Real-time load spectrum tracking control method and system

Abstract

The present application relates to the technical field of load spectrum tracking control, and discloses a real-time load spectrum tracking control method and system, which comprises the following steps: firstly, determining master-slave channels, collecting relevant curves, constructing channel dynamic fingerprints and storing; then, decomposing a target load spectrum, constructing a time sequence correlation label, generating and optimizing a collaborative frequency weight matrix; finally, inputting the optimized matrix into a hydraulic servo controller, and adopting master-slave synchronization and cross-coupling control strategies to realize multi-channel spectrum tracking output; the present application can significantly improve the time synchronization accuracy and phase consistency of load spectrum of each channel in multi-channel dynamic compression-shear test, so that the test results can more truly reflect the dynamic response of a complex structure or system, and break through the traditional multi-channel control precision bottleneck.

Description

Technical Field

[0001] This invention relates to the field of load spectrum tracking and control technology, and more specifically, to a real-time load spectrum tracking and control method and system. Background Technology

[0002] Multi-channel dynamic compression-shear testing is crucial for simulating the stress conditions of components under complex working conditions. It is widely used in scenarios such as seismic performance evaluation of large and complex structures, performance testing of multi-degree-of-freedom robot joints, and environmental simulation testing of aerospace structural components. In these tests, ensuring high synchronization of the load spectra of each channel in both time and phase is key to obtaining accurate test results. However, due to differences in the characteristics of the hydraulic systems of each channel, variations in frictional resistance, and the independence of the control algorithms, achieving high-precision time-synchronized load spectrum tracking control has become an extremely challenging task.

[0003] Chinese patent CN109405873B, authorized by the Ministry of Industry and Information Technology, provides a method for rapid and high-precision loading control of dynamic load spectra. It constructs B-spline basis functions, obtains a basis function control library through iterative learning, and performs piecewise rolling fitting on the repeatedly processed load spectrum to control an AC asynchronous variable frequency motor to generate loading torque. However, this existing technology mainly targets torque loading control of AC asynchronous variable frequency motors and has limitations in multi-channel dynamic compression-shear test scenarios. Multi-channel dynamic compression-shear tests involve multiple hydraulic servo channels working collaboratively, with complex and diverse dynamic characteristics for each channel. This method does not fully consider the impact of factors such as the hysteresis response of hydraulic valves and the potential distortion of the servo controller on the synchronization of the load spectra of each channel. In simulating bridge seismic performance tests, the differences in the response of hydraulic valves to control signals of different frequencies and the distortion of the controller output voltage may cause the magnitude and phase of the loading force in each channel to fail to accurately track the target load spectrum, making it difficult to meet the high-precision requirements for time synchronization and phase consistency under complex multi-channel conditions.

[0004] Chinese patent application CN118466196A discloses a modeling and control method for a force-controlled servo loading system used in a hydraulic loading platform. It designs the mechanical structure and loading method of the hydraulic loading system and builds a force-controlled servo loading system model to achieve control. While this method improves control accuracy to some extent, it fails to effectively address the timing deviations and phase differences between the load spectra of different channels in multi-channel dynamic compression-shear tests when handling multi-channel collaborative control. In the dynamic performance testing of multi-degree-of-freedom robot joints, when multiple channels operate simultaneously, the differences in the hydraulic system characteristics of each channel lead to varying processing capabilities for load signals of different frequencies. This method lacks refined analysis of these differences and targeted control strategies, making it impossible to accurately allocate weights and synchronize control based on the actual dynamic characteristics of each channel. It is therefore difficult to ensure that each channel simulates the complex force conditions during joint movement with high consistency in time and phase.

[0005] Existing technologies have limitations in addressing the challenges posed by differences in hydraulic system characteristics, frictional resistance, and the independence of control algorithms in multi-channel dynamic compression-shear tests. They cannot accurately handle the synchronization of load spectra across channels, making it difficult for test results to accurately reflect the actual stress state of complex structures or systems. Summary of the Invention

[0006] This invention has applications spanning multiple fields, including seismic performance testing of large and complex structures (such as bridges and building models), dynamic performance testing of multi-degree-of-freedom robot joints, and aerospace structural component testing simulating complex environmental loads. To overcome the aforementioned deficiencies of existing technologies, this invention provides a real-time load spectrum tracking and control method and system. This involves determining master and slave channels, collecting relevant data to construct and store channel dynamic fingerprints, decomposing the target load spectrum, constructing time-series correlation labels, generating and optimizing a collaborative frequency weighting matrix, and finally utilizing master-slave synchronization and cross-coupling control strategies to achieve multi-channel spectrum tracking output. This significantly improves the time synchronization accuracy and phase consistency of the load spectra of each channel in multi-channel dynamic compression-shear tests, breaking through the traditional multi-channel control accuracy bottleneck and providing more reliable and accurate results for related tests.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] Real-time load spectrum tracking and control methods include:

[0009] The master and slave channels of the hydraulic servo system in the dynamic compression-shear testing machine are determined, and the hysteresis response curves of the hydraulic valves in each master and slave channel and the potential distortion spectrum of the servo controller are collected. Based on the hysteresis response curves of the hydraulic valves in each master and slave channel, the channel dynamic fingerprints of each master and slave channel are constructed, and the channel dynamic fingerprints and potential distortion spectra are stored in the heterogeneous parameter library.

[0010] The target load spectrum is decomposed into three frequency band components, and time-series association labels for each frequency band component are constructed. The channel dynamic fingerprints of the master and slave channels are fused and combined with the three frequency band components to generate an initial cooperative frequency weight matrix. Based on the potential distortion spectrum, channel dynamic fingerprints and the constructed time-series association labels for each frequency band component stored in the heterogeneous parameter library, the initial cooperative frequency weight matrix is ​​corrected and weighted to obtain an optimized cooperative frequency weight matrix.

[0011] The optimized collaborative frequency weight matrix is ​​input into the hydraulic servo controller, and based on the master-slave synchronization and cross-coupling control strategy, the spectrum tracking output of multiple hydraulic servo channels is controlled in real time.

[0012] Furthermore, the method for determining the main channel and slave channels of the hydraulic servo system in the dynamic compression-shear testing machine is as follows: designate one hydraulic servo channel in the dynamic compression-shear testing machine as the main channel, and the remaining hydraulic servo channels as slave channels, wherein the number of slave channels is n1, and n1≥1.

[0013] Furthermore, the construction of the channel dynamic fingerprint of each master and slave channel based on the hysteresis response curves of the hydraulic valves in each channel includes:

[0014] Based on the hysteresis response curves of the master and slave hydraulic valves, the Hammerstein model is used to describe the nonlinear dynamic characteristics of the hydraulic valves. The model parameters are fitted by the least squares method to obtain the nonlinear dynamic models of the master and slave hydraulic valves. For the nonlinear dynamic models of the master and slave hydraulic valves, the time-domain phase offset margin and frequency-domain gain attenuation coefficient are extracted to construct the channel dynamic fingerprints of the master and slave channels.

[0015] Furthermore, the three frequency band components are high-frequency pulse components, mid-frequency harmonic components, and low-frequency base components;

[0016] The time-series association labels for each frequency band component include:

[0017] For the decomposed high-frequency pulse components, extract the timestamp sequence of the pulse peaks and construct high-frequency time series association tags;

[0018] For intermediate frequency harmonic components, phase reference points of each harmonic component are extracted, and intermediate frequency timing association tags are constructed.

[0019] For low-frequency base components, an absolute timestamp is directly added as a low-frequency timing association label;

[0020] The high-frequency timing association tags of the high-frequency pulse component, the intermediate-frequency timing association tags of the intermediate-frequency harmonic component, and the low-frequency timing association tags of the low-frequency base component are used as the timing association tags of each frequency band component.

[0021] Furthermore, the generation of the initial cooperative frequency weight matrix includes:

[0022] Based on the channel dynamic fingerprints of each master and slave channel, for high-frequency pulse components, according to the time-domain phase offset margin of each channel, the initial weight ratio of the high-frequency band of each master and slave channel is determined by fuzzy inference method.

[0023] For the intermediate frequency harmonic components, the initial weight ratio of the intermediate frequency band of each master and slave channel is determined by the frequency domain gain attenuation coefficient of each channel.

[0024] For the low-frequency base component, a balanced allocation strategy is adopted to distribute the weight of the low-frequency band equally among the master and slave channels.

[0025] By combining the weight allocation results of high-frequency, mid-frequency, and low-frequency bands, an initial collaborative frequency weight matrix is ​​constructed, with the matrix elements representing the initial weight percentage of each master and slave channel in different frequency bands.

[0026] Furthermore, the process of modifying and weighting the initial cooperative frequency weight matrix to obtain the optimized cooperative frequency weight matrix includes:

[0027] Based on the potential distortion spectrum stored in the heterogeneous parameter library, the initial cooperative frequency weighting matrix is ​​corrected to generate a cooperative frequency weighting matrix;

[0028] Based on the channel dynamic fingerprints in the heterogeneous parameter library and the time-series association labels of each frequency band component, an error probability cloud map is generated.

[0029] Based on the error probability cloud map, the weight allocation of the cooperative frequency weight matrix is ​​dynamically adjusted to obtain the optimized cooperative frequency weight matrix.

[0030] Furthermore, the step of modifying the initial cooperative frequency weight matrix to generate the cooperative frequency weight matrix includes:

[0031] Based on the potential distortion spectrum stored in the heterogeneous parameter library, the nonlinear distortion features of each master and slave channel are extracted. The nonlinear distortion features include amplitude distortion and phase distortion.

[0032] Calculate the amplitude distortion rate and phase distortion rate of each master and slave channel; based on the amplitude distortion rate and phase distortion rate, obtain the total distortion rate of each master and slave channel; channels with a total distortion rate greater than a preset distortion threshold are defined as severely distorted channels.

[0033] For the high-frequency and mid-frequency bands, an adaptive filtering algorithm is used to correct the corresponding elements in the initial cooperative frequency weight matrix based on the nonlinear distortion characteristics of each master and slave channel, thereby reducing the weight ratio of the severely distorted channel; for the low-frequency band, the balanced allocation scheme in the initial cooperative frequency weight matrix remains unchanged.

[0034] The final collaborative frequency weight matrix is ​​generated by combining the weight correction results of high frequency, mid frequency and low frequency bands.

[0035] Furthermore, the generation of the error probability cloud map includes:

[0036] For high-frequency pulse components, the predicted time offset values ​​of high-frequency pulse components in each master and slave channel are obtained based on the channel dynamic fingerprint and high-frequency timing association tag.

[0037] For intermediate frequency harmonic components, the phase separation prediction spectrum of intermediate frequency harmonic components in each master and slave channel is obtained based on the channel dynamic fingerprint and intermediate frequency timing association tag.

[0038] For the low-frequency base component, the steady-state error prediction range of the low-frequency base component of each master and slave channel is obtained based on the channel dynamic fingerprint and low-frequency timing association label.

[0039] By integrating the predicted time offset of high-frequency pulse components, the predicted phase separation of mid-frequency harmonic components, and the predicted steady-state error range of low-frequency base components, the error probability cloud map of the overall tracking performance of each master and slave channel is generated using the membership weighting method.

[0040] Furthermore, the step of obtaining the predicted timing offset values ​​of the high-frequency pulse components of each master and slave channel includes:

[0041] Based on high-frequency pulse component-based high-frequency time-series association tags, characteristic time-scale parameters of high-frequency pulses are extracted, including average pulse width and pulse interval time.

[0042] By correlating the temporal phase offset margin in the channel dynamic fingerprint with the characteristic time scale parameters of the high-frequency pulse, a prediction value of the temporal offset of the high-frequency pulse components in each master and slave channel is generated.

[0043] Furthermore, the step of dynamically adjusting the weight allocation of the cooperative frequency weight matrix based on the error probability cloud map to obtain the optimized cooperative frequency weight matrix includes:

[0044] Based on the predicted value A1 of the timing offset of the high-frequency pulse components of each master and slave channel in the error probability cloud map, the weight allocation of the high-frequency band is optimized.

[0045] Based on the phase separation prediction spectrum of the intermediate frequency harmonic components of each channel in the error probability cloud map, the weight allocation of the intermediate frequency band is optimized.

[0046] Based on the steady-state error prediction interval of the low-frequency basis components of each channel in the error probability cloud map, the weight allocation of the low-frequency band is optimized.

[0047] By combining the optimized weight allocation results of high-frequency, mid-frequency, and low-frequency bands, an optimized collaborative frequency weight matrix is ​​generated.

[0048] Furthermore, the optimization of low-frequency band weight allocation based on the steady-state error prediction interval of the low-frequency basis components of each channel in the error probability cloud map includes:

[0049] Define the upper limit of the steady-state error prediction interval as Eu, the lower limit as El, the midpoint of the interval as Ec=(Eu+El) / 2, and the radius of the interval as Er=(Eu-El) / 2;

[0050] For the steady-state error prediction interval of the i-th channel [E] min,i E max,i ], where E min,i E represents the minimum value of the steady-state error prediction interval for the i-th channel. max,i The maximum value of the steady-state error prediction interval for the i-th channel: i is the index of each master and slave channel;

[0051] If E max,i If Ec ≤ Ec, then the i-th channel is a low steady-state error channel. A reward-based low-frequency weight increment is applied to the i-th channel, where the low-frequency weight increment is proportional to (Ec - Ec). max,i It is directly proportional to Er;

[0052] If E min,i If the value is greater than or equal to Ec, then the i-th channel is a high steady-state error channel. A punitive low-frequency weight reduction is applied to the i-th channel, where the low-frequency weight reduction is proportional to (Ec). min,i -Ec) / Er is directly proportional;

[0053] If E min,i <Ec<E max,i， Then the i-th channel is the medium steady-state error channel, and the low-frequency weight of the i-th channel remains unchanged.

[0054] Furthermore, the master-slave synchronization and cross-coupling control strategy includes:

[0055] Using the real-time load signal of the main channel as the synchronization reference, the optimized cooperative frequency weight matrix is ​​input to the hydraulic servo controllers corresponding to each main and slave channel;

[0056] The hydraulic servo controller of the main channel directly uses the weights of each frequency band in the optimized cooperative frequency weight matrix of the main channel as the spectrum synthesis coefficients to synthesize the hydraulic control commands of the main channel and control the opening output of the hydraulic valve of the main channel.

[0057] The hydraulic servo controller of the slave channel introduces a cross-coupling control term during the spectrum synthesis process. The cross-coupling control term is obtained by adjusting the deviation of the real-time load signals of the master and slave channels by a PID controller. The cross-coupling control term is superimposed on the weighting coefficients of each frequency band of the slave channel to correct the spectrum synthesis result of the slave channel and finally generate the hydraulic control command of the slave channel to control the opening output of the hydraulic valve of the slave channel.

[0058] The coordinated opening output of the master and slave hydraulic valves is applied to the object under test via hydraulic cylinders, and dynamic loads are applied synchronously at multiple points to complete the closed-loop tracking control of the load spectrum.

[0059] A real-time load spectrum tracking control system, used to implement the above-described real-time load spectrum tracking control method, the system comprising:

[0060] Channel dynamic fingerprint construction module: Determine the master channel and slave channel of the hydraulic servo system in the dynamic compression and shear testing machine, collect the hysteresis response curves of the hydraulic valves of each master and slave channel and the potential distortion spectrum of the servo controller; Based on the hysteresis response curves of the hydraulic valves of each master and slave channel, construct the channel dynamic fingerprint of each master and slave channel, and store the channel dynamic fingerprint and potential distortion spectrum into the heterogeneous parameter library.

[0061] Frequency weight matrix construction module: Performs three-level decomposition of the target payload spectrum to obtain three frequency band components, constructs time-series association labels for each frequency band component; merges the channel dynamic fingerprints of the master and slave channels, and combines them with the three frequency band components to generate an initial collaborative frequency weight matrix;

[0062] Frequency weight matrix optimization module: Based on the potential distortion spectrum, channel dynamic fingerprint and time-series association labels of each frequency band component stored in the heterogeneous parameter library, the initial cooperative frequency weight matrix is ​​corrected and weighted to obtain the optimized cooperative frequency weight matrix.

[0063] Cooperative control module: The optimized cooperative frequency weight matrix is ​​input into the hydraulic servo controller. Based on the master-slave synchronization and cross-coupling control strategy, the spectrum tracking output of multiple hydraulic servo channels is controlled in real time.

[0064] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0065] This invention establishes a channel dynamic fingerprint by identifying master and slave channels and collecting relevant data, providing a foundation for subsequent precise control. In the decomposition of the target load spectrum, construction of time-series correlation labels, and generation and optimization of the cooperative frequency weight matrix, the characteristics of each channel and load are fully considered, effectively compensating for the effects caused by differences in hydraulic systems, varying frictional resistance, and independent control algorithms. Based on the optimized matrix and master-slave synchronization and cross-coupling control strategies, multi-channel spectrum tracking output is achieved. This overall scheme comprehensively improves the synchronization of the load spectrum of each channel in time and phase during multi-channel dynamic compression-shear tests, greatly reducing timing deviations and phase differences. It allows the test results to accurately reproduce the dynamic response of complex structures or systems under actual working conditions, providing reliable data support for seismic performance testing of large and complex structures, dynamic performance testing of robot joints, and aerospace structural component testing. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating the principle of the real-time load spectrum tracking control method of the present invention.

[0068] Figure 2 The flowchart of the method for constructing channel dynamic fingerprints for each master and slave channel in this invention is shown below;

[0069] Figure 3 The flowchart of the method for constructing time-series correlation labels for each frequency band component by performing three-level decomposition of the target load spectrum according to the present invention is shown below.

[0070] Figure 4 This is a flowchart of the method for generating the initial cooperative frequency weight matrix according to the present invention;

[0071] Figure 5 This is a flowchart of the method for modifying the initial cooperative frequency weight matrix to generate a cooperative frequency weight matrix according to the present invention;

[0072] Figure 6 This is a flowchart of the method for obtaining the timing offset prediction value of the high-frequency pulse components of each master and slave channel according to the present invention;

[0073] Figure 7 This is a flowchart of the method for obtaining the phase separation quantity prediction spectrum of the intermediate frequency harmonic components of the master and slave channels according to the present invention;

[0074] Figure 8 This is a functional block diagram of the real-time load spectrum tracking and control system of the present invention. Detailed Implementation

[0075] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0076] Example 1

[0077] Please see Figure 1 As shown, this embodiment provides a real-time load spectrum tracking control method, including:

[0078] Step S1000: Determine the master channel and slave channel of the hydraulic servo system in the dynamic compression and shear testing machine, collect the hysteresis response curves of the hydraulic valves in each master and slave channel and the potential distortion spectrum of the servo controller; based on the hysteresis response curves of the hydraulic valves in each master and slave channel, construct the channel dynamic fingerprint of each master and slave channel, and store the channel dynamic fingerprint and potential distortion spectrum into the heterogeneous parameter library.

[0079] Further, step S1000 includes:

[0080] Step S1100: Designate one hydraulic servo channel of the dynamic compression-shear testing machine as the main channel and the remaining hydraulic servo channels as slave channels; perform no-load dynamic scanning on the main channel and slave channels respectively, and collect the hysteresis response curves of the hydraulic valves of each main and slave channel and the potential distortion spectrum of the servo controller; the number of slave channels is n1, n1≥1;

[0081] Specifically, in dynamic compression-shear testing scenarios requiring multi-channel coordination, such as seismic performance testing of large and complex structures (e.g., bridges, building models), dynamic performance testing of multi-degree-of-freedom robot joints, and aerospace structural component testing simulating complex environmental loads, differences in the characteristics of the hydraulic systems of each channel, varying frictional resistance, and the independence of the control algorithms can lead to timing deviations and phase differences in the load spectra of each channel, affecting the accuracy of the test results. To address this issue, step S1100 involves determining the master and slave channels and acquiring relevant curves. When determining the master and slave channels, one hydraulic servo channel of the dynamic compression-shear testing machine is designated as the master channel, and the remaining hydraulic servo channels are designated as slave channels. The master channel serves as the benchmark for the entire system's load spectrum tracking, and its load spectrum tracking target determines the standard of the entire test loading. The slave channels coordinate with the master channel to apply the load spectrum, and each slave channel needs to maintain synchronization with the master channel while tracking its own target load spectrum. Taking bridge seismic performance testing as an example, if the channel simulating the loading of the main stress points at the bottom of the bridge pier is set as the main channel, then the channel simulating the loading of other stress points in the bridge body is the secondary channel. The main channel simulates the stress situation at the bottom of the bridge pier under the action of seismic waves, and the secondary channel simulates the stress situation of the corresponding parts of the bridge body according to the loading situation of the main channel, thus jointly restoring the stress state of the bridge in an earthquake.

[0082] After determining the master and slave channels, an unloaded dynamic scan (frequency domain 0.1-50Hz) is performed on both channels. The hysteresis response curves acquired during the unloaded dynamic scan are used to characterize the dynamic response characteristics of the hydraulic valve to control signals of different frequencies. Control signals of different frequencies are equivalent to excitations under different operating conditions, and the hydraulic valve's response under these excitations reflects its performance. For example, in aerospace structural component testing, when simulating vibration excitations experienced by an aircraft under different flight conditions, the hysteresis response curve of the hydraulic valve can demonstrate its ability to respond to control signals promptly and accurately, adjusting the loading force. The potential distortion spectrum reflects the distortion of the control voltage output by the servo controller in different frequency bands. Control voltage distortion is caused by factors such as the characteristics of internal circuit components of the servo controller and signal transmission interference; the potential distortion spectrum helps to understand the quality of the controller's output signal. In the dynamic performance testing of multi-degree-of-freedom robot joints, if the control voltage is distorted, it will cause deviations in the joint loading force, affecting the accurate testing of joint performance. By obtaining hysteresis response curves and potential distortion spectra through no-load calibration, the inherent dynamic characteristics of each master and slave channel can be comprehensively characterized, providing crucial data support for subsequent analysis of channel performance differences and the achievement of precise synchronous control. This operation effectively solves the synchronization difficulties caused by the inherent characteristics of the channels, laying the foundation for achieving high-precision time-synchronized load spectrum tracking control.

[0083] Step S1200: Based on the hysteresis response curves of the master and slave hydraulic valves, construct the channel dynamic fingerprints of the master and slave channels through nonlinear regression, and store the channel dynamic fingerprints and potential distortion spectra into the heterogeneous parameter library.

[0084] Furthermore, such as Figure 2 As shown, step S1200 includes:

[0085] Step S1210: Based on the hysteresis response curves of the master and slave hydraulic valves, the Hammerstein model is used to describe the nonlinear dynamic characteristics of the hydraulic valves. The model parameters are fitted by the least squares method to obtain the nonlinear dynamic model of the master and slave hydraulic valves.

[0086] Step S1220: For the nonlinear dynamic model of the hydraulic valves in each master and slave channel, extract the time-domain phase offset margin and frequency-domain gain attenuation coefficient to construct the channel dynamic fingerprint of each master and slave channel.

[0087] Step S1230: Store the channel dynamic fingerprints and potential distortion spectra of each master and slave channel into the heterogeneous parameter library.

[0088] Specifically, in actual multi-channel dynamic compression-shear tests, the operating characteristics of hydraulic valves are not a simple linear relationship, but rather exhibit complex nonlinear characteristics. To accurately describe these characteristics, step S1210 uses the Hammerstein model. The Hammerstein model consists of a static nonlinear element and a linear dynamic element connected in series, which can effectively simulate the output characteristics of a hydraulic valve under control signal input. When constructing the model, the model parameters are fitted using the least squares method. The principle of the least squares method is to determine the model parameters by minimizing the sum of squared errors between the model's predicted values ​​and the actual observed values. For example, in seismic performance tests of building models, a series of hydraulic valve output data are collected under different frequency control signal inputs. These data are substituted into the Hammerstein model, and the least squares method is used to continuously adjust the parameters in the model, such as the coefficients of the static nonlinear element and the transfer function parameters of the linear dynamic element, so that the model output is as close as possible to the actual hydraulic valve output. The nonlinear dynamic models of the master and slave hydraulic valves obtained in this way can more accurately reflect the working characteristics of the hydraulic valves under different working conditions, providing an accurate model basis for the subsequent construction of channel dynamic fingerprints, which helps to improve the analysis accuracy of the control performance of each channel, and thus improve the accuracy of multi-channel control.

[0089] To further quantify the nonlinear dynamic characteristics of the master and slave hydraulic valves, step S1220 extracts the time-domain phase offset margin and frequency-domain gain attenuation coefficient from the constructed nonlinear dynamic model to build a channel dynamic fingerprint. The time-domain phase offset margin reflects the degree of phase shift of the hydraulic valve relative to the input signal in the time domain. In multi-channel experiments, the consistency of phase across channels is crucial. For example, in a deformation test simulating an aircraft wing under complex airflow, if the time-domain phase offset margins of the hydraulic valves in each loading channel differ significantly, the application of loading forces to different parts of the wing will be asynchronous in time, failing to accurately simulate the actual stress conditions. The frequency-domain gain attenuation coefficient reflects the attenuation of signal gain by the hydraulic valve at different frequencies. During transmission, load signals of different frequencies experience gain attenuation due to the inherent characteristics of the hydraulic system. In bridge seismic tests simulating seismic wave loading, seismic waves contain multiple frequency components. If the hydraulic valve's gain attenuation for certain frequency signals is too large, the loading force will not accurately reflect the true excitation of the seismic wave. By extracting these two key parameters to construct a channel dynamic fingerprint, the dynamic characteristics of each channel hydraulic valve can be comprehensively and uniquely identified. This helps to adjust the control strategy in a targeted manner according to the differences in characteristics of each channel during the subsequent control process, so as to achieve more precise synchronous control and effectively solve the problem of load spectrum synchronization caused by the differences in channel characteristics.

[0090] Step S1230 stores the channel dynamic fingerprints and potential distortion spectra of each master and slave channel into a heterogeneous parameter library. A heterogeneous parameter library is a database capable of storing data of different types and formats, with the advantage of integrating data resources from multiple sources and in various forms. In this experimental scenario, the channel dynamic fingerprints and potential distortion spectra have different formats and types; the heterogeneous parameter library can uniformly store these data, facilitating subsequent retrieval and analysis. For example, during the dynamic performance testing of a multi-degree-of-freedom robot joint, new experimental data is continuously generated as the test progresses. The heterogeneous parameter library can store the new channel dynamic fingerprints and potential distortion spectra data together with the previous data. When it is necessary to analyze the performance of a certain channel or adjust the control strategy, the relevant data can be directly obtained from the heterogeneous parameter library. This storage method improves the efficiency of data management and the reusability of data, providing strong data support for achieving high-precision time-synchronous load spectrum tracking control in multi-channel dynamic compression-shear tests. It ensures that throughout the entire test process, the control algorithm can be continuously optimized based on this accurate data, improving control accuracy.

[0091] Step S2000: The target load spectrum is decomposed into three frequency band components: high-frequency pulse component, mid-frequency harmonic component, and low-frequency base component. Temporal correlation labels for each frequency band component are constructed. Channel dynamic fingerprints of the master and slave channels are fused, and combined with the three frequency band components, an initial cooperative frequency weighting matrix is ​​generated. Based on the potential distortion spectrum stored in the heterogeneous parameter library, the initial cooperative frequency weighting matrix is ​​corrected to generate a cooperative frequency weighting matrix. Based on the channel dynamic fingerprints in the heterogeneous parameter library and the constructed temporal correlation labels for each frequency band component, an error probability cloud map is generated. According to the error probability cloud map, the weight allocation of the cooperative frequency weighting matrix is ​​dynamically adjusted to obtain an optimized cooperative frequency weighting matrix.

[0092] Further, step S2000 includes:

[0093] Step S2100: The target load spectrum is decomposed into three levels using the wavelet packet transform algorithm to obtain three frequency band components, namely, high-frequency pulse component, mid-frequency harmonic component and low-frequency base component; and time-series association labels for each frequency band component are constructed.

[0094] In multi-channel dynamic compression-shear tests, differences in hydraulic system characteristics, frictional resistance, and the independence of control algorithms in each channel can easily lead to timing deviations and phase differences in the load spectrum of each channel, affecting the accuracy of the test results. Step S2100 aims to lay the foundation for achieving high-precision time-synchronized load spectrum tracking control by performing a three-level decomposition of the target load spectrum and constructing timing association labels for each frequency band component.

[0095] Furthermore, such as Figure 3 As shown, step S2100 includes:

[0096] Step S2110: Perform three-level decomposition on the target load spectrum to obtain three frequency band components, as shown in Table 1. The three frequency band components are high-frequency pulse component, mid-frequency harmonic component and low-frequency base component.

[0097] Table 1. Target Load Spectrum Decomposition Frequency Band Component Characteristics

[0098]

[0099] Specifically, as shown in Table 1, this step uses the wavelet packet transform algorithm to decompose the target load spectrum into three levels: high-frequency pulse components (>10Hz), mid-frequency harmonic components (1-10Hz), and low-frequency base components (<1Hz). The wavelet packet transform algorithm is a time-frequency analysis method capable of analyzing signals at different time and frequency resolutions. In multi-channel dynamic compression-shear testing, load signals of different frequency bands have different effects on the test results. For example, in tests simulating bridge seismic response, the high-frequency pulse component may correspond to short-period pulse excitation in seismic waves, which can cause significant impact on the bridge structure in a short time; the mid-frequency harmonic components may reflect various resonance phenomena during bridge vibration; and the low-frequency base components are related to the slow deformation or low-frequency swaying of the bridge as a whole. By decomposing the target load spectrum using the wavelet packet transform algorithm, the characteristics of signals in different frequency bands can be studied in more detail, and corresponding control strategies can be formulated for each frequency band, providing a prerequisite for subsequent precise synchronous control of the load spectrum of each channel. This decomposition operation helps to accurately grasp the load changes of different frequency components during the experiment, making subsequent control more targeted and improving the overall experimental system's ability to handle complex load spectra.

[0100] Step S2120: For the decomposed high-frequency pulse components, extract the timestamp sequence of the pulse peaks and construct high-frequency time series association tags;

[0101] Specifically, for the decomposed high-frequency pulse components, this step uses an extremum point tracking algorithm to extract the timestamp sequence of the pulse peaks, thereby constructing a high-frequency time-series association label. The extremum point tracking algorithm is used to identify extreme points (such as maximum and minimum values) in a signal; in this embodiment, it is used to accurately locate the peak points in the high-frequency pulse components. In actual multi-channel dynamic compression-shear tests, such as those simulating complex flight environment loads on aerospace structural components, high-frequency pulses may simulate the instantaneous impact loads experienced by an aircraft when encountering sudden airflow changes or rapid adjustments in flight attitude. The peak occurrence time of these pulses is crucial for accurately simulating the stress conditions of the aircraft. By extracting the timestamp sequence of the pulse peaks to construct a high-frequency time-series association label, the time information of each high-frequency pulse can be accurately recorded. This label can be used in subsequent control processes to determine the time synchronization of each channel when reproducing the high-frequency pulse load. For example, if the timestamps in the high-frequency time-series association labels of two channels differ significantly, it indicates a significant time asynchrony problem between the two channels when loading the high-frequency pulse load. This method of quantifying time synchronization by constructing high-frequency time-series correlation tags provides crucial data support for subsequent adjustment of control parameters of each channel and the achievement of high-precision time synchronization, effectively solving the time synchronization problem in the process of high-frequency pulse load loading.

[0102] Step S2130: For the intermediate frequency harmonic components, extract the phase reference points of each harmonic component and construct intermediate frequency timing association tags;

[0103] Specifically, for intermediate frequency (IF) harmonic components, this step uses Hilbert transform to extract phase reference points (such as the zero point of a sine wave) for each harmonic component, and then constructs IF timing association tags. The Hilbert transform is a mathematical transformation commonly used in signal processing to analyze signal phase information. In multi-channel dynamic compression-shear tests, taking the simulation of multi-degree-of-freedom robot joint dynamic performance testing as an example, IF harmonic components may be generated by the complex motion of the robot joint, and their phase information reflects the relative position and time relationship of the joint motion. Extracting phase reference points and constructing IF timing association tags through Hilbert transform can accurately capture the phase characteristics of each harmonic component. In subsequent control, this phase information can be used to determine the phase consistency of each channel when applying IF harmonic loads. For example, in the dynamic performance testing of robot joints, if the phase information reflected by the IF timing association tags of two channels differs significantly, it indicates that there is a phase deviation between the two channels when simulating joint motion, which will affect the accurate evaluation of the robot joint's dynamic performance. Therefore, constructing intermediate frequency timing association tags helps to achieve precise control of the loading phase of intermediate frequency harmonic loads in each channel, improves the control accuracy of intermediate frequency harmonic load synchronization during the test, and enables the test results to more accurately reflect the actual situation.

[0104] Step S2140: For the low-frequency base component, directly add an absolute timestamp as a low-frequency timing association tag;

[0105] Specifically, for the low-frequency base component, this step directly adds an absolute timestamp as a low-frequency time-series association label. In multi-channel dynamic compression-shear test scenarios, such as seismic performance tests of large building models, the low-frequency base component typically reflects the slow deformation or low-frequency vibration of the building structure over a longer timescale, and its changes are relatively stable. Adding an absolute timestamp provides a unified time reference for the low-frequency base component, used to measure the time accuracy of each channel when applying low-frequency base loads. In the test, if the absolute timestamps of the low-frequency base components of each channel are the same, it means that they are applied low-frequency loads at the same time, ensuring the time consistency of low-frequency loading. Unlike high-frequency pulse components and mid-frequency harmonic components, the low-frequency base component has a lower frequency of change. It does not require complex algorithms to extract specific features to construct time-series association labels as the former two. Directly adding an absolute timestamp is simple and meets the requirements of low-frequency loading time synchronization control, improving the efficiency and accuracy of the entire test system in low-frequency load loading and helping to accurately simulate the response of building structures under low-frequency excitation.

[0106] Step S2150: Use the high-frequency timing association tag of the high-frequency pulse component, the intermediate-frequency timing association tag of the intermediate-frequency harmonic component, and the low-frequency timing association tag of the low-frequency base component as the timing association tag of each frequency band component.

[0107] Specifically, this step uses the high-frequency timing association tags of the high-frequency pulse component, the mid-frequency timing association tags of the mid-frequency harmonic component, and the low-frequency timing association tags of the low-frequency base component as the timing association tags for each frequency band component. Throughout the multi-channel dynamic compression-shear test, load signals from different frequency bands are interconnected, collectively forming a complex load spectrum. By integrating these timing association tags from different frequency bands, a complete time reference system can be formed, enabling precise synchronization of multi-channel control commands. Taking bridge seismic performance testing as an example, under seismic wave action, the bridge structure simultaneously bears loads from multiple frequency components, including high-frequency pulses, mid-frequency harmonics, and low-frequency base components. Each channel performs load control based on its respective frequency band's timing association tags, ensuring high consistency in time and phase, accurately simulating the actual stress state of the bridge during an earthquake. This integrated operation solves the key problem of synchronous control of multi-channel load spectrum in different frequency bands. Through a unified time-series association tag system, each channel can work closely together when loading loads in different frequency bands, improving the reliability and accuracy of the test results. It provides a unified and accurate time basis for the subsequent generation and adjustment of the frequency weight matrix based on these time-series association tags, thereby achieving high-precision time-synchronized load spectrum tracking control.

[0108] Step S2200: Fuse the channel dynamic fingerprints of the master and slave channels and combine them with the three frequency band components to generate an initial collaborative frequency weighting matrix; the matrix elements of the initial collaborative frequency weighting matrix are the weight ratios of the master and slave channels in different frequency bands.

[0109] In multi-channel dynamic compression-shear tests, achieving high synchronization of the load spectra of each channel in terms of time and phase is a key objective. Step S2200 aims to generate an initial cooperative frequency weighting matrix by fusing the channel dynamic fingerprints of the master and slave channels and combining them with the three frequency band components obtained from the decomposition of the target load spectrum. This provides crucial foundational support for subsequent high-precision time-synchronized load spectrum tracking control. This matrix is ​​used to determine the weight ratio of each master and slave channel when loads are applied at different frequency bands. A reasonable weight allocation can optimize the cooperative working effect of each channel and ensure the synchronization of the load spectra of each channel during the test.

[0110] Furthermore, such as Figure 4 As shown, step S2200 includes:

[0111] Step S2210: Based on the channel dynamic fingerprints of each master and slave channel, for the high-frequency pulse component, according to the time-domain phase offset margin of each channel, the initial weight ratio of the high-frequency band of each master and slave channel is determined by fuzzy inference method.

[0112] Specifically, the time-domain phase offset margin reflects the degree of phase shift of the hydraulic valve relative to the input signal in the time domain, reflecting the differences in dynamic characteristics of each channel in response to high-frequency pulse signals. Due to the different characteristics of each channel's hydraulic system, their phase offset responses to high-frequency pulse signals also differ. Fuzzy inference is a decision-making method based on fuzzy logic, capable of handling uncertainties and fuzzy information. In this case, the fuzzy inference method uses the time-domain phase offset margin of each channel as input, and infers the initial weight ratio of each channel in the high-frequency band through pre-defined fuzzy rules and membership functions. For example, in a test scenario simulating high-frequency vibration of aerospace structural components, assume there is one master channel and two slave channels. The master channel has a smaller time-domain phase offset margin, meaning it can follow signal changes more quickly and accurately when responding to high-frequency pulse signals; while slave channel 1 has a larger time-domain phase offset margin, and slave channel 2 has a time-domain phase offset margin at an intermediate level. Using fuzzy inference, the main channel is given a relatively high weight in the high-frequency band because it performs better in high-frequency response. For slave channel 1, due to its large phase shift, which may cause signal transmission delay, a lower weight is given. The weight of slave channel 2 is set at a moderate level based on its phase shift. The purpose of this is to allow the channel with better response characteristics to undertake more loading tasks when multiple channels are collaboratively loading high-frequency pulse components, thereby optimizing the overall loading effect.

[0113] This weight determination method, based on time-domain phase offset margin and fuzzy inference, can fully consider the differences in high-frequency response among the channels. Its beneficial effect lies in improving the synchronization of channels during high-frequency pulse component loading by rationally allocating weights. By distributing more weights to channels with more accurate and rapid responses, it reduces the loading asynchrony problem caused by large phase offsets in some channels, allowing high-frequency pulse loads to be applied more precisely to the test object. This effectively solves the synchronization problem during multi-channel high-frequency pulse loading, helps improve the accuracy of experimental results, and ensures that the experiment can more realistically simulate the stress conditions of aerospace structural components under actual working conditions.

[0114] Step S2220: For the intermediate frequency harmonic components, the initial weight ratio of the intermediate frequency band of each master and slave channel is determined by the frequency weighting method based on the frequency domain gain attenuation coefficient of each channel.

[0115] Specifically, the frequency domain gain attenuation coefficient reflects the degree of signal gain attenuation of the hydraulic valve at different frequencies, reflecting the differences in the ability of each channel to process intermediate frequency harmonic signals. Different channels of the hydraulic system exhibit varying gain attenuation during the transmission of intermediate frequency harmonic signals due to their inherent characteristics. The frequency weighting method allocates weights based on the signal frequency and the processing capability of each channel for that frequency. Taking the simulation of the dynamic performance test of a multi-degree-of-freedom robot joint as an example, in the experiment, each channel is responsible for simulating the forces acting on different parts of the joint. Assume that the intermediate frequency harmonic signals generated by the joint movement are mainly concentrated in a few specific frequency bands. The main channel exhibits smaller gain attenuation in one frequency band, indicating its strong transmission and amplification capability for that band; channel 1 exhibits larger gain attenuation in another frequency band, while channel 2 shows moderate gain attenuation across all intermediate frequency harmonic bands. The frequency weighting method calculates weights for each frequency band based on the gain attenuation coefficients of each channel in different intermediate frequency harmonic bands. For frequency bands with low gain attenuation in the main channel, higher weights are assigned because they can more effectively transmit and load signals in that band. For frequency bands with high gain attenuation from channel 1, lower weights are assigned. The weight of channel 2 is determined comprehensively based on its gain attenuation across various frequency bands. The purpose of this method is to ensure that when loading intermediate frequency (IF) harmonic components, each channel can reasonably distribute the loading task according to its processing capabilities for different frequency signals. Its beneficial effect is improved accuracy and synchronization of IF harmonic loading. By rationally allocating weights, loading deviations caused by insufficient processing capabilities of some channels for certain IF harmonic frequencies are avoided. This allows each channel to load IF harmonic loads more coordinatedly, more accurately simulating the force conditions of robot joints during actual movement, thereby improving the reliability of experimental results and providing more accurate data support for evaluating the dynamic performance of robot joints.

[0116] Step S2230: For the low-frequency base component, a balanced allocation strategy is adopted to evenly distribute the weight of the low-frequency band to each master and slave channel.

[0117] Specifically, in multi-channel dynamic compression-shear tests, the low-frequency base component typically reflects the slow deformation or low-frequency vibration of the tested object over a longer timescale. In practical applications, such as seismic performance tests of large building models, the low-frequency base component simulates the overall swaying or slow deformation of the building structure under the action of low-frequency seismic waves. Because the changes in the low-frequency base component are relatively stable, the differences in dynamic characteristics among the channels have a relatively small impact on the loading effect when processing low-frequency signals. For example, in a test simulating a large bridge under the action of low-frequency seismic waves, a main channel and multiple slave channels are set up to simulate the forces on different parts of the bridge. When loading the low-frequency base component, it is not necessary to allocate weights based on the complex characteristics of each channel, as is done with high-frequency pulse components or mid-frequency harmonic components. Instead, a balanced allocation strategy is adopted, distributing the weight of the low-frequency band evenly among the channels, so that each channel undertakes the same workload when loading the low-frequency base component. The purpose of this is to simplify the weight allocation process of the low-frequency band while ensuring that each channel maintains a relatively consistent working state during low-frequency loading. The advantages of this strategy are twofold. First, it reduces the complexity of weight allocation, decreasing computational load and control difficulty. When processing low-frequency signals, complex parameter analysis and weight calculations are unnecessary, improving the efficiency of experimental control. Second, by distributing weights evenly, it ensures consistency across channels during low-frequency loading, avoiding overloading or underloading of some channels due to unreasonable weight allocation. This helps to more accurately simulate the overall response of large building structures under low-frequency excitation, improving the reliability and stability of experimental results in the low-frequency range.

[0118] Step S2240: Based on the weight allocation results of high frequency band, mid frequency band and low frequency band, construct an initial collaborative frequency weight matrix. The matrix elements are the initial weight ratios of each master and slave channel in different frequency bands.

[0119] Specifically, in the aforementioned steps, different weighting methods were adopted for the high-frequency pulse component, mid-frequency harmonic component, and low-frequency base component, based on the different characteristics of each channel. These weighting results reflect the capability and workload of each channel under load at different frequency bands. Taking a multi-channel dynamic compression-shear test system containing one main channel and two slave channels as an example, assuming that step S2210 determines the weight ratio of the main channel in the high-frequency band to be 0.5, slave channel 1 to 0.2, and slave channel 2 to 0.3; step S2220 determines the weight ratio of the main channel in the mid-frequency band to be 0.4, slave channel 1 to 0.3, and slave channel 2 to 0.3; and step S2230, the weights in the low-frequency band are evenly distributed, with the main channel, slave channel 1, and slave channel 2 each having a weight ratio of 1 / 3. These weighting results for different frequency bands are integrated to construct a matrix, where each row of the matrix represents a channel, each column represents a frequency band, and the matrix elements are the weight ratio of each channel in the corresponding frequency band.

[0120] The purpose of constructing the initial cooperative frequency weight matrix is ​​to provide a unified basis for weight allocation in the subsequent control process. Its beneficial effects are multifaceted. First, it integrates weight information from different frequency bands, enabling the entire multi-channel system to work collaboratively according to pre-set weight ratios when loading loads of different frequency components, thus improving system coordination. For example, in aerospace structural component tests simulating complex environmental loads, loads from different frequency bands act on the structural component. Through the initial cooperative frequency weight matrix, each channel can orderly load loads of different frequency bands, avoiding loading chaos. Second, it provides a foundation for subsequent correction of the frequency weight matrix based on the potential distortion spectrum and optimization of weight allocation based on the error probability cloud map. The initial cooperative frequency weight matrix is ​​the starting point of the entire weight adjustment process; subsequent corrections and optimizations are all based on this, ensuring the coherence and systematic nature of the entire control process. This helps achieve high-precision time-synchronized load spectrum tracking control, allowing the test results to more accurately reflect the stress state of the tested object under actual working conditions.

[0121] Step S2300: Based on the potential distortion spectrum stored in the heterogeneous parameter library, the initial cooperative frequency weighting matrix is ​​corrected to generate a cooperative frequency weighting matrix.

[0122] In multi-channel dynamic compression-shear tests, the control voltage output by the servo controller is distorted, i.e., the potential is distorted. This affects the accuracy of loading on each channel and thus interferes with the synchronization of the multi-channel load spectrum. Step S2300 aims to correct the initial cooperative frequency weighting matrix based on the potential distortion spectrum stored in the heterogeneous parameter library, generating a more accurate cooperative frequency weighting matrix. This ensures that, even considering the influence of potential distortion, each channel can still be synchronously loaded with high precision, making the test results more accurately reflect the actual working conditions.

[0123] Furthermore, such as Figure 5 As shown, step S2300 includes:

[0124] Step S2310: Based on the potential distortion spectrum stored in the heterogeneous parameter library, extract the nonlinear distortion features of each master and slave channel. The nonlinear distortion features include amplitude distortion and phase distortion.

[0125] Specifically, this step extracts the nonlinear distortion features of each master and slave channel based on the potential distortion spectrum stored in the heterogeneous parameter library. These features include amplitude distortion and phase distortion. The potential distortion spectrum records the distortion of the control voltage output by the servo controller in different frequency bands. Amplitude distortion refers to the difference between the signal amplitude and the ideal amplitude. In multi-channel dynamic compression-shear tests, this can cause the actual loading force to deviate from the expected value. For example, in simulating bridge seismic tests, amplitude distortion in a certain channel may cause the loading force at the simulated pier stress point to be too large or too small, failing to realistically simulate the stress situation during an earthquake. Phase distortion refers to the deviation between the signal phase and the ideal phase. In the experiment, this can cause inconsistencies in the application time of the loading force in each channel, also affecting the accuracy of the test. Taking the simulation of the dynamic performance test of a multi-degree-of-freedom robot joint as an example, if a certain channel has phase distortion, it will cause the time sequence of force application to different parts of the joint to deviate, making it impossible to accurately simulate the stress state during actual movement. By extracting these nonlinear distortion features, we can gain a detailed understanding of the specific ways and extent in which the potential distortion of each channel affects the loading process. This provides a basis for subsequent weight adjustments and loading optimization, helps improve the accuracy of loading in each channel, and ultimately enhances the overall precision of multi-channel experiments.

[0126] Step S2320: Calculate the amplitude distortion rate and phase distortion rate of each master and slave channel; based on the amplitude distortion rate and phase distortion rate, obtain the total distortion rate of each master and slave channel; channels with a total distortion rate greater than a preset distortion threshold are defined as severely distorted channels.

[0127] Specifically, after extracting the nonlinear distortion characteristics of each master and slave channel, this step calculates the amplitude distortion rate and phase distortion rate of each master and slave channel, and obtains the total distortion rate of each master and slave channel based on these two distortion rates. The amplitude distortion rate is obtained by calculating the difference between the actual amplitude and the ideal amplitude, and then dividing by the ideal amplitude, and is used to quantify the degree of amplitude distortion; the phase distortion rate is obtained by calculating the difference between the actual phase and the ideal phase, and quantifying it in an appropriate way, reflecting the severity of phase distortion. The total distortion rate is an index obtained by comprehensively considering the amplitude distortion rate and the phase distortion rate, and can generally be calculated by a specific mathematical formula, such as adding the amplitude distortion rate and the phase distortion rate according to a certain weight. The distortion threshold can be set based on the 95% confidence interval of error statistics in historical experiments, preferably a value between 0.2 and 0.3, for example, 0.25. When the total distortion rate of a channel exceeds the preset distortion threshold, the channel is defined as a severely distorted channel. This allows for the rapid screening of channels that have a significant impact on the experimental results, providing a clear target for subsequent targeted weight adjustments. It also helps avoid significant deviations in experimental results due to channel distortion, thereby improving the reliability and accuracy of the experiment.

[0128] Step S2330: For the high-frequency and mid-frequency bands, based on the nonlinear distortion characteristics of each master and slave channel, an adaptive filtering algorithm is used to correct the corresponding elements in the initial cooperative frequency weight matrix to reduce the weight ratio of the severely distorted channel.

[0129] Specifically, adaptive filtering is an algorithm that automatically adjusts filter parameters based on changes in the input signal. It adjusts the weights of corresponding channels in the initial cooperative frequency weighting matrix for the high and mid-frequency bands based on the amplitude and phase distortion of each channel. For example, in an experimental system with one master channel and two slave channels, assuming slave channel 1 has a high amplitude distortion rate and significant phase distortion in the high-frequency band, the adaptive filtering algorithm would reduce its weight in this band. This is because the distortion of this channel affects the accuracy of high-frequency loading; if the loading task is still allocated according to the initial weights, the overall loading effect in the high-frequency band will deteriorate. Reducing its weight and relatively increasing the weights of other channels with smaller distortions allows the overall system to achieve a loading effect closer to the ideal state in the high-frequency band. A similar approach is used for the mid-frequency band. This approach optimizes the weight distribution of each channel in the high-frequency and mid-frequency bands, reduces the negative impact of severely distorted channels on the overall loading effect, improves the accuracy and synchronization of load loading in the high-frequency and mid-frequency bands of multiple channels, and makes the test results more accurately reflect the response of the tested object under high-frequency and mid-frequency loads in actual working conditions.

[0130] Step S2340: For the low-frequency band, keep the balanced allocation scheme in the initial cooperative frequency weight matrix unchanged;

[0131] Specifically, in multi-channel dynamic compression-shear tests, the low-frequency base component typically reflects the slow deformation or low-frequency vibration of the tested object over a longer timescale, and its changes are relatively stable. Compared to high-frequency pulse components and mid-frequency harmonic components, the low-frequency base component is less sensitive to potential distortion. For example, in seismic performance tests simulating large building models, the low-frequency base component simulates the overall slow swaying of the building structure under the action of low-frequency seismic waves. Even with some potential distortion, the difference in loading in the low-frequency band among the channels has a relatively small impact on the overall test results. Therefore, maintaining the initial balanced allocation scheme and distributing the weight of the low-frequency band equally among the master and slave channels can simplify the calculation process, ensure relatively balanced loading in the low-frequency band for each channel, avoid introducing new errors due to excessive weight adjustment, help maintain the stability of the low-frequency loading, and enable the test results to accurately reflect the actual working conditions in the low-frequency band.

[0132] Step S2350: Combine the weight correction results of high frequency band, mid frequency band and low frequency band to generate the final cooperative frequency weight matrix.

[0133] Specifically, in the preceding steps, weights were adjusted for the high-frequency and mid-frequency bands based on the nonlinear distortion characteristics of the channels, while the initial balanced allocation scheme was maintained for the low-frequency band. These weight adjustment results for different frequency bands were integrated to form a new matrix, the final cooperative frequency weight matrix. The elements of this matrix represent the weight percentages of each master and slave channel in different frequency bands, considering the influence of potential distortion. The purpose of generating the final cooperative frequency weight matrix is ​​to provide a precise weight allocation basis that comprehensively considers potential distortion factors for subsequent multi-channel loading control. Its beneficial effect is that, through reasonable adjustment and integration of the weights of each frequency band, the entire multi-channel system can optimize the allocation according to the actual situation of each channel when loading loads with different frequency components, effectively reducing the impact of potential distortion on loading, improving the synchronization accuracy and loading accuracy of the load spectrum of each channel, and thus enabling the test results to more realistically reflect the dynamic response of complex structures or systems under actual working conditions, providing strong support for high-precision control of multi-channel dynamic compression-shear tests.

[0134] Step S2400: Based on the channel dynamic fingerprint in the heterogeneous parameter library and the constructed time-series association labels of each frequency band component, generate an error probability cloud map;

[0135] In multi-channel dynamic compression-shear tests, due to differences in the characteristics of the hydraulic systems of each channel, varying frictional resistance, and the independence of the control algorithms, uncertainties exist in the tracking of the target load spectrum by each channel. This can lead to discrepancies between the test results and the actual stress state. Step S2400 aims to generate an error probability cloud map by analyzing the channel dynamic fingerprints in the heterogeneous parameter library and the time-series correlation labels of each frequency band component constructed in step S2100. This quantifies the uncertainty risk of the tracking performance of each channel, providing a basis for subsequent adjustment of the cooperative frequency weight matrix, thereby achieving high-precision time-synchronized load spectrum tracking control.

[0136] Further, step S2400 includes:

[0137] Step S2410: For the high-frequency pulse component, based on the channel dynamic fingerprint and high-frequency timing association tag, obtain the predicted value of the timing offset of the high-frequency pulse component of each master and slave channel.

[0138] Furthermore, such as Figure 6 As shown, step S2210 includes:

[0139] Step S2411: Based on the high-frequency time-series association label of the high-frequency pulse component, extract the characteristic time scale parameters of the high-frequency pulse, including the average pulse width and pulse interval time.

[0140] Step S2412: Correlation analysis is performed between the temporal phase offset margin in the channel dynamic fingerprint and the characteristic time scale parameters of the high-frequency pulse to generate the predicted value of the temporal offset of the high-frequency pulse components of each master and slave channel.

[0141] Specifically, in multi-channel dynamic compression-shear tests, high-frequency pulse components often simulate instantaneous impacts or high-frequency vibrations experienced by the test object. For example, in tests simulating high-frequency vibrations caused by sudden airflow changes during flight of aerospace structural components, each channel needs to accurately reproduce these high-frequency pulse loads to accurately assess the performance of the structural components. Accurately obtaining the predicted time offset of the high-frequency pulse components in each channel is crucial for determining the time synchronization of each channel when reproducing the high-frequency pulse loads.

[0142] The high-frequency time-series association tags are constructed from the timestamp sequence of pulse peaks extracted using the extreme point tracking algorithm in step S2120. In practice, taking a series of high-frequency pulse signals as an example, the extreme point tracking algorithm traverses the entire signal sequence, identifies the peak point of each pulse, and records its timestamp. By analyzing these timestamps, the time interval between adjacent peak points is calculated, and the average of the time intervals between all adjacent peak points is calculated to obtain the pulse interval time. Simultaneously, the duration of each pulse from start to end is measured, and the average duration of all pulses is calculated to obtain the average pulse width. These characteristic timescale parameters reflect the temporal characteristics of high-frequency pulses and are of great significance for subsequently evaluating the time synchronization of each channel when reproducing high-frequency pulse loads.

[0143] The time-domain phase offset margin in the channel dynamic fingerprint is extracted from the nonlinear dynamic model of the hydraulic valves in each master and slave channel in step S1220. It reflects the degree of phase offset of the hydraulic valve to the input signal in the time domain, and reflects the differences in dynamic characteristics of each channel when responding to high-frequency pulse signals. Here, the grey relational analysis algorithm is used for correlation analysis. The grey relational analysis algorithm is a multi-factor statistical analysis method that measures the degree of correlation between factors by calculating the correlation degree between factors. This algorithm takes the time-domain phase offset margin in the channel dynamic fingerprint and the characteristic time scale parameter of the high-frequency pulse as inputs and calculates the correlation degree between them. The higher the correlation degree, the closer the relationship between the time-domain phase offset margin and the characteristic time scale parameter of the high-frequency pulse, which means that the time-domain phase offset margin has a greater impact on the time characteristics of the high-frequency pulse. For example, if the time-domain phase offset margin of a certain channel has a high correlation with the average pulse width of the high-frequency pulse, it indicates that the phase offset of the hydraulic valve in that channel has a significant impact on the duration of the high-frequency pulse. When the phase offset of the hydraulic valve is large, it may cause a large deviation between the actual action time of the high-frequency pulse and the ideal time, which will lead to a large time delay when the channel reproduces the high-frequency pulse load, that is, a higher risk of dynamic hysteresis. Typically, the predicted time series offset value is determined based on the magnitude of the correlation, using a specific functional relationship. For example, a function can be defined. The calculated correlation degree and The function takes input as input and outputs the predicted time offset value, where... This is used to measure the correlation between the temporal phase offset margin and the average pulse width of high-frequency pulses in channel dynamic fingerprints. This measures the correlation between the temporal phase offset margin in a channel dynamic fingerprint and the pulse interval time of high-frequency pulses. This function can be determined based on the characteristics of the experimental data and experience; a common approach is to obtain it through analysis and fitting of a large amount of experimental data. Generally, the correlation... and The larger the function The larger the output value, the larger the predicted time series offset; conversely, the smaller the correlation, the smaller the predicted time series offset.

[0144] The purpose of this step is to evaluate the time synchronization of each channel when reproducing high-frequency pulse loads. Its beneficial effect is that by quantifying the predicted time offset of each channel, the time deviation of each channel during the high-frequency pulse loading process can be clearly understood. For channels with large predicted time offsets, their weight in the high-frequency band can be appropriately reduced when adjusting the collaborative frequency weighting matrix later, or other compensation measures can be taken, such as adjusting control algorithm parameters, to improve the time synchronization accuracy of each channel during high-frequency pulse loading. This effectively avoids test result deviations caused by time asynchrony between channels, allowing the test results to more accurately reflect the actual response of the tested object under high-frequency pulse loads, thus solving the time synchronization problem during high-frequency pulse loading in multi-channel dynamic compression-shear tests.

[0145] Step S2420: For the intermediate frequency harmonic components, based on the channel dynamic fingerprint and intermediate frequency timing association tag, obtain the phase separation amount prediction spectrum of the intermediate frequency harmonic components of each master and slave channel.

[0146] Furthermore, such as Figure 7 As shown, step S2420 includes:

[0147] Step S2421: Based on the time-series correlation tags of the intermediate frequency harmonic components, extract the amplitude-frequency characteristic parameters of the intermediate frequency harmonics, including the amplitude, frequency and initial phase of each harmonic.

[0148] Step S2422: Convolve the frequency domain gain attenuation coefficient in the channel dynamic fingerprint with the amplitude-frequency characteristic parameters of the intermediate frequency harmonics to generate the phase separation prediction spectrum of the intermediate frequency harmonic components of each master and slave channel.

[0149] Specifically, in multi-channel dynamic compression-shear testing applications, such as simulating the stress on multi-degree-of-freedom robot joints during complex motion, intermediate frequency (IF) harmonic components reflect the vibrations and force changes of various frequency components generated during joint movement. Accurately grasping the phase consistency of each channel when reproducing IF harmonic loads is crucial for accurately simulating the actual motion state of the joint. The IF timing association label is constructed in step S2130 using Hilbert transform to extract the phase reference points (such as the zero point of a sine wave) of each harmonic component. In practice, for an IF signal containing multiple harmonic components, Hilbert transform can convert the signal from the time domain to the complex frequency domain. In the complex frequency domain, each harmonic component has corresponding amplitude, frequency, and phase information. By analyzing the complex frequency domain data, the amplitude, frequency, and initial phase of each harmonic can be accurately extracted. These parameters comprehensively describe the characteristics of the IF harmonics; different combinations of amplitude, frequency, and initial phase determine the specific form of the IF harmonics and their effect on the tested object.

[0150] The frequency domain gain attenuation coefficient in the channel dynamic fingerprint is extracted in step S1220. It reflects the degree of signal gain attenuation by the hydraulic valve at different frequencies, demonstrating the difference in the ability of each channel to process intermediate frequency harmonic signals. Here, a cross-spectral density estimation algorithm is used for convolution operations. This algorithm can analyze the correlation between two signals in the frequency domain. Specifically, the frequency domain gain attenuation coefficient and the amplitude-frequency characteristic parameters of the intermediate frequency harmonics are first expressed as frequency domain functions. Then, according to the rules of convolution operations, these two frequency domain functions are multiplied point-by-point and integrated (or summed in the discrete case). The result is the phase dissociation prediction spectrum. Each value in the phase dissociation prediction spectrum represents the degree of phase dissociation of the intermediate frequency harmonic signal by the channel at the corresponding frequency. The larger the value, the more likely the channel is to exhibit frequency-selective distortion when reproducing the intermediate frequency harmonic load at that frequency.

[0151] The purpose of this step is to evaluate the phase consistency of the master and slave channels when reproducing mid-frequency harmonic loads. Its beneficial effect lies in the fact that by generating a phase segregation prediction spectrum, the phase deviation of each channel when processing mid-frequency harmonics can be intuitively understood. For channels with large phase deviations in the phase segregation prediction spectrum, their weight ratio in the mid-frequency band can be adjusted specifically when adjusting the collaborative frequency weight matrix subsequently, or a phase compensation algorithm can be used for correction, such as adding a phase adjustment stage to the control algorithm. This can improve the phase consistency of each channel under mid-frequency harmonic loading, making the experimental results more accurately simulate the force state of multi-degree-of-freedom robot joints in actual motion, and effectively solving the phase synchronization problem under mid-frequency harmonic loading in multi-channel dynamic compression-shear tests.

[0152] Step S2430: For the low-frequency substrate components, based on the channel dynamic fingerprint and low-frequency timing association tag, the steady-state error prediction range of the low-frequency substrate components of each master and slave channel is obtained.

[0153] Specifically, in multi-channel dynamic compression-shear tests, the low-frequency base component typically simulates the slow deformation or low-frequency vibration of the tested object over a longer timescale, such as the overall swaying of a large building model under the action of low-frequency seismic waves. Accurately assessing the accuracy of each channel in reproducing the low-frequency base load is crucial for studying the performance of the tested object under low-frequency loads. Here, a trend extrapolation algorithm is used to obtain the steady-state error prediction interval through multi-step advance prediction. The low-frequency time-series association label is constructed by directly adding absolute timestamps in step S2140, providing the time-series data foundation for the trend extrapolation algorithm. The trend extrapolation algorithm is a method of inferring the future development trend of a phenomenon based on historical data. First, historical data on the reproduction of the low-frequency base load by each channel is collected. This data contains error information between the actual load output and the target load at different time points. Then, based on this historical error data, a suitable trend extrapolation model, such as a linear regression model or an exponential smoothing model, is selected for training and parameter estimation. During training, the model parameters are continuously adjusted to optimize the model's fit to the historical error data. After model training is completed, multi-step forward prediction is performed using the trained model to predict errors at multiple future time points. For example, predicting error values ​​for five future time points yields a set of prediction error data. Based on this set of prediction error data, the maximum and minimum values ​​are calculated to determine the steady-state error prediction interval. This interval characterizes the absolute accuracy limit of each channel in reproducing low-frequency base loads. For example, in an experiment simulating a large bridge under low-frequency seismic waves, if the steady-state error prediction interval of a certain channel is small, it indicates that the channel has high accuracy in reproducing low-frequency base loads and can more accurately simulate the stress situation of the bridge under low-frequency vibrations; conversely, if the steady-state error prediction interval is large, it indicates that the channel may have large errors under low-frequency loading and needs to be adjusted in subsequent control processes.

[0154] The purpose of this step is to evaluate the steady-state tracking accuracy of each channel in reproducing low-frequency base loads. Its beneficial effect is that by determining the steady-state error prediction interval, the accuracy level of each channel under low-frequency loading can be clearly defined. For channels with larger steady-state error prediction intervals, their weight in the low-frequency band can be appropriately reduced when adjusting the cooperative frequency weight matrix subsequently, or their control parameters can be optimized, such as adjusting the gain and integral time of the hydraulic system, to improve the loading accuracy of each channel in the low-frequency band. This allows the test results to more accurately reflect the actual response of the tested object, such as a large building model, under low-frequency loads, solving the accuracy problem of low-frequency base load loading in multi-channel dynamic compression-shear tests.

[0155] Step S2440: The timing offset prediction value of the high-frequency pulse component, the phase separation prediction spectrum of the mid-frequency harmonic component, and the steady-state error prediction interval of the low-frequency base component are fused, and the error probability cloud map of the overall tracking performance of each master and slave channel is generated by the membership weighting method.

[0156] Specifically, in multi-channel dynamic compression-shear tests, it is necessary to comprehensively evaluate the tracking performance of each channel under load at different frequency bands to fully understand the system's operating status and uncertainty risks. The membership-weighted method is a data processing method based on fuzzy set theory. Fuzzy set theory allows elements to belong to a set to varying degrees, described by defining a membership function. In this step, the predicted time offset of the high-frequency pulse component, the predicted phase separation of the mid-frequency harmonic component, and the predicted steady-state error interval of the low-frequency base component are first normalized. Normalization maps data with different dimensions and value ranges to a unified interval, such as [0,1], for comparison and comprehensive calculation. Then, based on the importance of each frequency band in the entire test, different weights are assigned to these three types of data for weighted aggregation. For example, in tests simulating aerospace structural components, if the high-frequency pulse component has a significant impact on the fatigue life of the structural component, then a larger weight can be assigned to the predicted value of the timing offset of the high-frequency pulse component; if the low-frequency base component mainly affects the overall stability of the structural component, then an appropriate weight can be assigned to its steady-state error prediction interval according to the focus of the test.

[0157] Assume the weight of the high-frequency pulse component is w1, the weight of the mid-frequency harmonic component is w2, and the weight of the low-frequency base component is w3, and w1 + w2 + w3 = 1. Through weighted aggregation, these three types of data are fused to form a unified probability distribution surface, i.e., an error probability cloud map. The specific calculation process is as follows: For each channel, the normalized predicted time offset value is multiplied by w1, the value of each frequency point in the predicted phase separation quantity is multiplied by w2, and the normalized value of the steady-state error prediction interval is multiplied by w3. Then, these three results are added together to obtain the comprehensive error value of each channel at different frequency points or different time points. These comprehensive error values ​​constitute the error probability cloud map.

[0158] Error probability cloud plots intuitively characterize the dynamic adaptation level of each master and slave channel to the current load spectrum. For example, in the error probability cloud plot, if a channel has a low error probability in the high-frequency, mid-frequency, and low-frequency regions, it indicates that the channel has a high level of dynamic adaptation to the current load spectrum and can track the target load spectrum well. Conversely, if a channel has a high error probability in one or more frequency bands, it indicates that the channel's tracking performance in the corresponding frequency band is poor. Error probability cloud plots quantify the uncertainty risk of achieving high-precision tracking control, providing a decision-making basis for subsequent dynamic adjustment of frequency weights. For example, if the error probability cloud plot shows that a channel has a high error probability in the high-frequency band, then when adjusting the cooperative frequency weight matrix, it is possible to consider reducing the weight of that channel in the high-frequency band and increasing the weight of other channels with better tracking performance, thereby optimizing the tracking performance of the entire system, improving the time synchronization accuracy and phase consistency of the load spectrum of each channel in the multi-channel dynamic compression-shear test, and making the test results more realistically reflect the dynamic response of complex structures or systems, thus solving the key problems of comprehensive evaluation and optimization adjustment in multi-channel synchronous control.

[0159] Step S2500: Based on the error probability cloud map, dynamically adjust the weight distribution of the cooperative frequency weight matrix to obtain the optimized cooperative frequency weight matrix.

[0160] In multi-channel dynamic compression-shear tests, due to the differences in characteristics among the channels, even after the preliminary processing of the cooperative frequency weighting matrix in the preceding steps, it is still difficult to guarantee that each channel will achieve ideal synchronization when loading loads at different frequency bands. Step S2500 aims to dynamically adjust the weight allocation of the cooperative frequency weighting matrix based on the tracking performance information of each channel at different frequency bands provided by the error probability cloud map, thereby obtaining an optimized cooperative frequency weighting matrix. This further improves the time synchronization accuracy and phase consistency of the load spectrum of each channel, making the test results more accurately reflect the actual stress state of complex structures or systems.

[0161] Further, step S2500 includes:

[0162] Step S2510: Based on the predicted value A1 of the timing offset of the high-frequency pulse components of each master and slave channel in the error probability cloud map, optimize the high-frequency band weight allocation.

[0163] Set the timing offset prediction threshold to T, and linearly tilt the weight of the high-frequency band towards the channel where the timing offset prediction value A1 is less than T. The tilt ratio is proportional to (T-A1) / T.

[0164] Specifically, the predicted time offset value A1 quantifies the dynamic hysteresis risk of each channel when reproducing high-frequency pulse loads. It is obtained in step S2410 by correlating the temporal phase offset margin in the channel dynamic fingerprint with the characteristic time scale parameters of the high-frequency pulse. The predicted time offset threshold T can be set according to the loading response synchronization accuracy requirements, preferably between 0.2 and 0.4, and exemplarily 0.3. A nonlinear programming algorithm is used to optimize the high-frequency band weight allocation. The principle is to find the optimal solution through mathematical methods under certain constraints, that is, to find the weight allocation scheme that optimizes the high-frequency band loading effect. The weight of the high-frequency band is linearly tilted towards channels with a predicted time offset value A1 less than T, and the tilt ratio is proportional to (T-A1) / T. This is because channels with a predicted time offset value A1 less than T have a relatively low dynamic hysteresis risk when reproducing high-frequency pulse loads, that is, they perform better in terms of time synchronization. For example, suppose there is one master channel and two slave channels. The A1 value of slave channel 1 is 0.2, and the A1 value of slave channel 2 is 0.5. Let T be 0.4. According to the above rules, the tilt ratio of slave channel 1 is (0.4-0.2) / 0.4 = 0.5, and the tilt ratio of slave channel 2 is (0.4-0.5) / 0.4 = -0.25 (the negative sign indicates the direction of weight reduction). Thus, the weight of slave channel 1 will relatively increase in the high-frequency band, while the weight of slave channel 2 will relatively decrease.

[0165] The purpose of this operation is to optimize the weight allocation of the high-frequency band, making the loading of high-frequency pulse components in each channel closer to the ideal synchronization state. Its beneficial effect is that by allocating more weight to channels with better high-frequency response performance, it can effectively reduce the asynchrony problem caused by time delays in some channels during high-frequency loading. This helps improve the loading accuracy of the entire system for high-frequency pulse loads, making the load on the tested object under high-frequency vibration simulation closer to the actual working condition, thereby improving the accuracy and reliability of the test results and solving the key problem of synchronous loading of high-frequency loads in multi-channel dynamic compression-shear tests.

[0166] Step S2520: Based on the phase separation amount prediction spectrum of the intermediate frequency harmonic components of each channel in the error probability cloud map, optimize the weight allocation of the intermediate frequency band.

[0167] Set the phase separation amount prediction threshold as P', perform spectral clustering on the phase separation amount prediction spectrum, group the weights of the mid-frequency band according to the clustering results, and allocate the weights to the channel group whose phase separation amount prediction value A2 is less than P' at the cluster center. The allocation ratio is proportional to (P'-A2) / P'.

[0168] Specifically, a phase segregation prediction threshold P' is set. This threshold P' can be determined by statistically analyzing typical phase deviation data from multiple master-slave hydraulic servo channels during mid-frequency harmonic load tests, preferably between 0.1 and 0.3, with 0.2 being an example. Spectral clustering is performed on the phase segregation prediction spectrum. Spectral clustering is a graph-based clustering method that treats data points as nodes in a graph, constructs the graph by calculating the similarity between nodes, and then performs clustering based on the graph's structure. In this embodiment, spectral clustering is performed on the phase segregation prediction spectrum based on Euclidean distance. Euclidean distance measures the distance between two data points in the feature space; the closer the distance, the higher the similarity between the two data points. Through spectral clustering, the weights of the mid-frequency band are grouped according to the clustering results. The weights are then allocated to the channel group whose phase segregation prediction value A2 is less than P' at the cluster center, with the allocation ratio proportional to (P'-A2) / P'. For example, suppose there are three channels that are divided into two groups after spectral clustering. One group has an A2 value of 0.3, and the other has an A2 value of 0.6. Let P' be 0.5. Then the distribution ratio of the channel group with an A2 value of 0.3 is (0.5-0.3) / 0.5=0.4, and the distribution ratio of the channel group with an A2 value of 0.6 is (0.5-0.6) / 0.5=-0.2 (the negative sign indicates the direction of weight reduction). That is, the weight will be tilted towards the channel group with an A2 value of 0.3.

[0169] The purpose of this approach is to improve the phase consistency of each channel under mid-frequency harmonic loading by adjusting the weight allocation of the mid-frequency band. The beneficial effect is that it reduces mid-frequency harmonic loading deviations caused by phase separation in some channels, enabling each channel to load mid-frequency harmonic loads more coordinatedly. In the dynamic performance testing of multi-degree-of-freedom robot joints, this means that the stress conditions of the joint during actual movement can be simulated more accurately, improving the reliability of test results, providing more precise data support for evaluating robot joint performance, and solving the phase synchronization problem under mid-frequency harmonic loading in multi-channel dynamic compression-shear tests.

[0170] Step S2530: Based on the steady-state error prediction interval of the low-frequency basis components of each master and slave channel in the error probability cloud map, optimize the weight allocation of the low-frequency band.

[0171] Define the upper limit of the steady-state error prediction interval as Eu, the lower limit as El, the midpoint of the interval as Ec=(Eu+El) / 2, and the radius of the interval as Er=(Eu-El) / 2;

[0172] For the steady-state error prediction interval of the i-th channel [E] min,i E max,i ], where E min,i E represents the minimum value of the steady-state error prediction interval for the i-th channel. max,iThe maximum value of the steady-state error prediction interval for the i-th channel: i is the index of each master and slave channel;

[0173] If E max,i If Ec ≤ Ec, then the i-th channel is a low steady-state error channel. A reward-based low-frequency weight increment is applied to the i-th channel, where the low-frequency weight increment is proportional to (Ec - Ec). max,i It is directly proportional to Er;

[0174] If E min,i If the value is greater than or equal to Ec, then the i-th channel is a high steady-state error channel. A punitive low-frequency weight reduction is applied to the i-th channel, where the low-frequency weight reduction is proportional to (Ec). min,i -Ec) / Er is directly proportional;

[0175] If E min,i <Ec<E max,i， Then the i-th channel is the medium steady-state error channel, and the low-frequency weight of the i-th channel remains unchanged.

[0176] Specifically, in multi-channel dynamic compression-shear tests, the low-frequency base component typically simulates the slow deformation or low-frequency vibration of the tested object over a longer timescale, such as the overall swaying of a large building model under the action of low-frequency seismic waves. Due to differences in the characteristics of the hydraulic systems of each channel, variations in frictional resistance, and the independence of the control algorithms, the accuracy of each channel in reproducing the low-frequency base load varies. Step S2530 aims to optimize the weight allocation of the low-frequency band based on the steady-state error prediction interval of the low-frequency base components of each master and slave channel in the error probability cloud diagram, thereby improving the accuracy of each channel in low-frequency loading and making the test results more accurately reflect the actual response of the tested object under low-frequency load.

[0177] When the maximum value of the steady-state error prediction interval of a certain channel is E max,i When the error is less than or equal to the midpoint Ec of the interval, the channel is considered a low steady-state error channel. This means that when reproducing low-frequency base loads, the error of this channel remains at a relatively low level, and the tracking accuracy of low-frequency base loads is high. For example, in simulating a seismic test of a large bridge, suppose a channel is responsible for simulating the stress on the bottom of the pier under the action of low-frequency seismic waves, and its steady-state error prediction interval is [0.05, 0.15], while the calculated Ec is 0.2. In this case, the channel's Ec... max,i =0.15≤0.2, belonging to the low steady-state error channel. For this type of channel, a rewarding low-frequency weight increment is given, and the low-frequency weight increment is proportional to (Ec-E). max,iThe weight is proportional to Er. This is because this channel performs well in low-frequency loading, and increasing its weight allows it to undertake more loading tasks in the low-frequency band, thus better leveraging its high-precision tracking advantage. For example, if Er = 0.1, according to the formula, the low-frequency weight increment of this channel is (0.2 - 0.15) / 0.1 = 0.5. By increasing the weight, in subsequent experiments, the channel's contribution to the low-frequency base load increases, helping to more accurately simulate the actual stress at the bottom of the bridge pier, thereby improving the accuracy of the overall experiment in simulating the response of large bridges under low-frequency seismic waves.

[0178] If the minimum value of the steady-state error prediction interval of a certain channel is E min,i If the steady-state error is greater than or equal to the midpoint Ec of the interval, the channel is considered a high steady-state error channel. This indicates that the channel has a relatively large error and poor tracking accuracy when reproducing low-frequency base loads. For example, in a test simulating aerospace structural components under low-frequency vibration, the steady-state error prediction interval of a certain channel is [0.3, 0.4], and Ec is 0.25. In this case, the channel's Ec... min,i =0.3≥0.25, belonging to the high steady-state error channel. For this type of channel, a punitive reduction in low-frequency weights is applied, the reduction in low-frequency weights being proportional to (E... min,i The weight of this channel is proportional to (Ec) / Er. This is to reduce the loading ratio of this channel in the low-frequency range and reduce its negative impact on the overall test results. Assuming Er=0.1, according to the formula, the reduction in low-frequency weight of this channel is (0.3-0.25) / 0.1=0.5. By reducing the weight, the effect of this channel in the low-frequency range is weakened in subsequent tests, avoiding its large error from causing excessive interference to the overall test results, and ensuring that the test results can more accurately reflect the real situation of aerospace structural components under low-frequency vibration.

[0179] When the minimum steady-state error prediction interval of a certain channel is less than Ec and the maximum value is greater than Ec, the channel is considered a medium steady-state error channel. This indicates that the channel's error is at a moderate level when reproducing low-frequency base loads, and its tracking accuracy is generally average. For example, in an experiment simulating the low-frequency motion of a multi-degree-of-freedom robot joint, the steady-state error prediction interval of a certain channel is [0.1, 0.3], and Ec is 0.2. The Ec value of this channel... min,i =0.1<0.2, and E max,i =0.3 > 0.2, belonging to the medium steady-state error channel. For this type of channel, its low-frequency weight remains unchanged. This is because the error level of this channel is in the middle range, and there is no need to adjust its weight for the time being, so as to maintain the overall weight distribution balance, ensure that the loading of each channel in the low-frequency range is relatively stable during the experiment, and enable the experimental results to reflect the force state of the multi-degree-of-freedom robot joints during low-frequency motion more stably.

[0180] The fuzzy inference algorithm based on the steady-state error prediction interval is used to optimize the weight allocation in the low-frequency band. It can reasonably adjust the weight according to the actual tracking accuracy of each channel in the low-frequency band, improve the loading accuracy of multi-channel dynamic compression and shear test in the low-frequency band, and solve the problem of inaccurate test results caused by the difference in accuracy of each channel when loading low-frequency base load. This makes the test results more realistically reflect the actual working conditions of the tested object under low-frequency load.

[0181] Step S2540: Combine the optimized weight allocation results of high frequency band, mid frequency band and low frequency band to generate the optimized collaborative frequency weight matrix.

[0182] Specifically, the optimized weight allocation results from different frequency bands are integrated to construct a new matrix, namely the optimized cooperative frequency weight matrix. The elements of this matrix represent the weight ratios of the master and slave channels in different frequency bands, adjusted after considering the tracking performance of each channel in different frequency bands. The purpose of generating the optimized cooperative frequency weight matrix is ​​to provide a more reasonable and accurate basis for weight allocation in subsequent cooperative tracking control of multi-channel load spectra. Its beneficial effect is that the optimized cooperative frequency weight matrix can fully utilize the information provided by the error probability cloud map to proactively guide the cooperative tracking control of multi-channel load spectra. In actual experiments, each channel is loaded according to this optimized matrix, which can significantly improve the time synchronization accuracy and phase consistency of the load spectra of each channel in multi-channel dynamic compression-shear tests, enabling the test load to more accurately reproduce the actual working conditions, effectively improving the performance of the entire test system, and providing strong support for accurately studying the response of complex structures or systems under multi-channel dynamic compression-shear action.

[0183] In step S3000, the optimized collaborative frequency weight matrix is ​​input into the hydraulic servo controller, and the spectrum tracking output of multiple hydraulic servo channels is controlled in real time based on the master-slave synchronization and cross-coupling control strategy.

[0184] After the analysis of the target load spectrum and optimization of the cooperative frequency weight matrix are completed in the multi-channel dynamic compression-shear test, the optimized cooperative frequency weight matrix needs to be input into the hydraulic servo controller. Through the master-slave synchronization and cross-coupling control strategy, the spectrum tracking output of multiple hydraulic servo channels is controlled in real time to achieve a high degree of consistency of the load spectrum of each channel in terms of time, amplitude and phase, so that the test load can accurately reproduce the actual working condition.

[0185] Further, step S3000 includes:

[0186] Step S3100: Using the real-time load signal of the main channel as the synchronization reference, the optimized cooperative frequency weight matrix is ​​input to the hydraulic servo controllers corresponding to each main and slave channel.

[0187] Specifically, this step uses the real-time load signal of the main channel as the synchronization benchmark and inputs the optimized cooperative frequency weighting matrix into the hydraulic servo controllers corresponding to each master and slave channel. In multi-channel dynamic compression-shear tests, the main channel plays a crucial role, providing a benchmark reference for the entire system. Taking a simulated bridge seismic performance test as an example, the main channel may simulate the stress on key parts of the bridge (such as the bottom of the piers) under seismic waves, and its load spectrum tracking target is the standard for loading the entire test. Inputting the optimized cooperative frequency weighting matrix into the hydraulic servo controllers corresponding to each master and slave channel allows each channel to precisely control its own loading based on the weights determined by this matrix. The weight information in the cooperative frequency weighting matrix is ​​optimized after several steps, comprehensively considering factors such as the characteristics of each channel, potential distortion, and load characteristics of different frequency bands. By inputting it into the hydraulic servo controllers, each channel can accurately adjust its loading signal according to its weight ratio in different frequency bands, ensuring that each channel remains synchronized with the main channel during loading. This step is the key starting point for achieving multi-channel coordinated control, laying the foundation for the precise loading of each channel in the future. It ensures that the entire test system works under a unified benchmark, which helps to improve the time synchronization accuracy and phase consistency of the load spectrum of each channel, and makes the test results more accurately reflect the actual stress state of the bridge in the earthquake.

[0188] In step S3200, the hydraulic servo controller of the main channel directly uses the weights of each frequency band of the main channel in the optimized cooperative frequency weight matrix as the spectrum synthesis coefficients to synthesize the hydraulic control command of the main channel and control the opening output of the hydraulic valve of the main channel.

[0189] Specifically, the hydraulic servo controller of the main channel directly uses the weights of each frequency band in the optimized cooperative frequency weight matrix as spectrum synthesis coefficients to synthesize the hydraulic control commands for the main channel, controlling the opening output of the main channel hydraulic valves. In the simulation of the dynamic performance test of a multi-degree-of-freedom robot joint, it is assumed that the main channel is responsible for simulating the loading conditions of the main force points of the joint. The weights of each frequency band in the optimized cooperative frequency weight matrix are determined based on the force characteristics of the joint at different motion frequencies and the comprehensive performance of each channel. For example, in the high-frequency band, if the main channel has good response performance to high-frequency vibrations, its weight in the high-frequency band is relatively high. The hydraulic servo controller uses these weights as spectrum synthesis coefficients and synthesizes signals from different frequency bands according to a certain algorithm to generate the hydraulic control commands for the main channel. This command controls the opening output of the main channel hydraulic valves, and the opening size determines the magnitude and frequency of the load applied to the main force points of the joint. In this way, the main channel can accurately output a load spectrum that meets the test requirements based on the optimized weights. Since the main channel serves as the benchmark of the entire system, its accurate load output is crucial to ensuring the synchronization of each channel. This not only ensures the accuracy of the main channel in simulating joint stress, but also provides a reliable reference for the secondary channels, which helps improve the accuracy of the entire multi-channel system in simulating joint dynamic performance, and makes the test results more realistically reflect the stress situation of the joint in actual movement.

[0190] In step S3300, the hydraulic servo controller of the slave channel introduces a cross-coupling control term during the spectrum synthesis process. The cross-coupling control term is obtained by adjusting the deviation of the real-time load signals of each master and slave channel by a PID controller. The cross-coupling control term is superimposed on the weighting coefficients of each frequency band of the slave channel to correct the spectrum synthesis result of the slave channel and finally generate the hydraulic control command of the slave channel to control the opening output of the hydraulic valve of the slave channel.

[0191] Specifically, the hydraulic servo controller of the slave channel introduces a cross-coupling control term during the spectrum synthesis process. This cross-coupling control term is obtained by adjusting the deviation of the real-time load signals from the master and slave channels using a PID controller. The cross-coupling control term is then superimposed on the weighting coefficients of each frequency band in the slave channel to correct the spectrum synthesis result, ultimately generating the hydraulic control command for the slave channel and controlling the opening of its hydraulic valves. In simulated aerospace structural component tests, the slave channel is responsible for simulating the stress conditions of other parts of the structure. Due to differences in the characteristics of the hydraulic systems of each channel, the slave channel may become asynchronous with the master channel when tracking its own target load spectrum independently. Introducing a cross-coupling control term effectively solves this problem. The deviation of the real-time load signals from the master and slave channels reflects the degree of difference between them. The PID controller is a commonly used feedback controller that adjusts the control quantity based on the error of the input signal (i.e., the deviation of the real-time load signals from the master and slave channels) through proportional (P), integral (I), and derivative (D) operations. The hydraulic servo controller of the slave channel adds the deviation (i.e., cross-coupling control term) adjusted by the PID controller to the weighting coefficients of each frequency band of the slave channel. This allows the slave channel to adjust its loading signal in real time based on the deviation from the master channel during spectrum synthesis. For example, if the loading signal of the slave channel lags behind the master channel at a certain moment, the PID controller calculates a suitable cross-coupling control term based on the deviation, increasing the weighting coefficient of the slave channel in the corresponding frequency band, thus accelerating the loading speed of the slave channel and correcting the spectrum synthesis result. The final generated hydraulic control command of the slave channel controls the opening of the hydraulic valve, ensuring that the loading of the slave channel is highly consistent with the master channel in terms of time, amplitude, and phase. This process, by adjusting the loading of the slave channel in real time, improves the coordination of the multi-channel system, effectively reduces the timing deviation and phase difference between the slave and master channels, and enables the test load to more accurately simulate the actual stress conditions of aerospace structural components in complex environments, improving the reliability and accuracy of the test results.

[0192] In step S3400, the coordinated opening output of the master and slave hydraulic valves is applied to the object under test via the hydraulic cylinder, and dynamic loads are applied synchronously at multiple points to complete the closed-loop tracking control of the load spectrum.

[0193] Specifically, the coordinated opening outputs of the master and slave hydraulic valves are applied to the test object via hydraulic cylinders, synchronously applying dynamic loads at multiple points to achieve closed-loop tracking control of the load spectrum. In simulated seismic performance tests of large building models, the master channel and multiple slave channels simulate the stress conditions of different parts of the building model (such as piers, bridge bodies, etc.). The master and slave hydraulic valves generate coordinated opening outputs according to their respective hydraulic control commands, and these opening changes control the hydraulic cylinders to apply dynamic loads to the building model. Because the load spectra of each channel have achieved a high degree of consistency in time, amplitude, and phase through the master-slave synchronization and cross-coupling control strategy in the previous steps, the dynamic loads applied at multiple points can accurately simulate the actual stress state of the building model during an earthquake. For example, under the action of different frequency components of seismic waves, the master channel simulates the main stress at the bottom of the pier, while the slave channels simulate the stress on other parts of the bridge body. The hydraulic valves of each channel work in concert to ensure that the dynamic loads on each part of the building model match the stress conditions in an actual earthquake. This closed-loop tracking control is the final execution link of the entire test system. It transforms the optimization and control results of the previous steps into actual loading effects. By monitoring and adjusting the load output of each channel in real time, it ensures accurate tracking of the load spectrum during the test. This effectively solves the problem of inaccurate test results caused by the asynchronous load spectrum of each channel in multi-channel dynamic compression-shear test. It enables the test results to truly reflect the dynamic response of large building models in earthquakes and provides a reliable basis for the seismic performance evaluation of building structures.

[0194] Example 2

[0195] This embodiment, based on Embodiment 1, provides a real-time load spectrum tracking and control system, such as... Figure 8 As shown, it includes:

[0196] Channel dynamic fingerprint construction module: Determine the master channel and slave channel of the hydraulic servo system in the dynamic compression and shear testing machine, collect the hysteresis response curves of the hydraulic valves of each master and slave channel and the potential distortion spectrum of the servo controller; Based on the hysteresis response curves of the hydraulic valves of each master and slave channel, construct the channel dynamic fingerprint of each master and slave channel, and store the channel dynamic fingerprint and potential distortion spectrum into the heterogeneous parameter library.

[0197] Frequency weight matrix construction module: Performs three-level decomposition of the target payload spectrum to obtain three frequency band components, constructs time-series association labels for each frequency band component; merges the channel dynamic fingerprints of the master and slave channels, and combines them with the three frequency band components to generate an initial collaborative frequency weight matrix;

[0198] Frequency weight matrix optimization module: Based on the potential distortion spectrum, channel dynamic fingerprint and time-series association labels of each frequency band component stored in the heterogeneous parameter library, the initial cooperative frequency weight matrix is ​​corrected and weighted to obtain the optimized cooperative frequency weight matrix.

[0199] Cooperative control module: The optimized cooperative frequency weight matrix is ​​input into the hydraulic servo controller. Based on the master-slave synchronization and cross-coupling control strategy, the spectrum tracking output of multiple hydraulic servo channels is controlled in real time.

[0200] In the channel dynamic fingerprint construction module, the construction of channel dynamic fingerprints for each master and slave channel includes:

[0201] Step S1210: Based on the hysteresis response curves of the master and slave hydraulic valves, the Hammerstein model is used to describe the nonlinear dynamic characteristics of the hydraulic valves. The model parameters are fitted by the least squares method to obtain the nonlinear dynamic model of the master and slave hydraulic valves.

[0202] Step S1220: For the nonlinear dynamic model of the hydraulic valves in each master and slave channel, extract the time-domain phase offset margin and frequency-domain gain attenuation coefficient to construct the channel dynamic fingerprint of each master and slave channel.

[0203] Step S1230: Store the channel dynamic fingerprints and potential distortion spectra of each master and slave channel into the heterogeneous parameter library.

[0204] In the frequency weight matrix construction module, the time-series association labels for constructing each frequency band component include:

[0205] Step S2110: Perform three-level decomposition on the target load spectrum to obtain three frequency band components, namely, high-frequency pulse component, mid-frequency harmonic component and low-frequency base component.

[0206] Step S2120: For the decomposed high-frequency pulse components, extract the timestamp sequence of the pulse peaks and construct high-frequency time series association tags;

[0207] Step S2130: For the intermediate frequency harmonic components, extract the phase reference points of each harmonic component and construct intermediate frequency timing association tags;

[0208] Step S2140: For the low-frequency base component, directly add an absolute timestamp as a low-frequency timing association tag;

[0209] Step S2150: Use the high-frequency timing association tag of the high-frequency pulse component, the intermediate-frequency timing association tag of the intermediate-frequency harmonic component, and the low-frequency timing association tag of the low-frequency base component as the timing association tag of each frequency band component.

[0210] In the frequency weight matrix optimization module, the step of correcting and weighting the initial cooperative frequency weight matrix to obtain an optimized cooperative frequency weight matrix includes: correcting the initial cooperative frequency weight matrix based on the potential distortion spectrum stored in the heterogeneous parameter library to generate a cooperative frequency weight matrix; generating an error probability cloud map based on the channel dynamic fingerprint in the heterogeneous parameter library and the constructed time-series association labels of each frequency band component; and dynamically adjusting the weight allocation of the cooperative frequency weight matrix according to the error probability cloud map to obtain the optimized cooperative frequency weight matrix.

[0211] The step of dynamically adjusting the weight allocation of the cooperative frequency weight matrix based on the error probability cloud map to obtain the optimized cooperative frequency weight matrix includes:

[0212] Step S2510: Based on the predicted value A1 of the timing offset of the high-frequency pulse components of each master and slave channel in the error probability cloud map, optimize the high-frequency band weight allocation.

[0213] Step S2520: Based on the phase separation amount prediction spectrum of the intermediate frequency harmonic components of each channel in the error probability cloud map, optimize the weight allocation of the intermediate frequency band.

[0214] Step S2530: Based on the steady-state error prediction interval of the low-frequency basis components of each master and slave channel in the error probability cloud map, optimize the weight allocation of the low-frequency band.

[0215] Step S2540: Combine the optimized weight allocation results of high frequency band, mid frequency band and low frequency band to generate the optimized collaborative frequency weight matrix.

[0216] The methods and systems of this application may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of this application are not limited to the order specifically described above, unless otherwise specifically stated.

[0217] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0218] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A real-time load spectrum tracking control method, characterized by, The method comprises: determining the master channel and the slave channel of the hydraulic servo system in the dynamic compression-shear testing machine, collecting the hysteresis response curves of the hydraulic valves in the master and slave channels and the potential distortion spectrum of the servo controller; based on the hysteresis response curves of the hydraulic valves in the master and slave channels, constructing the channel dynamic fingerprints of the master and slave channels, and storing the channel dynamic fingerprints and the potential distortion spectrum into the heterogeneous parameter library; The construction of the channel dynamic fingerprints of the master and slave channels comprises: based on the hysteresis response curves of the hydraulic valves in the master and slave channels, using the Hammerstein model to describe the nonlinear dynamic characteristics of the hydraulic valves, fitting the model parameters by the least square method to obtain the nonlinear dynamic model of the hydraulic valves in the master and slave channels; for the nonlinear dynamic model of the hydraulic valves in the master and slave channels, extracting the time domain phase shift margin and the frequency domain gain attenuation coefficient to construct the channel dynamic fingerprints of the master and slave channels; performing three-level decomposition on the target load spectrum to obtain three frequency band components, and constructing the time sequence correlation labels of the frequency band components; fusing the channel dynamic fingerprints of the master and slave channels, and combining the three frequency band components to generate an initial collaborative frequency weight matrix; based on the potential distortion spectrum, the channel dynamic fingerprints stored in the heterogeneous parameter library and the constructed time sequence correlation labels of the frequency band components, correcting and weighting the initial collaborative frequency weight matrix to obtain an optimized collaborative frequency weight matrix; the three frequency band components are high-frequency pulse components, medium-frequency harmonic components and low-frequency base components; The construction of the time sequence correlation labels of the frequency band components comprises: for the decomposed high-frequency pulse component, extracting the timestamp sequence of the pulse peak to construct a high-frequency time sequence correlation label; for the medium-frequency harmonic component, extracting the phase reference point of each harmonic component to construct a medium-frequency time sequence correlation label; for the low-frequency base component, directly adding an absolute timestamp as a low-frequency time sequence correlation label; the high-frequency time sequence correlation label of the high-frequency pulse component, the medium-frequency time sequence correlation label of the medium-frequency harmonic component and the low-frequency time sequence correlation label of the low-frequency base component are used as the time sequence correlation labels of the frequency band components; The generation of the initial collaborative frequency weight matrix comprises: based on the channel dynamic fingerprints of the master and slave channels, for the high-frequency pulse component, determining the initial weight proportion of the high-frequency band of the master and slave channels according to the time domain phase shift margin of each channel by using a fuzzy reasoning method; for the medium-frequency harmonic component, determining the initial weight proportion of the medium-frequency band of the master and slave channels according to the frequency domain gain attenuation coefficient of each channel by using a frequency weighting method; for the low-frequency base component, using an equal distribution strategy to evenly distribute the weight of the low-frequency band to the master and slave channels; synthesizing the weight distribution results of the high-frequency band, the medium-frequency band and the low-frequency band to construct the initial collaborative frequency weight matrix, and the matrix elements are the initial weight proportions of the master and slave channels in different frequency bands; inputting the optimized collaborative frequency weight matrix into the hydraulic servo controller, and based on the master-slave synchronization and cross-coupling control strategy, real-time collaborative controlling the frequency spectrum tracking output of multiple hydraulic servo channels.

2. The real-time load spectrum tracking control method of claim 1, wherein, The method for determining the master channel and the slave channel of the hydraulic servo system in the dynamic compression-shear testing machine comprises: designating one hydraulic servo channel in the dynamic compression-shear testing machine as the master channel, and the remaining hydraulic servo channels as the slave channels, the number of the slave channels being n1, n1≥1.

3. The real-time load spectrum tracking control method of claim 2, wherein, The modifying the initial cooperative frequency weight matrix and the weight distribution to obtain an optimized cooperative frequency weight matrix comprises: The initial cooperative frequency weight matrix is modified based on the potential distortion spectrum stored in the heterogeneous parameter library to generate the cooperative frequency weight matrix; Based on the channel dynamic fingerprint and the time sequence correlation label of each frequency band component in the heterogeneous parameter library, an error probability cloud map is generated; According to the error probability cloud map, the weight distribution of the cooperative frequency weight matrix is dynamically adjusted to obtain an optimized cooperative frequency weight matrix.

4. The real-time load spectrum tracking control method of claim 3, wherein, The modifying the initial cooperative frequency weight matrix to generate the cooperative frequency weight matrix comprises: Based on the potential distortion spectrum stored in the heterogeneous parameter library, the nonlinear distortion characteristics of the master and slave channels are extracted, including amplitude distortion and phase distortion; The amplitude distortion rate and the phase distortion rate of the master and slave channels are calculated; according to the amplitude distortion rate and the phase distortion rate, the total distortion rate of the master and slave channels is obtained; the channel with a total distortion rate greater than a preset distortion threshold is defined as a severe distortion channel; For high and medium frequency bands, the initial cooperative frequency weight matrix is modified according to the nonlinear distortion characteristics of the master and slave channels, and the weight proportion of the severe distortion channel is reduced; for low frequency bands, the initial cooperative frequency weight matrix is kept unchanged; The weight correction results of high, medium and low frequency bands are integrated to generate the final cooperative frequency weight matrix.

5. The real-time load spectrum tracking control method of claim 4, wherein, The generation of the error probability cloud map comprises: For high frequency pulse components, the time sequence offset degree prediction value of the high frequency pulse components of the master and slave channels is obtained according to the channel dynamic fingerprint and the high frequency time sequence correlation label; For medium frequency harmonic components, the phase segregation quantity prediction spectrum of the medium frequency harmonic components of the master and slave channels is obtained according to the channel dynamic fingerprint and the medium frequency time sequence correlation label; For low frequency base components, the steady state error prediction interval of the low frequency base components of the master and slave channels is obtained according to the channel dynamic fingerprint and the low frequency time sequence correlation label; The time sequence offset degree prediction value of the high frequency pulse components, the phase segregation quantity prediction spectrum of the medium frequency harmonic components, and the steady state error prediction interval of the low frequency base components are fused to generate the error probability cloud map of the overall tracking performance of the master and slave channels by using the membership weighted method.

6. The real-time load spectrum tracking control method of claim 5, wherein, The time sequence offset degree prediction value of the high frequency pulse components of the master and slave channels comprises: Based on the high frequency time sequence correlation label of the high frequency pulse components, the characteristic time scale parameters of the high frequency pulse are extracted, including the average pulse width and the pulse interval time; The time domain phase offset margin in the channel dynamic fingerprint is associated with the characteristic time scale parameters of the high frequency pulse to generate the time sequence offset degree prediction value of the high frequency pulse components of the master and slave channels.

7. The real-time load spectrum tracking control method of claim 6, wherein, The dynamic adjustment of the weight distribution of the cooperative frequency weight matrix according to the error probability cloud map to obtain an optimized cooperative frequency weight matrix comprises: Based on the time sequence offset degree prediction value A1 of the high frequency pulse components of the master and slave channels in the error probability cloud map, the high frequency band weight distribution is optimized; Based on the phase segregation quantity prediction spectrum of the medium frequency harmonic components of the master and slave channels in the error probability cloud map, the medium frequency band weight distribution is optimized; Based on the steady state error prediction interval of the low frequency base components of the master and slave channels in the error probability cloud map, the low frequency band weight distribution is optimized; The optimized weight distribution results of the high frequency band, the medium frequency band and the low frequency band are integrated to generate an optimized collaborative frequency weight matrix.

8. The real-time load spectrum tracking control method of claim 7, wherein, The low-frequency base component of the error probability cloud map is used to optimize the low-frequency weight distribution. Define the upper limit of the steady-state error pre-judgment interval as Eu, the lower limit as E l , the midpoint of the interval as Ec=(Eu+E l ) / 2, and the interval radius as Er=(Eu-E l ) / 2; The steady-state error prediction interval [E min,i ,E max,i ] of the i-th channel, wherein E min,i is the minimum value of the steady-state error prediction interval of the i-th channel, and E max,i is the maximum value of the steady-state error prediction interval of the i-th channel: i is the index of the master and slave channels; If E max,i ≤ Ec, the ith channel is a low steady-state error channel, and a rewarding low-frequency weight increment is given to the ith channel, which is proportional to (Ec-E max,i ) / Er. If E min,i ≥ Ec, the ith channel is a high steady-state error channel, and a penalizing low frequency weight decrement is given to the ith channel, said low frequency weight decrement being proportional to (E min,i -Ec) / Er. If E min,i <Ec<E max,i， The ith channel is a mesostable error channel, and the low frequency weight of the ith channel remains unchanged.

9. The real-time load spectrum tracking control method of claim 8, wherein, The master-slave synchronization and cross-coupling control strategy includes: The real-time load signal of the master channel is used as a synchronization reference, and the optimized collaborative frequency weight matrix is input into the hydraulic servo controller of each channel. The hydraulic servo controller of the master channel directly uses the weights of each frequency band in the optimized collaborative frequency weight matrix as the frequency spectrum synthesis coefficients to synthesize the hydraulic control command of the master channel and control the opening output of the master channel hydraulic valve. The hydraulic servo controller of the slave channel introduces a cross-coupling control term in the frequency spectrum synthesis process, which is obtained by adjusting the deviation of the real-time load signals of the master and slave channels through a PID controller. The cross-coupling control term is superimposed on the weight coefficients of each frequency band of the slave channel to correct the frequency spectrum synthesis result of the slave channel, and finally generate the hydraulic control command of the slave channel to control the opening output of the slave channel hydraulic valve.

10. A real-time load spectrum tracking control system for implementing the real-time load spectrum tracking control method of any one of claims 1-9, characterized by, The collaborative opening outputs of the master and slave channel hydraulic valves act on the measured object through the hydraulic cylinder to synchronously apply dynamic load at multiple points and complete the closed-loop tracking control of the load spectrum. The system includes: The channel dynamic fingerprint construction module determines the master and slave channels of the hydraulic servo system in the dynamic compression shear testing machine, collects the hysteresis response curves of the master and slave channel hydraulic valves and the potential distortion spectrum of the servo controller, constructs the channel dynamic fingerprints of the master and slave channels based on the hysteresis response curves of the master and slave channel hydraulic valves, and stores the channel dynamic fingerprints and the potential distortion spectrum in the heterogeneous parameter library. The frequency weight matrix construction module performs three-level decomposition on the target load spectrum to obtain three frequency band components, constructs the time sequence association labels of each frequency band component, fuses the channel dynamic fingerprints of the master and slave channels, and generates an initial collaborative frequency weight matrix in combination with the three frequency band components. The frequency weight matrix optimization module corrects and distributes the weights of the initial collaborative frequency weight matrix based on the potential distortion spectrum, the channel dynamic fingerprints and the constructed time sequence association labels of each frequency band component stored in the heterogeneous parameter library to obtain an optimized collaborative frequency weight matrix. The collaborative control module inputs the optimized collaborative frequency weight matrix into the hydraulic servo controller and controls the frequency spectrum tracking output of multiple hydraulic servo channels based on the master-slave synchronization and cross-coupling control strategy.

Images