Multiple-input multiple-output non-stationary random vibration environment test system and control method

Abstract

The application discloses a kind of multiple-input multiple-output non-stationary random vibration test system and control method, system includes system, including digital control subsystem, digital signal generation and acquisition subsystem and vibration test subsystem.Control method includes setting reference information, generating pseudo reference random signal and reference non-stationary signal, and generating drive signal through frequency domain inverse system, and vibration test subsystem responds non-stationary random vibration signal.A kind of multiple-input multiple-output non-stationary random vibration test system and control method of the application can simultaneously realize good control to power spectral density and kurtosis, accurately simulate real engineering environment, test condition is more close to actual engineering situation, and test result is more practically significant.

Description

Technical Field

[0001] This invention belongs to the field of mechanical vibration environment testing technology, specifically a multi-input multi-output non-stationary random vibration test system and control method. Background Technology

[0002] Environmental testing refers to the use of ground-based tests such as static, vibration, combined environmental, and aerodynamic thermal tests to reproduce the actual working environment, boundaries, and loads of a product, thereby predicting, verifying, and calibrating the mechanical and thermal properties and reliability of the product, and providing a basis for product design verification and improvement optimization.

[0003] In most cases, the actual vibration environment in engineering projects is random. Therefore, random vibration environment testing is essential to verify the reliability, safety, and comfort of products under random excitation. Random vibration environment testing can be divided into single-input single-output (SSO) vibration tests and multiple-input multiple-output (MIMO) vibration tests. MIMO tests can simulate more complex vibration environments. Based on the type of random signal, random vibration environment testing can also be divided into stationary random vibration tests and non-stationary random vibration tests. Currently, the signals reproduced in random vibration tests are mainly stationary Gaussian signals; however, in practical engineering applications, many random vibration environments are non-stationary. Stationary signals differ significantly from the vibration environments in actual engineering projects, thus there is a considerable gap between simulated tests and real-world engineering applications. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a multi-input multi-output non-stationary random vibration test system and control method. The system and method can provide a non-stationary random signal generation method and a non-stationary random signal control strategy used in conjunction with it, so that the random vibration test is closer to the real engineering environment.

[0005] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0006] A multi-input multi-output non-stationary random vibration testing system includes a digital control subsystem, a digital signal generation and acquisition subsystem, and a vibration testing subsystem. The vibration testing subsystem is connected to the digital control subsystem via the digital signal generation and acquisition subsystem. The digital subsystem is implemented by a computer, which includes an algorithm module. The digital signal generation and acquisition subsystem includes a control module, a signal input module, and a signal output module. The control module is connected to the computer, and both the signal input and signal output modules are connected to the control module. The vibration testing subsystem includes an excitation device, a power amplifier, an accelerometer, a fixture, and a test specimen.

[0007] Furthermore, the method for controlling multi-input multi-output non-stationary random vibration signals used in the algorithm module includes the following steps:

[0008] 1) Set reference information, including setting the reference spectrum S rr and reference kurtosis. For S rr Perform Cholesky decomposition:

[0009]

[0010] The superscript H represents the complex conjugate transpose.

[0011] 2) Add a random phase to obtain the spectrum of the desired reference pseudo-random signal x(t):

[0012] X=LΘ(44)

[0013] Where L is the spectral correction matrix, and the original L before correction is L. r Θ is the random phase matrix, as follows:

[0014]

[0015] Where, θ i (i = 1, 2, ..., n) are random numbers in the range [-π, π], and n is the number of control points in the experiment.

[0016] 3) Perform an inverse Fourier transform on the signal spectrum X to obtain the required reference pseudo-random signal x(t), and select a suitable amplitude modulation function A(t) based on the reference kurtosis, where the parameter t represents the time series of the signal.

[0017] 4) Calculate the spectrum of the driving signal for each frame using the frequency domain inverse system method:

[0018] D i =ZX i (46)

[0019] Among them, D i Let X be the spectrum of the driving signal in the i-th frame, Z be the impedance matrix of the experimental system, and X be the signal spectrum in the i-th frame. i For the i-th frame, refer to the pseudo-random signal x i (t) is the signal spectrum obtained by performing a Fourier transform on D. i By performing the inverse Fourier transform, the required driving signal d for each frame can be obtained. i (t).

[0020] 5) By using an improved time-domain randomization method, a reference non-stationary random signal x with time-varying root mean square value and controllable kurtosis is generated. Ref-NS(t), and then through the frequency domain inverse system of step 4), the non-stationary driving signal d(t) required for the experiment can be obtained. A reference stationary random signal x is generated using the classical time-domain randomization method. Ref-S (t).

[0021] 6) The driving signal d(t) is output through the signal output module to drive the vibration control test object of the vibration test subsystem to vibrate.

[0022] 7) Acquire vibration response signal y using an accelerometer and signal input module.

[0023] 8) Extract the non-stationary envelope of the acquired vibration response signal. The non-stationary envelope of the vibration response signal is defined as:

[0024] x e =x Ref-NS -x Refs-S (47)

[0025] The obtained stationary signal is:

[0026] y st =yx e (48)

[0027] The response power spectral density was then calculated:

[0028] S yy =YY H (49)

[0029] Where Y is a stationary signal y st The Fourier transform of , where the superscript H represents the complex conjugate transpose.

[0030] 9) The acquired response signal is subject to dual control of both non-stationary and stationary characteristics. Non-stationary characteristics are controlled by adjusting the amplitude modulation function. Controlling the stationary characteristics of the response signal involves controlling the power spectral density of the non-stationary random signal after envelope extraction; the algorithm for controlling the response signal employs a matrix power control algorithm. The response spectral density S of the stationary signal... yy Perform Cholesky decomposition:

[0031]

[0032] Calculate the correction matrix Δ:

[0033]

[0034] 10) Determine the error between the power spectral density and kurtosis and the target value. If the error is reached, exit control; otherwise, continue correction and proceed to step 11.

[0035] 11) Calculate the new spectral correction matrix:

[0036] L new =Δ η L (52)

[0037] Where η is the convergence power exponent, which controls the convergence speed, and its value ranges from (0,1). The calculated L... new Substitute back to step 2) and calculate the new drive signal.

[0038] Furthermore, the improved time-domain randomization method involves introducing an amplitude modulation function and adjusting its statistical characteristics to control the non-stationary characteristics of the generated signal. Based on the improved time-domain randomization principle, the expression for the generated non-stationary random signal can be obtained:

[0039]

[0040] Among them, A i (t) is the amplitude modulation function, w i (t) is the window function, and l is the superposition factor. The generated non-stationary random signal x Ref-NS The kurtosis formula for (t) is:

[0041]

[0042] Where K x Let x(t) be the kurtosis of the pseudorandom signal, and let the coefficients a, b, c be defined as follows:

[0043]

[0044]

[0045]

[0046] Where T is the time length of the pseudo-random signal x(t). Once the amplitude modulation function, window function, and superposition factor are determined, the coefficients a, b, and c will be constants, thus the pseudo-random signal x(t) and the generated true random signal xt will be... Ref-NS The kurtosis relationship between (t) is linear.

[0047] Let the superposition factor be 2 (l = 2), which is also for the convenience of superposition. The kurtosis of the generated sufficiently long non-stationary random signal can also be further simplified to:

[0048] K NS =αK A K x +β(58)

[0049] In the above formula, K A The kurtosis of the amplitude modulation function is defined by the coefficients α and β as follows:

[0050]

[0051]

[0052]

[0053]

[0054]

[0055] A method for controlling multi-input multi-output non-stationary random vibration signals using the above-mentioned test system includes the following steps:

[0056] 1) Set reference information, including setting the reference spectrum S rr and reference kurtosis. For S rr Perform Cholesky decomposition:

[0057]

[0058] The superscript H represents the complex conjugate transpose.

[0059] 2) Add a random phase to obtain the spectrum of the desired reference pseudo-random signal x(t):

[0060] X=LΘ(65)

[0061] Where L is the spectral correction matrix, and the original L before correction is L. r Θ is the random phase matrix, as follows:

[0062]

[0063] Where, θ i (i = 1, 2, ..., n) are random numbers in the range [-π, π], and n is the number of control points in the experiment.

[0064] 3) Perform an inverse Fourier transform on the signal spectrum X to obtain the required reference pseudo-random signal x(t), and select a suitable amplitude modulation function A(t) based on the reference kurtosis, where the parameter t represents the time series of the signal.

[0065] 4) Calculate the spectrum of the driving signal for each frame using the frequency domain inverse system method:

[0066] D i =ZX i (67)

[0067] Among them, D i Let X be the spectrum of the driving signal in the i-th frame, Z be the impedance matrix of the experimental system, and X be the signal spectrum in the i-th frame. iFor the i-th frame, refer to the pseudo-random signal x i (t) is the signal spectrum obtained by performing a Fourier transform on D. i By performing the inverse Fourier transform, the required driving signal d for each frame can be obtained. i (t).

[0068] 5) By using an improved time-domain randomization method, a reference non-stationary random signal x with time-varying root mean square value and controllable kurtosis is generated. Ref-NS (t), and then through the frequency domain inverse system of step 4), the non-stationary driving signal d(t) required for the experiment can be obtained. A reference stationary random signal x is generated using the classical time-domain randomization method. Ref-S (t).

[0069] 6) The driving signal d(t) is output through the signal output module to drive the vibration control test object of the vibration test subsystem to vibrate.

[0070] 7) Acquire vibration response signal y using an accelerometer and signal input module.

[0071] 8) Extract the non-stationary envelope of the acquired vibration response signal. The non-stationary envelope of the vibration response signal is defined as:

[0072] x e =x Ref-NS -x Refs-S (68)

[0073] The obtained stationary signal is:

[0074] y st =yx e (69)

[0075] The response power spectral density was then calculated:

[0076] S yy =YY H (70)

[0077] Where Y is a stationary signal y st The Fourier transform of , where the superscript H represents the complex conjugate transpose.

[0078] 9) The acquired response signal is subject to dual control of both non-stationary and stationary characteristics. Non-stationary characteristics are controlled by adjusting the amplitude modulation function. Controlling the stationary characteristics of the response signal involves controlling the power spectral density of the non-stationary random signal after envelope extraction; the algorithm for controlling the response signal employs a matrix power control algorithm. The response spectral density S of the stationary signal... yy Perform Cholesky decomposition:

[0079]

[0080] Calculate the correction matrix Δ:

[0081]

[0082] 10) Determine the error between the power spectral density and kurtosis and the target value. If the error is reached, exit control; otherwise, continue correction and proceed to step 11.

[0083] 11) Calculate the new spectral correction matrix:

[0084] L new =Δ η L (73)

[0085] Where η is the convergence power exponent, which controls the convergence speed, and its value ranges from (0,1). The calculated L... new Substitute back to step 2) and calculate the new drive signal.

[0086] Furthermore, the improved time-domain randomization method involves introducing an amplitude modulation function and adjusting its statistical characteristics to control the non-stationary characteristics of the generated signal. Based on the improved time-domain randomization principle, the expression for the generated non-stationary random signal can be obtained:

[0087]

[0088] Among them, A i (t) is the amplitude modulation function, w i (t) is the window function, and l is the superposition factor. The generated non-stationary random signal x Ref-NS The kurtosis formula for (t) is:

[0089]

[0090] Where K x Let x(t) be the kurtosis of the pseudorandom signal, and let the coefficients a, b, c be defined as follows:

[0091]

[0092]

[0093]

[0094] Where T is the time length of the pseudo-random signal x(t). Once the amplitude modulation function, window function, and superposition factor are determined, the coefficients a, b, and c will be constants, thus the pseudo-random signal x(t) and the generated true random signal xt will be... Ref-NS The kurtosis relationship between (t) is linear.

[0095] Let the superposition factor be 2 (l = 2), which is also for the convenience of superposition. The kurtosis of the generated sufficiently long non-stationary random signal can also be further simplified to:

[0096] K NS =αK A K x +β(79)

[0097] In the above formula, K A The kurtosis of the amplitude modulation function is defined by the coefficients α and β as follows:

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] The present invention provides a multi-input multi-output non-stationary random vibration test system and control method, which can simultaneously achieve good control of power spectral density and kurtosis, accurately simulate the real engineering environment, and make the test conditions closer to the actual engineering situation, and the test results more practically significant. Attached Figure Description

[0104] Figure 1 This is a flowchart illustrating the operation of a multi-input multi-output non-stationary random vibration testing system according to the present invention.

[0105] Figure 2 This is a flowchart of a multi-input multi-output non-stationary random vibration test method according to the present invention;

[0106] Figure 3 This is a block diagram of the multi-input multi-output non-stationary random vibration test system in an embodiment of the present invention;

[0107] Figure 4 This is a diagram illustrating the power spectral density control effect in response to a non-stationary random signal in the X control direction in an embodiment of the present invention.

[0108] Figure 5 This is a diagram illustrating the power spectral density control effect in the Y-direction of a non-stationary random signal in an embodiment of the present invention.

[0109] Figure 6 This is a diagram illustrating the kurtosis control effect in response to non-stationary random signals in the X and Y control directions according to an embodiment of the present invention. Detailed Implementation

[0110] The following detailed description, in conjunction with the accompanying drawings, provides a multi-input multi-output non-stationary random vibration test system and test algorithm proposed in this invention. In the description of this invention, it should be understood that terms such as "left side," "right side," "upper part," "lower part," and "bottom," indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation. Terms such as "first" and "second" do not indicate the importance of components and therefore should not be construed as limiting the invention. The specific dimensions used in this embodiment are only for illustrating the technical solution and do not limit the scope of protection of this invention.

[0111] This example provides a test case of two-input two-output nonstationary random vibration control.

[0112] like Figure 1 and 3 As shown, this example uses the Shinken G-6080-3HT-020 triaxial vibration table as the non-stationary random vibration test object. The non-stationary response signal is analyzed and processed by a VXI signal transmission and acquisition system and a computer to achieve non-stationary random vibration control.

[0113] The Shinken G-6080-3HT-020 triaxial vibration table is a small, electrically powered vibration table capable of simultaneously providing translational motion in the X, Y, and Z directions. The table surface dimensions are 0.2 × 0.2 m, the frequency range is 5-1000 Hz, and the maximum displacement is 20 mm. The maximum load capacity is 20 kg, and the maximum acceleration under no-load conditions is 52.9 m / s². 2 The maximum acceleration when fully loaded with 20kg is 22.6m / s². 2 The Agilent VXI signal transmission and acquisition system mainly consists of three parts: the EX2500A, VT1436, and VT1434A boards. The EX2500A connects to the computer via a network cable, with a data transfer speed of up to 40MB / s, enabling the computer to process data quickly. It plays a role in information exchange between the data acquisition module and the computer in the VXI signal transmission and acquisition system.

[0114] The VT1436 is a data acquisition board with a total of 16 acquisition channels. The board itself has some DSP functionality, capable of conditioning sensor signals and performing anti-aliasing filtering. The VT1436 also has a 32MB FIFO memory to buffer continuous signals, thus preventing data loss.

[0115] The VT1434A is a signal source board in VXI with four signal source channels. The VT1434A comes with many built-in signal source modes, including random and sinusoidal modes. The signal source mode can also be set to a custom signal, which is used in random vibration control experiments to send specified drive signals.

[0116] The vibration control test object is the three-axis vibration table surface, and the response control point is in the horizontal X and Y directions on the vibration table surface.

[0117] like Figure 2 As shown, the experiment mainly consists of the following steps:

[0118] 1. Set reference information for the non-stationary random vibration control test. This reference information includes the reference spectrum and reference kurtosis. In this example, the control frequency bandwidth is 20-2000Hz, the alarm limit of the reference spectrum is set to ±3dB, the stopping limit is set to ±6dB, and the amplitude modulation function distribution follows a zero-mean Gaussian distribution.

[0119] 2. Measure the frequency response function matrix of the vibration control test object. The frequency response function matrix of the control test object is obtained by the H1 estimation method in the frequency response function estimation method. Save the measured frequency response function matrix for subsequent generation of the driving signal required in the vibration control test.

[0120] 3. Start the experiment. Add an amplitude modulation function to generate the non-stationary drive signal required for the experiment. In this example, it is then applied to the X and Y directions of the triaxial vibration table surface.

[0121] 4. Acceleration signals in both control directions of the vibration table surface are acquired using accelerometers. The acquired non-stationary response signals are transmitted to the computer control system for analysis and processing, and their power spectral density and kurtosis are controlled simultaneously.

[0122] 5. Determine the error between the power spectrum and kurtosis and the target value. If the test control requirements are not met, continue iterative control; otherwise, exit control.

[0123] The control effect of this example multi-axis non-stationary random vibration control experiment is as follows: Figures 4 to 6 As shown. Figure 4 and Figure 5 The outermost solid line represents the ±6dB stopping limit of the reference spectrum, the next outermost dashed line represents the ±3dB alarm limit, the dotted dashed line in the middle is the reference spectrum line, and the solid line is the response power spectrum line. As can be seen from the figure, the response power spectrum lines in both directions of the vibration table are within the ±3dB alarm limit of the reference spectrum, indicating excellent control performance. Figure 6 It can be seen that the kurtosis of the non-stationary response signals in both directions of the vibration table is completely controlled near the target kurtosis value during the control iteration process, with very small fluctuations.

[0124] Based on the description of preferred embodiments of the present invention, it should be clear that the present invention as defined by the appended claims is not limited to the specific details set forth in the above description, and many obvious modifications to the present invention without departing from its spirit or scope may also achieve the purpose of the present invention.

Claims

1. A multiple-input, multiple-output non-stationary random vibration test system, characterized by, The system includes a digital control subsystem, a digital signal generation and acquisition subsystem, and a vibration testing subsystem. The vibration testing subsystem is connected to the digital control subsystem via the digital signal generation and acquisition subsystem. The digital control subsystem is implemented by a computer, which includes an algorithm module. The digital signal generation and acquisition subsystem includes a control module, a signal input module, and a signal output module. The control module is connected to the computer, and both the signal input module and the signal output module are connected to the control module. The vibration testing subsystem includes an excitation device, a power amplifier, an accelerometer, a fixture, and a test specimen. The algorithm module employs a method for controlling multi-input multi-output non-stationary random vibration signals, which includes the following steps: 1) Set reference information, including setting reference spectrum S rr and reference kurtosis; perform Cholesky decomposition on S rr ​ （1）， The superscript H represents the complex conjugate transpose; 2) Add a random phase to obtain the spectrum of the desired reference pseudo-random signal x(t): （2）， Wherein, L is a spectrum correction matrix, the original L before correction is L r ; Θ is a random phase matrix, which is as follows: （3）， Where, θ i (i=1,2,…,n) are random numbers in the range [-π, π], and n is the number of control points in the experiment; 3) Perform an inverse Fourier transform on the signal spectrum X to obtain the required reference pseudo-random signal x(t), and select a suitable amplitude modulation function A(t) based on the reference kurtosis, where the parameter t represents the time series of the signal; 4) Calculate the spectrum of the driving signal for each frame using the frequency domain inverse system method: （4）， Among them, D i Let X be the spectrum of the driving signal in the i-th frame, Z be the impedance matrix of the experimental system, and X be the signal spectrum in the i-th frame. i For the i-th frame, refer to the pseudo-random signal x i (t) is the signal spectrum obtained by performing a Fourier transform; for D i By performing the inverse Fourier transform, the required driving signal d for each frame can be obtained. i (t); 5) By using an improved time-domain randomization method, a reference non-stationary random signal x with time-varying root mean square value and controllable kurtosis is generated. Ref-NS (t), and then through the frequency domain inverse system of step 4), the non-stationary driving signal d(t) required for the experiment is obtained; through the classical time-domain randomization method, a reference stationary random signal x is generated. Ref-S (t); 6) The driving signal d(t) is output through the signal output module to drive the vibration control test object of the vibration test subsystem to vibrate; 7) Acquire vibration response signal y using an accelerometer and signal input module; 8) Extract the non-stationary envelope of the acquired vibration response signal; the non-stationary envelope of the vibration response signal is defined as: （16）， The obtained stationary signal is: （17）， The response power spectral density was then calculated: （18）， where Y is a stationary signal y st the Fourier transform of X, and the superscript H represents the complex conjugate transpose. 9) The acquired response signal is subject to dual control of both non-stationary and stationary characteristics. The non-stationary characteristic is controlled by adjusting the amplitude modulation function. Controlling the stationary characteristic of the response signal involves controlling the power spectral density of the non-stationary random signal after envelope extraction. The algorithm for controlling the response signal employs a matrix power control algorithm. The response spectral density S of the stationary signal... yy Perform Cholesky decomposition: （19）， Calculate the correction matrix ∆: （20）， 10) Determine the error between the power spectral density and kurtosis and the target value. If the error is reached, exit control; otherwise, continue correction and proceed to step 11). 11) Calculate the new spectral correction matrix: （21）， where η is the convergence power index, which can control the convergence speed, and its value range is (0, 1]; the calculated L new The new driving signal is calculated (step 2).

2. The multiple-input, multiple-output non-stationary random vibration test system of claim 1, wherein, The improved time-domain randomization method involves introducing an amplitude modulation function and adjusting its statistical characteristics to control the non-stationary characteristics of the generated signal. Based on the improved time-domain randomization principle, the expression for the generated non-stationary random signal is obtained: （5）， where A i (t) is the amplitude modulation function, w i (t) is the window function, and l is the superposition factor; the kurtosis formula of the generated non-stationary random signal x Ref-NS (t) is （6）， where K x is the kurtosis of the pseudo-random signal x(t), and the coefficients a, b, c are defined as: （7）， （8）， （9）， Where T is the time length of the pseudo-random signal x(t); once the amplitude modulation function, window function, and superposition factor are determined, the coefficients a, b, and c will be constants, thus the pseudo-random signal x(t) and the generated true random signal xt will be... Ref-NS The kurtosis relationship between (t) is linear; Let the superposition factor be 2 (l=2), which is also for the convenience of superposition; the kurtosis of the generated sufficiently long non-stationary random signal can also be further simplified to: （10）， In the above formula, K A The kurtosis of the amplitude modulation function is defined by the coefficients α and β as follows: （11）， （12）， （13） ， （14） ， （15）。

Images