Multi-vibrator vibration test control method and device, terminal and medium

Abstract

The application provides a multi-excitation vibrator vibration test control method, device, terminal and medium, the method comprises the following steps: obtaining a multi-channel white noise signal and a multi-channel output signal, and obtaining an adaptive inverse matrix according to the multi-channel white noise signal and the multi-channel output signal; the driving signal of the current frame is applied to the test piece through the excitation vibrator, and the test piece response signal is obtained; the dynamic target phase is obtained according to the test piece response signal; the driving signal of the next frame is obtained according to the dynamic target phase, the test piece response signal, the driving signal of the current frame and the adaptive inverse matrix, and the test piece is controlled according to the driving signal of the next frame through the excitation vibrator, and the test piece response signal of the next frame is obtained. The inter-channel antagonistic effect is eliminated, the error is efficiently eliminated, the number of iterations required to reach the target spectrum tolerance band is shortened, and the control convergence speed is improved.

Description

Technical Field

[0001] This invention relates to the field of vibration test control technology, specifically to a multi-exciter vibration test control method, device, terminal, and medium. Background Technology

[0002] In reliability testing of aerospace, automotive, and electronic products, multiple-exciter (MIMO) vibration testing is widely used to simulate the vibration environment of real large and complex structures. Traditional MIMO stochastic vibration control is usually based on inverse control using the frequency response function matrix (FRF). The basic process is as follows: using the pilot measurement system's FRF matrix, calculate its inverse matrix; during control, calculate the error between the target spectrum and the response spectrum, use the inverse matrix to solve for the correction amount of the driving signal, and iterate to approximate the target response.

[0003] Traditional control strategies often attempt to control both the amplitude (power spectral density) and phase of the response simultaneously. However, in random vibrations, due to the nonlinearity of the system or the uncertainty of boundary conditions, the phase relationship between each control point is often difficult to reproduce accurately. Forcing phase control can lead to "internal friction" or "antagonism" between exciters, resulting in wasted driving energy, and amplitude control convergence is extremely slow. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a multi-exciter vibration test control method, device, terminal, and medium, aiming to eliminate the antagonistic effect between channels, efficiently eliminate errors, shorten the number of iterations required to reach the target spectral tolerance band, and improve the control convergence speed.

[0005] In a first aspect, embodiments of this application provide a multi-exciter vibration test control method, including: Acquire multi-channel white noise signal and multi-channel output signal, and obtain adaptive inverse matrix based on the multi-channel white noise signal and the multi-channel output signal; The driving signal of the current frame is applied to the specimen through the exciter to obtain the specimen response signal; Based on the specimen response signal, the target vector is reconstructed using the measured response phase to obtain the dynamic target phase; Based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix, the ratio of increase or decrease of the driving signal in each iteration is limited to obtain the driving signal of the next frame. The specimen is then controlled by the exciter according to the driving signal of the next frame to obtain the specimen response signal of the next frame.

[0006] Optionally, obtaining the adaptive inverse matrix based on the multi-channel white noise signal and the multi-channel output signal includes: Based on the multi-channel white noise signal and the multi-channel output signal, the frequency response function matrix is ​​obtained; For each frequency point f, the condition number and adaptive regularization parameter of the frequency response function matrix are dynamically calculated based on the ill-conditionedness of the frequency response function matrix at the current frequency. An improved Tikhonov cost function is constructed, and an adaptive inverse matrix is ​​obtained based on the frequency response function matrix and the adaptive regularization parameter.

[0007] Optionally, the step of reconstructing the target vector using the measured response phase based on the specimen response signal to obtain the dynamic target phase includes: The current measured phase of the specimen response signal is unwrapped to obtain the predicted phase value of the current frame; Using the current measured phase as the phase observation value, and updating it by Kalman filtering by predicting the current frame phase value, a smooth phase estimate is obtained; The target vector is reconstructed based on the smoothed phase estimate to obtain the dynamic target phase.

[0008] Optionally, the step of obtaining the driving signal for the next frame by limiting the increase or decrease ratio of the driving signal in each iteration based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix includes: Based on the dynamic target phase and the specimen response signal, the frequency domain error signal is obtained; The driving signal for the next frame is obtained based on the frequency domain error signal, the driving signal of the current frame, and the adaptive inverse matrix.

[0009] Optionally, it also includes: The driving signal of the historical frame is obtained, and the convergence state factor is obtained based on the specimen response signal, the dynamic target phase, and the driving signal of the historical frame. Based on the preset base gain limit, the preset gain margin, the condition number of the frequency response function matrix, the convergence state factor, the drive signal of the current frame, and the drive signal of the next frame, the dynamic gain limit, the drive ratio, and the safe drive are obtained. If the drive ratio corresponding to the frequency point is greater than the upper limit of the dynamic gain, then based on the dynamic target phase, the specimen response signal, the drive signal, and the adaptive inverse matrix, the drive signal for the next frame is obtained, including: Force the drive signal of the next frame corresponding to the frequency point to be restricted to safe drive.

[0010] Optionally, obtaining the dynamic gain upper limit, drive ratio, and safe drive based on a preset base gain upper limit, a preset gain margin, the convergence state factor, the drive signal of the current frame, and the drive signal of the next frame includes: The dynamic gain limit is obtained based on the preset base gain limit, the preset gain margin, and the convergence state factor. The drive ratio is obtained based on the drive signal of the current frame and the drive signal of the next frame; The safety drive is obtained based on the drive signal corresponding to the frequency point and the upper limit of the dynamic gain.

[0011] Optionally, it also includes: Based on the driving signal of the current frame and the response signal of the specimen, a multi-feature monitoring vector is obtained; The driving signal and the specimen response signal of the historical frame are obtained, and the applied driving increment and the generated response increment are obtained based on the driving signal of the current frame, the specimen response signal, the driving signal of the historical frame and the specimen response signal of the historical frame. In each frame control loop, the adaptive inverse matrix is ​​modified using the applied driving increment, the generated response increment, the convergence state factor, and the adaptive regularization parameter, and the quasi-Newton method is employed to obtain the modified adaptive inverse matrix. When each parameter in the multi-feature monitoring vector satisfies the corresponding preset condition, the adaptive inverse matrix is ​​updated by correcting the adaptive inverse matrix.

[0012] Secondly, embodiments of this application provide a multi-exciter vibration test control device, comprising: The data acquisition module is used to acquire multi-channel white noise signals and multi-channel output signals, and to obtain an adaptive inverse matrix based on the multi-channel white noise signals and the multi-channel output signals; The signal driving module is used to apply the driving signal of the current frame to the specimen through the exciter to obtain the specimen response signal; The phase determination module is used to reconstruct the target vector based on the measured response phase of the specimen response signal to obtain the dynamic target phase. The test control module is used to limit the increase or decrease ratio of the driving signal in each iteration according to the dynamic target phase, the specimen response signal, the driving signal of the current frame and the adaptive inverse matrix, to obtain the driving signal of the next frame, and to control the specimen through the exciter according to the driving signal of the next frame to obtain the specimen response signal of the next frame.

[0013] Thirdly, embodiments of this application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the multi-exciter vibration test control method as described in any one of the first aspects above.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the multi-exciter vibration test control method as described in any one of the first aspects above.

[0015] Fifthly, embodiments of this application provide a computer program product that, when run on a terminal device, causes the terminal device to execute the multi-exciter vibration test control method described in any one of the first aspects.

[0016] This application acquires multi-channel white noise signals and multi-channel output signals, and obtains an adaptive inverse matrix based on the multi-channel white noise signals and multi-channel output signals; applies the driving signal of the current frame to the specimen through the exciter to obtain the specimen response signal; reconstructs the target vector using the measured response phase based on the specimen response signal to obtain the dynamic target phase; limits the increase or decrease ratio of the driving signal in each iteration based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix to obtain the driving signal of the next frame; and controls the specimen through the exciter based on the driving signal of the next frame to obtain the specimen response signal of the next frame. This eliminates the inter-channel antagonistic effect, efficiently eliminates errors, shortens the number of iterations required to reach the target spectral tolerance band, and improves the control convergence speed. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 This is a flowchart illustrating the first embodiment of the multi-exciter vibration test control method provided in this application. Figure 2 This is an overall flowchart of a multi-exciter vibration test control method provided in an embodiment of this application; Figure 3 This is the state-aware model dynamic hierarchical amplitude limiting logic in the multi-exciter vibration test control method provided in one embodiment of this application.

[0019] Figure 4 (a) is a schematic diagram comparing the theoretical frequency response function of the main channel and the observed frequency response function after distortion after the introduction of disturbance in the multi-exciter vibration test control method provided in an embodiment of this application.

[0020] Figure 4(b) is a schematic diagram comparing the theoretical frequency response function of the coupling channel with the observed frequency response function that is distorted after the introduction of disturbance in a multi-exciter vibration test control method provided in an embodiment of this application.

[0021] Figure 5 (a) is a schematic diagram of the PSD spectrum control result of channel 1 when the FRF disturbance is 0.4 in the multi-exciter vibration test control method provided in an embodiment of this application.

[0022] Figure 5 (b) is a schematic diagram of the PSD spectrum control result of channel 2 when the FRF disturbance is 0.4 in the multi-exciter vibration test control method provided in an embodiment of this application.

[0023] Figure 6 (a) is a schematic diagram of the PSD spectrum control result of channel 1 when the FRF disturbance is 0.4 in the multi-exciter vibration test control method provided in an embodiment of this application.

[0024] Figure 6 (b) is a schematic diagram of the PSD spectrum control result of channel 2 when the FRF disturbance is 0.4 in the multi-exciter vibration test control method provided in an embodiment of this application.

[0025] Figure 6 (c) is a schematic diagram of the time-domain waveform signal actually output by channel 1 during the control process in a multi-exciter vibration test control method provided in an embodiment of this application; Figure 6 (d) is a schematic diagram of the time-domain waveform signal actually output by channel 2 during the control process in a multi-exciter vibration test control method provided in an embodiment of this application.

[0026] Figure 7 This is a block diagram illustrating the system hardware architecture of a multi-exciter vibration test control method provided in an embodiment of this application. Figure 8 This is a schematic diagram of the structure of the multi-exciter vibration test control device provided in the embodiments of this application; Figure 9 This is a schematic diagram of the structure of the terminal device provided in the embodiments of this application. Detailed Implementation

[0027] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0028] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0029] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0030] Figure 1 The diagram shown is a flowchart of a first embodiment of the multi-exciter vibration test control method provided in this application. It is provided as an example and not as a limitation. Figure 1 As shown, the method may include: S10, acquire the multi-channel white noise signal and the multi-channel output signal, and obtain the adaptive inverse matrix based on the multi-channel white noise signal and the multi-channel output signal; To eliminate inter-channel antagonism and efficiently eliminate errors, the number of iterations required to reach the target spectral tolerance band is reduced, thus improving the control convergence speed, such as... Figure 2 As shown, multi-channel white noise signals can be input to multiple exciters respectively, and multi-channel output signals can be obtained through multiple exciters; then, an adaptive inverse matrix is ​​obtained based on the multi-channel white noise signals and the multi-channel output signals. As one implementation method, an adaptive inverse matrix is ​​obtained based on the multi-channel white noise signal and the multi-channel output signal, specifically including: A1. Based on the multi-channel white noise signal and the multi-channel output signal, the frequency response function matrix is ​​obtained; The frequency response function matrix will be established using multi-channel white noise signal excitation and the acquired multi-channel output signal. .

[0031] A2, for each frequency point f, dynamically calculate the condition number and adaptive regularization parameter of the frequency response function matrix based on the ill-conditionedness of the frequency response function matrix at the current frequency; For each frequency point Based on the system frequency response function matrix at the current frequency The degree of ill-conditioning is dynamically calculated using the frequency response function matrix, condition number, and adaptive regularization parameters. Adaptive regularization parameters The specific calculation formula is as follows: in, The frequency response function matrix At frequency point The condition number at a given point (i.e., the condition number of the frequency response function matrix); and These are the preset upper and lower limits of the regularization parameters, which are usually determined based on experience with the system noise floor and control accuracy; The frequency response function matrix at frequency points The maximum singular value at (in the algorithm, by adjusting the maximum singular value at) (obtained by singular value decomposition (SVD) calculation). The frequency response function matrix at frequency points The minimum singular value at (in the algorithm, by adjusting the minimum singular value at the minimum singular value) (obtained by singular value decomposition (SVD) calculation). and These are pre-set upper and lower limits of system conditions used to determine the degree of pathology.

[0032] A3. Construct the improved Tikhonov cost function, and obtain the adaptive inverse matrix based on the frequency response function matrix and the adaptive regularization parameter. .

[0033] Further, by introducing the response coherence confidence matrix and the structural dynamic stiffness constraint matrix, an improved Tikhonov cost function is constructed. The final adaptive inverse matrix calculation formula, derived from the frequency response function matrix and adaptive regularization parameters, is as follows: in, It is an adaptive inverse matrix; This is a confidence weighting matrix based on the frequency response coherence function, whose diagonal elements are the multicoherence function values ​​(ranging from 0 to 1) of each channel, used to reduce the weight of the coherent signal-to-noise ratio channels; To drive the energy penalty matrix, an identity matrix is ​​typically chosen. ; The generalized inverse of the FRF obtained from the initial low-order random excitation is used to obtain the pre-identified structural dynamic stiffness constraint matrix. This is the prior weighting coefficient for physical stiffness, typically ranging from [0.01, 0.1], calibrated through pre-tests; The frequency response function matrix The conjugate transpose of . This is a weighted matrix based on the energy distribution of the diagonal elements, where the diagonal elements are... ,in, Indicates at frequency point Next, the The excitation channel to the first Frequency response function of each response channel; For the first The maximum amplitude transmissibility of the response channel across all excitation channels. The physical implication of this design is that for "ill-conditioned / weakly excited response channels" with very low transmissibility in all excitation directions, their weight in the weighting matrix can be amplified by taking their reciprocal, thus achieving targeted suppression of ill-conditioned channels. A weighted matrix based on the energy distribution of the diagonal elements of the matrix. The conjugate transpose of; This is the frequency response function matrix.

[0034] S20, the driving signal of the current frame The exciter is applied to the specimen to obtain the specimen's response signal. ; like Figure 7 As shown, after obtaining the adaptive inverse matrix Then, the driving signal of the current frame will be... (That is, the driving signal of the kth frame) is applied to the specimen through the exciter to obtain the specimen response signal. Among them, the specimen response signal This can be the signal after Fourier transform. Among them, the driving signal of the current frame... It can be the signal after Fourier transform.

[0035] S30, Based on the specimen response signal, the target vector is reconstructed using the measured response phase to obtain the dynamic target phase; like Figure 2 As shown, in one implementation method, the target vector is reconstructed using the measured response phase based on the specimen response signal to obtain the dynamic target phase. Specifically, it includes: B1, unwrap the current measured phase of the test specimen response signal to obtain the predicted current frame phase value; That is, by introducing structural dynamics equations and unwrapping the current measured phase of the specimen's response signal, the predicted phase value of the current frame is obtained. ; Introducing structural dynamics equations (in For the quality matrix, Here is the damping matrix. (where the stiffness matrix is ​​used). The above dynamic parameters can be extracted through finite element modal analysis (FEA) before the experiment, or obtained through parameter identification by low-level modal sweep frequency test. The excitation force vector (in actual experiments, this excitation force vector can be obtained in real time by force sensor / impedance head installed at the connection between the vibrator and the specimen, or it can be calculated based on the current output drive signal and the electromechanical conversion coefficient of the vibrator system). Given the angular frequency, calculate the theoretical prior value of the physical phase in the frequency domain. By integrating the state of the previous frame with prior physical information, the response signal of the specimen is determined. Current measured phase Perform unwrapping and establish phase state equations. ; and according to the phase state equation The predicted phase value of the current frame is obtained. ;in, The state transition matrix, in stationary random vibrations, has a gradual phase change between adjacent frames and is usually set as the identity matrix. It is used to describe the phase variation between two adjacent frames. For the first The posterior phase estimate of the frame, i.e., the optimal estimate combining the measurements, is the initial posterior phase estimate. Set as the measured phase of frame 0 when the experiment starts; For rational phase prior values; This is the physical model confidence factor, which has a value range of [0,1] and can be dynamically adjusted according to the model accuracy.

[0036] B2, based on the current measured phase The phase observations are used, and the Kalman filter is applied to update the phase values ​​of the predicted current frame to obtain a smoothed phase estimate. Among them, the specimen response signal The current measured phase observation value is the frequency domain complex sequence obtained by performing a Fast Fourier Transform (FFT) on the time domain data. The calculation formula is: , In response signal The imaginary part; In response signal The real part.

[0037] That is, combining the current measured phase observation values Update the phase estimate: The smoothed phase estimate is obtained. .in, Kalman gain is used to balance the reliability of predicted and measured values. The measured phase value of the k-th frame is the value derived from the specimen response signal. The phase is obtained directly from the calculation.

[0038] B3, based on the smoothed phase estimate Reconstruct the target vector to obtain the dynamic target phase. .

[0039] That is, using the filtered smooth phase Reconstruct the target vector: To obtain the dynamic target phase .in, Let f be the target amplitude at frequency point f. j is the phase factor; j is the phase coefficient. This is the smoothed phase estimate (posterior phase estimate) at frequency point f in the k-th frame.

[0040] S40, based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix, the ratio of increase / decrease of the driving signal in each iteration is limited to obtain the driving signal for the next frame. ; and according to the drive signal of the next frame The specimen is controlled by the exciter to obtain the specimen response signal of the next frame. ; In other words, based on the error between the current response and the target, the driving signal for the next frame is calculated, propelling the system closer to the target.

[0041] The next frame drive signal obtained The D / A converter performs a D / A conversion and outputs the signal to a power amplifier, forming a closed-loop control.

[0042] After receiving the drive signal for the next frame and adaptive inverse matrix Then, according to the drive signal of the next frame The specimen is controlled by the exciter to obtain the specimen response signal of the next frame. .

[0043] As one implementation method, based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix, the driving signal of the next frame is obtained by limiting the proportion of increase or decrease of the driving signal in each iteration, including: C1, based on the dynamic target phase and the test specimen response signal The frequency domain error signal is obtained. ; Wherein, according to the dynamic target phase and the test specimen response signal The frequency domain error signal is obtained. Specifically, it includes: ; in, The dynamic target phase at frequency point f; The specimen response signal at frequency point f; This is the frequency domain error signal at frequency point f.

[0044] C2, based on the frequency domain error signal drive signal and adaptive inverse matrix To obtain the drive signal for the next frame. .

[0045] Wherein, according to the frequency domain error signal drive signal and adaptive inverse matrix To obtain the drive signal for the next frame. Specifically, it includes: ; in, This is the driving signal for the current frame, which is also the driving signal for the kth frame. This is the step size factor, used to control the update intensity and prevent overshoot; It is an adaptive inverse matrix; Let be the frequency domain error signal of the k-th frame; This is the driving signal for the next frame, that is, the driving signal for the (k+1)th frame.

[0046] As a necessary component of closed-loop control, such as Figure 7 As shown, after the next frame drive signal is output as excitation through D / A conversion, the system will acquire the corresponding next frame specimen response signal. This allows the system to proceed to the next control iteration.

[0047] Compared to traditional fixed-phase control strategies, this application significantly improves the control convergence speed. Traditional MIMO control strategies typically force the response signal to simultaneously approximate the target in both amplitude and phase. This is prone to "internal friction" and antagonism between exciters due to system nonlinearity or uncertain boundary conditions, thus slowing down the convergence process. This invention employs a phase-adaptive strategy, abandoning the forced phase locking, allowing the control system to focus on the accurate reproduction of energy (amplitude). Compared to traditional methods, this invention eliminates the antagonistic effect between channels, enabling more efficient error elimination and shortening the number of iterations required to reach the target spectral tolerance band.

[0048] As one implementation method, this solution aims to intelligently upgrade the "amplitude limiting protection." Traditional fixed amplitude limiting cannot simultaneously achieve both "rapid initial convergence" and "fine-grained stability in the later stages." Therefore, this solution introduces a "funnel-shaped" adaptive amplitude limiting logic, such as... Figure 2 As shown, the multi-exciter vibration test control method may further include: D1, acquire the drive signals of historical frames, and based on The test specimen response signal Dynamic target phase And the driving signals of historical frames, to obtain the convergence state factor; As one implementation method, the driving signal of historical frames is acquired, and based on... The test specimen response signal Dynamic target phase And the driving signals of historical frames, to obtain the convergence state factor, specifically including: E1, based on the specimen response signal and dynamic target phase The root mean square value of the error is obtained. and historical error sequences; According to the test specimen response signal and dynamic target phase Calculate the root mean square value of the relative error within the current frame frequency range. Specifically, it includes: ; in, is the root mean square error value; N is the number of frequency points; The specimen response signal at frequency point f; Frequency point Dynamic target phase at the location Let f be the dynamic target amplitude at frequency point f.

[0049] Among them, the historical error sequence is the current frame and the previous frame calculated by the preceding steps. The root mean square error values ​​of the frames arranged chronologically, i.e. .

[0050] E2 utilizes a lightweight LSTM time series model, taking the historical error sequence as input, to predict the error change rate for the next frame. The historical error sequence is input into a pre-trained LSTM network, and the hidden layer extracts temporal features, while the fully connected layer outputs the single-step relative rate of change. The LSTM model was trained offline using error sequences from historical typical vibration tests before the experiment. During actual control, the model parameters remained fixed to ensure the stability of the online closed-loop system. Its input sequence length is typically the root mean square error of approximately L frames, and the output is the predicted single-step relative rate of change.

[0051] E3 acquires the drive signals of historical frames and, based on these signals, calculates the cumulative amount of experimental drive energy. ; Calculate the cumulative amount of test-driven energy. The calculation starts from the first frame (t=1) of the test and accumulates the sum of squares of the frequency domain amplitude of the driving signal frame by frame, reflecting the total energy injected into the specimen from the start of the test to the current frame; For drive signal The frequency domain amplitude.

[0052] E4, based on the root mean square error value, the error change rate of the next frame, and the cumulative amount of test driving energy, divide each by its corresponding maximum reference value to obtain the dimensionless normalized parameter that eliminates the dimension difference, and then perform weighting to obtain the composite index. The formula for calculating the composite index is as follows: ; in, It is a composite index that characterizes the overall approximation risk of the current control system; This is the first normalized weighting coefficient; This is the second normalized weighting coefficient; This is the third normalized weighting coefficient. , , These are dimensionless normalization parameters obtained by dividing the corresponding variable by its reference maximum value, in order to eliminate the influence of dimensional differences.

[0053] E5, based on the preset composite index threshold and the composite index The convergence state factor is obtained. ; in, A preset dimensionless error threshold is used to distinguish between the "rapid approximation stage" and the "fine locking stage." It is set by the test personnel before the test based on the control accuracy requirements and system characteristics.

[0054] like Figure 3 As shown, based on the preset composite index threshold and the composite index The convergence state factor is obtained. Specifically, it includes: ; in, The compliance index is the one calculated above; For the preset compliance threshold, when When the value is less than this threshold, the system determines that it has entered a fine control state; The sensitivity coefficient for state switching determines the steepness of the transition from wide-limiting to strict-limiting. Before the test begins, the target power spectral density (PSD) given by the test standard or specification is loaded into this unit and converted into the target amplitude in the frequency domain. The initial regularization parameter range and error convergence threshold are also set. , When the error is large (Wide limiting mode); when the error is small, (Strict amplitude limiting mode).

[0055] D2, based on the preset base gain upper limit Preset gain margin The condition number of the frequency response function matrix The convergence state factor The driving signal of the current frame and the drive signal for the next frame To obtain the upper limit of dynamic gain Drive ratio and security drivers ; As one implementation method, the dynamic gain limit, drive ratio, and safe drive are obtained based on a preset base gain limit, a preset gain margin, the convergence state factor, the drive signal of the current frame, and the drive signal of the next frame, including: F1, based on the preset base gain upper limit Preset gain margin and convergence state factor To obtain the upper limit of dynamic gain ; As one implementation method, such as Figure 3 As shown, based on the preset base gain upper limit Preset gain margin The condition number of the frequency response function matrix and convergence state factor To obtain the upper limit of dynamic gain Specifically, it includes: ; in, This is the upper limit of the dynamic gain. The preset upper limit of the base gain is the maximum driving rate of change allowed during the steady-state control phase. The value range is usually 1.05 to 1.2, which means that the maximum increase in a single iteration is allowed to be 5% to 20%. This is the preset gain margin. The physical meaning of the preset gain margin is the additional gain relaxation allowed during the rapid approximation stage in the early stage of the experiment. The value range is usually 0.5~2.0. Smaller values ​​are taken for strongly nonlinear systems and larger values ​​are taken for systems with good linearity. This is the convergence state factor. This is the pathological penalty coefficient. The condition number of the frequency response function matrix calculated in the preceding steps, i.e. When the matrix at a certain frequency point is extremely ill-conditioned ( When the frequency band is very high, the system automatically tightens the upper limit of the drive in that frequency band to ensure the absolute safety of the exciter at the anti-resonance point or the point of severe coupling.

[0056] Dynamic gain limit Level I (Fast Approximation Region): Allows for large gain jumps to rapidly increase the magnitude.

[0057] Dynamic gain limit Level II (Fine Locking Region): Strictly limits the rate of change of a single spectral line to ensure control precision.

[0058] F2, based on the driving signal of the current frame. and the drive signal for the next frame To obtain the driving ratio ; The formula for calculating the drive ratio is as follows: ;in, The driving signal for the current frame at frequency point f. ; This is the driving signal for the next frame at frequency point f.

[0059] F3, based on the driving signal corresponding to the frequency point. and the aforementioned dynamic gain upper limit Get security driver .

[0060] Calculate the drive ratio between the estimated drive signal (the drive signal of the next frame at frequency point f) and the drive signal of the previous frame (the drive signal of the next frame at frequency point f). ,like Then it will be forcibly truncated to Generate the final security driver ;in, This is the driving signal for the current frame at frequency point f.

[0061] D3, if the drive ratio corresponding to the frequency point Greater than the upper limit of the dynamic gain Then, based on the dynamic target phase... Specimen response signal drive signal and adaptive inverse matrix To obtain the drive signal for the next frame. This includes: forcibly changing the drive signal of the next frame corresponding to the frequency point. Restricted to security drivers .

[0062] Through security driver This invention overcomes the transient impact defects of traditional "hard limiting" mechanisms, providing a smoother safety guarantee than single voltage limiting. Existing safety protection technologies typically employ "hard limiting" (i.e., limiting only the absolute value of the driving voltage). When the update amount calculated by the algorithm is too large, although no overvoltage occurs, it can cause severe transient physical shocks, posing a risk of damaging precision test pieces. This invention proposes a nonlinear limiting mechanism based on the "rate of change." Compared to traditional hard limiting, this scheme not only limits the absolute value but also limits the percentage increase or decrease of the driving signal in each iteration. This "soft protection" mechanism eliminates the possibility of sudden changes in the driving signal from the algorithm's underlying layer, providing smoother and more reliable safety protection for equipment and test pieces when system characteristics change drastically or are disturbed.

[0063] As one implementation method, to enhance the online correction function of the model, and to address the model mismatch problem caused by specimen characteristic drift (such as screw loosening or fatigue) during long-term testing, a "control-and-correct" approach is adopted; for example... Figure 2 As shown, the multi-exciter vibration test control method may further include: G1, based on the driving signal of the current frame The response signal of the specimen This yields a multi-feature monitoring vector; In addition to the conventional PSD error and time-domain RMS error, "coherent function drop monitoring" and "drive-response nonlinearity" are added as safety feature indicators.

[0064] As one implementation, the multi-feature monitoring vector includes: power spectral density error, time-domain data error, coherence function drop monitoring, and drive response nonlinearity. The step of obtaining the multi-feature monitoring vector based on the drive signal of the current frame and the specimen response signal includes: H1, based on the specimen response signal The power spectral density error was obtained. ; ; This is a test PSD file; The target power spectral density; where, The specimen response signal at frequency point f; The frequency period; This represents the power spectral density error.

[0065] H2, based on the specimen response signal The time-domain data error was obtained. ,in ,in, The number of temporal sampling points within a frame. The aforementioned test specimen response signal The corresponding original time-domain sampled data sequence, This is the time-domain target reference signal.

[0066] H3, based on the driving signal of the current frame The response signal of the specimen The coherence function is obtained. ; using signals Drive and specimen response signals Calculate the coherence function to determine whether the coherence coefficient has experienced a precipitous drop of a predetermined percentage compared to the previous period (i.e., calculate the coherence function). (fall monitoring). The cross-power spectrum of the driving force and the response. (for (Complex conjugate of the driving signal); self-power spectrum of the driving signal. The self-power spectrum of the response signal .

[0067] H4, according to the coherence function and the test specimen response signal self-power spectrum The nonlinearity of the driving response is obtained. .

[0068] Specifically, it includes: the coherence function calculated in step H3. Specimen response signal self-power spectrum Extract the response signal of the specimen. The proportion of incoherent energy that cannot be linearly explained by the driving signal is obtained. A band-wide integrated nonlinearity index is defined. (That is, the nonlinearity of the driving response), its calculation formula is: in, For the first The nonlinearity index of the frame's driving response; the closer its value is to 1, the more severe the nonlinear response of the system due to structural gaps, friction, or large deformation. When the preset safety threshold is exceeded, the system will trigger a protection mechanism; It is a coherence function; For the test specimen response signal self-power spectrum ; This refers to a frequency point.

[0069] G2 acquires the driving signal and specimen response signal of the historical frame, and obtains the applied driving increment based on the driving signal of the current frame, the specimen response signal, the driving signal of the historical frame, and the specimen response signal of the historical frame. and generate response increment ; Among them, apply driving increment This is the increment vector of the driving signal between the current frame and the previous frame; it generates the response increment. This is the increment vector of the response signal between the current frame and the previous frame.

[0070] G3, in each frame control loop, utilizes the applied drive increment and the generated response increment The convergence state factor and the adaptive regularization parameter The quasi-Newton method is used to adapt the inverse matrix. After making corrections, we obtain the corrected adaptive inverse matrix. ; As one implementation, in each frame control loop, the applied drive increment is utilized. The generation of response increments The convergence state factor and the adaptive regularization parameter The quasi-Newton method is used to adapt the inverse matrix. After making corrections, we obtain the corrected adaptive inverse matrix. Specifically, it includes: ; in, Step size learning rate; ; in, The inverse system matrix before correction; This is the corrected inverse system matrix; The learning rate, which is based on the base step size, is usually set as an empirical constant according to the initial noise level of the system. To generate response increment The conjugate transpose of; It is a very small positive number, used to prevent the denominator from being zero and to ensure the stability of numerical calculations; The convergence state factor; The confidence weighting matrix for the observation increment is a diagonal matrix, with the diagonal elements corresponding to the multicoherence function values ​​of each response channel at the current frequency point. Used to freeze or slow down model updates when the error has not converged; Used to reduce the correction weight when the frequency is severely ill-conditioned; The regularization penalty coefficient is calibrated through pre-tests; For adaptive regularization parameters.

[0071] By correcting the adaptive inverse matrix This enables the control system to track changes in the physical properties of the measured object in real time, forming a Model Reference Adaptive (MRAC) closed loop.

[0072] G4 updates the adaptive inverse matrix by correcting the adaptive inverse matrix when each parameter in the multi-feature monitoring vector satisfies the corresponding preset condition, that is, by using the corrected adaptive inverse matrix calculated in the current frame. Replace the adaptive inverse matrix Used for control calculations in the next frame.

[0073] The preset conditions are: power spectral density error, time-domain data error, and driving response nonlinearity are all less than a preset safety threshold, and the coherence function does not exhibit a preset percentage drop. The formula for determining whether the coherence function exhibits a preset percentage drop is: ;in This is a preset drop ratio threshold, such as 20%.

[0074] like Figure 4 , Figure 5 , Figure 6 and Figure 7 As shown, this application, by modifying the adaptive inverse matrix, exhibits stronger robustness under model mismatch and large error environments compared to traditional inverse control algorithms. Traditional frequency domain inverse control algorithms are highly sensitive to the accuracy of the frequency response function matrix (FRF). Engineering experience shows that when the model error exceeds 20%, traditional algorithms are prone to control divergence. This invention combines frequency adaptive sparse regularization and model online correction (MRAC) techniques, significantly improving tolerance to model errors. Simulation and experimental data show that even with a stochastic multiplicative disturbance of up to 40% (i.e., 0.4) introduced into the FRF model, this invention can still maintain stable convergence of the system. This means that when the pilot measurement environment is not ideal or the specimen characteristics drift significantly during the experiment (such as screw loosening or fatigue), this invention has superior stability and engineering adaptability compared to traditional control algorithms.

[0075] The specific verification results are as follows: Figure 4 , Figure 5 , Figure 6 As shown.

[0076] in, Figure 4This diagram illustrates the frequency response function (FRF) results when a random multiplicative perturbation of 0.4 (40%) is introduced into the FRF. This diagram is used to illustrate the harsh control environment of the system under severe model mismatch. The diagram shows the main channel (e.g., Figure 4 (a) shown) and coupling channel ( Figure 4 (b) Frequency response characteristics. The figure includes curves in two states: the curve exhibiting smooth characteristics represents the system's true theoretical frequency response function curve (H_theoretical); while the curve exhibiting violent fluctuations and spikes represents the observed frequency response function curve after artificially introducing a 40% multiplicative perturbation (H_perturbation). Figure 4 It can be seen that the frequency response function after introducing the perturbation has a significant distortion at each resonance and anti-resonance peak.

[0077] Figure 5 This diagram illustrates the results of PSD (Power Spectral Density) closed-loop control of the system under the aforementioned FRF disturbance environment of 0.4. The purpose of this diagram is to verify the high-precision control capability of the present invention under model mismatch conditions. Figure 5 It includes channel 1 (e.g.) Figure 5 (a) shown) and channel 2 (as shown) Figure 5 (b) shows the frequency domain control results. To clearly distinguish the multiple curves in the figure, the following explanations are provided: The target curve (target) represents the standard target power spectral density required by the test, which is the central reference that the control system needs to approximate; the control curve (control) represents the actual output power spectral density of the specimen response after the system has undergone iterations of the multi-exciter control algorithm. In the figure, this curve closely follows and fits the fluctuation of the target curve; the alarm curve (alarm) is distributed on the outer side of the inner layer of the target curve, divided into upper and lower limits, representing the control error warning boundary allowed by the specification (i.e., the alarm tolerance band); the stop curve (stop) is distributed on the outermost side of the overall figure (i.e., the periphery of the alarm curve), also divided into upper and lower limits, representing the limit error safety boundary allowed by the specification (i.e., the stop tolerance band). Figure 5 It can be seen that, despite the perturbation of up to 40% in the FRF model, the "control curve" generated by the method of this application can still closely fit the "target curve", and the entire frequency band is completely within the tolerance band set by the "alarm curve" without touching the outermost "stop curve", which verifies the excellent performance of the algorithm in eliminating inter-channel antagonistic effects and improving convergence stability.

[0078] Figure 6 This is a schematic diagram showing the result of time-domain playback control of the system under an FRF disturbance of 0.4. Figure 6 (a) and Figure 6 (b) and Figure 5 (a) and Figure 5 (b) The same frequency domain monitoring and control results, Figure 6 (c) and Figure 6(d) The time-domain waveform signals actually output by Channel 1 and Channel 2 during the control process are shown in the supplementary illustration. Figure 6 The time-domain waveform diagram shows that, under the control of the method proposed in this application, even if the system characteristics are drastically disturbed, the physical time-domain sequence output by each channel remains smooth and stable, without high-frequency oscillations or severe transient physical shocks caused by algorithm divergence, thus ensuring the absolute safety of the precision specimen during the vibration test.

[0079] This application acquires multi-channel white noise signals and multi-channel output signals, and obtains an adaptive inverse matrix based on the multi-channel white noise signals and multi-channel output signals; applies the driving signal of the current frame to the specimen through the exciter to obtain the specimen response signal; reconstructs the target vector using the measured response phase based on the specimen response signal to obtain the dynamic target phase; limits the increase or decrease ratio of the driving signal in each iteration based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix to obtain the driving signal of the next frame; and controls the specimen through the exciter based on the driving signal of the next frame to obtain the specimen response signal of the next frame. This eliminates the inter-channel antagonistic effect, efficiently eliminates errors, shortens the number of iterations required to reach the target spectral tolerance band, and improves the control convergence speed.

[0080] For those consistent with the above, please refer to Figure 8 , Figure 8 This application provides a schematic diagram of the structure of a multi-exciter vibration test control device. (See attached diagram.) Figure 8 As shown, the device includes: The data acquisition module 801 is used to acquire multi-channel white noise signals and multi-channel output signals, and to obtain an adaptive inverse matrix based on the multi-channel white noise signals and the multi-channel output signals; The signal driving module 802 is used to apply the driving signal of the current frame to the specimen through the exciter to obtain the specimen response signal; The phase determination module 803 is used to reconstruct the target vector based on the measured response phase of the specimen response signal to obtain the dynamic target phase. The test control module 804 is used to limit the increase or decrease ratio of the driving signal in each iteration according to the dynamic target phase, the specimen response signal, the driving signal of the current frame and the adaptive inverse matrix, to obtain the driving signal of the next frame, and to control the specimen through the exciter according to the driving signal of the next frame to obtain the specimen response signal of the next frame.

[0081] like Figure 9As shown, this application embodiment also provides a terminal device 2, which includes: at least one processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the at least one processor. The processor 20 and the memory 21 are connected. When the processor 20 executes the computer program 22, it implements the steps in the robot motion planning method embodiment.

[0082] This application also provides a computer storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the multi-exciter vibration test control methods described in the above method embodiments.

[0083] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps of any of the multi-exciter vibration test control methods described in the above method embodiments.

[0084] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium can include at least: any entity or device capable of carrying computer program code to a device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable storage media cannot be electrical carrier signals or telecommunication signals.

[0085] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0086] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0087] In the embodiments provided in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0088] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

Claims

1. A method for controlling vibration testing with multiple exciters, characterized in that, include: Acquire multi-channel white noise signal and multi-channel output signal, and obtain adaptive inverse matrix based on the multi-channel white noise signal and the multi-channel output signal; The driving signal of the current frame is applied to the specimen through the exciter to obtain the specimen response signal; Based on the specimen response signal, the target vector is reconstructed using the measured response phase to obtain the dynamic target phase; Based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix, the ratio of increase or decrease of the driving signal in each iteration is limited to obtain the driving signal of the next frame. The specimen is then controlled by the exciter according to the driving signal of the next frame to obtain the specimen response signal of the next frame. The step of obtaining the adaptive inverse matrix based on the multi-channel white noise signal and the multi-channel output signal includes: Based on the multi-channel white noise signal and the multi-channel output signal, the frequency response function matrix is ​​obtained; For each frequency point f, the condition number and adaptive regularization parameter of the frequency response function matrix are dynamically calculated based on the ill-conditionedness of the frequency response function matrix at the current frequency. An improved Tikhonov cost function is constructed, and an adaptive inverse matrix is ​​obtained based on the frequency response function matrix and the adaptive regularization parameter; The step of reconstructing the target vector using the measured response phase based on the specimen response signal to obtain the dynamic target phase includes: The current measured phase of the specimen response signal is unwrapped to obtain the predicted phase value of the current frame; Using the current measured phase as the phase observation value, and updating it by Kalman filtering by predicting the current frame phase value, a smooth phase estimate is obtained; The target vector is reconstructed based on the smoothed phase estimate to obtain the dynamic target phase.

2. The multi-exciter vibration test control method according to claim 1, characterized in that, The step of obtaining the driving signal for the next frame by limiting the increase or decrease ratio of the driving signal in each iteration based on the dynamic target phase, the specimen response signal, the driving signal of the current frame, and the adaptive inverse matrix includes: Based on the dynamic target phase and the specimen response signal, the frequency domain error signal is obtained; The driving signal for the next frame is obtained based on the frequency domain error signal, the driving signal of the current frame, and the adaptive inverse matrix.

3. The multi-exciter vibration test control method according to claim 1, characterized in that, Also includes: The driving signal of the historical frame is obtained, and the convergence state factor is obtained based on the specimen response signal, the dynamic target phase, and the driving signal of the historical frame. Based on the preset base gain limit, the preset gain margin, the condition number of the frequency response function matrix, the convergence state factor, the drive signal of the current frame, and the drive signal of the next frame, the dynamic gain limit, the drive ratio, and the safe drive are obtained. If the drive ratio corresponding to the frequency point is greater than the upper limit of the dynamic gain, then based on the dynamic target phase, the specimen response signal, the drive signal, and the adaptive inverse matrix, the drive signal for the next frame is obtained, including: Force the drive signal of the next frame corresponding to the frequency point to be restricted to safe drive.

4. The multi-exciter vibration test control method according to claim 3, characterized in that, The process of obtaining the dynamic gain limit, drive ratio, and safe drive based on the preset base gain limit, preset gain margin, convergence state factor, drive signal of the current frame, and drive signal of the next frame includes: The dynamic gain limit is obtained based on the preset base gain limit, the preset gain margin, and the convergence state factor. The drive ratio is obtained based on the drive signal of the current frame and the drive signal of the next frame; The safety drive is obtained based on the drive signal corresponding to the frequency point and the upper limit of the dynamic gain.

5. The multi-exciter vibration test control method according to claim 3 or 4, characterized in that, Also includes: Based on the driving signal of the current frame and the response signal of the specimen, a multi-feature monitoring vector is obtained; The driving signal and the specimen response signal of the historical frame are obtained, and the applied driving increment and the generated response increment are obtained based on the driving signal of the current frame, the specimen response signal, the driving signal of the historical frame and the specimen response signal of the historical frame. In each frame control loop, the adaptive inverse matrix is ​​modified using the applied driving increment, the generated response increment, the convergence state factor, and the adaptive regularization parameter, and the quasi-Newton method is employed to obtain the modified adaptive inverse matrix. When each parameter in the multi-feature monitoring vector satisfies the corresponding preset condition, the adaptive inverse matrix is ​​updated by correcting the adaptive inverse matrix.

6. A multi-exciter vibration test control device, characterized in that, include: The data acquisition module is used to acquire multi-channel white noise signals and multi-channel output signals, and obtain a frequency response function matrix based on the multi-channel white noise signals and multi-channel output signals; for each frequency point f, dynamically calculate the condition number and adaptive regularization parameter of the frequency response function matrix based on the ill-conditioned degree of the frequency response function matrix at the current frequency; construct an improved Tikhonov cost function, and obtain an adaptive inverse matrix based on the frequency response function matrix and the adaptive regularization parameter; The signal driving module is used to apply the driving signal of the current frame to the specimen through the exciter to obtain the specimen response signal; The phase determination module is used to: unwrap the current measured phase of the specimen response signal to obtain the predicted current frame phase value; use the current measured phase as the phase observation value and update it by Kalman filtering through the predicted current frame phase value to obtain a smoothed phase estimate; and reconstruct the target vector based on the smoothed phase estimate to obtain the dynamic target phase. The test control module is used to limit the increase or decrease ratio of the driving signal in each iteration according to the dynamic target phase, the specimen response signal, the driving signal of the current frame and the adaptive inverse matrix, to obtain the driving signal of the next frame, and to control the specimen through the exciter according to the driving signal of the next frame to obtain the specimen response signal of the next frame.

7. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the multi-exciter vibration test control method as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the multi-exciter vibration test control method as described in any one of claims 1 to 5.

Images