main control point, and the other points are selected as secondary control points.

3. The master control time domain reproduction and spectrum average random vibration test control method based on actual measurement according to claim 2 is characterized in that: The measured vibration response data is collected and then processed offline, or collected online in real time, and the length of the measured vibration response data is not limited by time.

4. The master control time domain reproduction and spectrum average random vibration test control method based on actual measurement according to claim 3 is characterized in that: For the measured vibration time domain data and control response vibration time domain data, the PSD method is calculated by frame processing and periodogram, DOF is selected as the statistical degree of freedom, and DOF is 50-80. The time-varying target PSD and control response PSD of all points of the main control point and the secondary control point can be obtained, and then the total time-varying target PSD and the total time-varying control response PSD are obtained by the weighted average method, wherein the weighting coefficient of the main control point is set to 0.5, and the weighting coefficient of the secondary control point is (1-0.5) / n, n is the number of secondary control points, and the total time-varying target PSD and the total time-varying control response PSD are used as the calculation input of the frequency domain correction window function.

5. The master control time domain reproduction and spectrum average random vibration test control method based on actual measurement according to claim 4 is characterized in that: In the process of driving signal calculation, the frequency domain window function and root mean square double correction calculation method are introduced.

Images