Sensor pulse pressure fatigue test closed-loop control system based on signal drift compensation

Abstract

The invention relates to the technical field of online material performance test instruments, and discloses a sensor pulse pressure fatigue test closed-loop control system based on signal drift compensation, which comprises a load excitation execution module, a signal feedback acquisition module and a closed-loop control logic module, the closed-loop control logic module selects a phase-locked characteristic observation window by calculating a load first-order derivative, extracts a dynamic gain coefficient reflecting the response characteristic of a controlled test object in the observation window, and inputs the dynamic gain coefficient into a filter to generate a characteristic variable; the variable is compared with the standard response envelope line, the deviation is calculated, the compensation factor is calculated and superposed to the driving instruction, the real-time correction of the output gain is realized, and through the identification and active compensation of the signal drift, the sensing deviation caused by the structural degradation of the sensitive component is effectively eliminated, and the sensing consistency of the test process is maintained.

Description

Technical Field

[0001] This invention relates to a closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation, belonging to the technical field of online material performance testing instruments. Background Technology

[0002] Current sensor pulse pressure fatigue testing is used to evaluate the reliability of sensitive components. The test system simulates periodically fluctuating pressure loads and monitors the sensor's output level. Due to the influence of stress-strain transformation within the sensor material, the industry typically adopts a dynamic-static separation monitoring method, performing zero-point sampling during the intervals of pressure pulses to avoid interference from dynamic loads on the reference level. The above monitoring mode has a physical observation blind zone, and the transient drift characteristics generated by the sensor during sudden pressure changes cannot be captured, thus masking the signal disturbances caused by microscopic fatigue damage to the material. Since the sensor is accompanied by hysteresis and thermo-elastic coupling phenomena during fatigue failure, failure characteristics often erupt at the rising edge of the pulse during drastic pressure changes. Existing test systems lack means to extract effective signal components under dynamic conditions and usually use large time windows for smoothing, causing hysteresis in the monitoring signal on the time axis, thus masking the abnormal performance of the sensor under extreme stress conditions.

[0003] While improvements such as increasing the sampling frequency or the filter order increase the amount of raw data, the inability to physically decouple the signal from the dynamic load leads to phase delay and load noise interference. This lag mechanism, which infers dynamic performance based on static data, constitutes an inherent contradiction in improving the accuracy of closed-loop control. Not only are there blind spots at the physical level, but the control end also exhibits logical lag when dealing with perception deviations caused by performance degradation. For example, the utility model patent with authorization announcement number CN218781945U discloses a pressure sensor pulse fatigue testing device. By optimizing the hydraulic station, test chamber, pressure reducing valve, and electromagnetic reversing valve components, it achieves accurate loading and unloading control over a large pressure range. Although the hardware supports the stability of the pulse cycle, the control logic remains at the mechanical maintenance of the pressure at the output end of the actuator, lacking feedback on the dynamic response characteristics of the controlled sensor. Under long-term pulse impact, the sensitive component experiences sensitivity drift due to slight changes in elastic modulus or thermo-elastic coupling. Existing equipment cannot decouple the dynamic deviation components in real time and cannot correct the micro-gain of the drive command within a single pulse cycle, thus interfering with the consistency of strain energy during the test process.

[0004] Therefore, how to achieve real-time decoupling of the sensor's essential performance parameters during the dynamic process of pulsed pressure loading, and how to construct a closed-loop intervention mechanism that can respond instantly to material degradation, is the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation, comprising a load excitation execution module, a signal feedback acquisition module, and a closed-loop control logic module: The signal feedback acquisition module is connected to the output of the controlled test object; The closed-loop control logic module is connected to both the load excitation execution module and the signal feedback acquisition module. The closed-loop control logic module executes the following steps: Step S101: Calculate the first derivative of the load output by the load excitation execution module and compare it with a preset gradient threshold; select a phase-locked feature observation window at the rising edge of each cycle of the pulse pressure load. Step S102: Obtain the load feedback value and test feedback signal within the feature observation window, and construct a dynamic correlation matrix reflecting the response characteristics of the controlled test object based on the load feedback value, test feedback signal, and the differential relationship of the time axis. Step S103: Perform least squares fitting on the dynamic correlation matrix to extract the dynamic gain coefficient of the test feedback signal to the load change, and input the dynamic gain coefficient into the first-order hysteresis filter to obtain the characteristic variable characterizing the degree of stiffness degradation of the controlled test object structure. Step S104: Compare the feature variable with the standard response envelope stored in the initialization phase. In response to the state where the feature variable deviates from the standard response envelope and the volatility exceeds a preset threshold, calculate the deviation of the feature variable relative to the standard response envelope. Step S105: Calculate the compensation factor based on the deviation and the preset image table, and superimpose the compensation factor into the drive control command of the load excitation execution module. By correcting the output gain of the load excitation execution module, the perception consistency closed-loop control of the controlled test object during the fatigue test process is realized.

[0006] Preferably, the closed-loop control logic module further performs the following steps: Step S201: Extract sampling feature points using the high-speed sampling channel established by the feature observation window, and perform phase locking and logic reconstruction on the sampling feature points to separate the signal disturbance caused by the micro-fatigue damage of the controlled test object; Step S202: Monitor the amplitude and phase characteristics of the signal disturbance. When the signal median shift, peak attenuation, or waveform distortion exceeds the preset limit, the closed-loop control logic module generates an abnormal drift warning command and locks the current output state of the load excitation execution module to avoid transient failure of the controlled test object.

[0007] Preferably, the sampling time interval of the feature observation window is set to 10μs, and the closed-loop control logic module calculates the dynamic gain coefficient within the feature observation window. The calculation formula is: ,in, as well as These represent the test feedback signal levels at the end and beginning of the feature observation window, respectively. as well as These are the measured load feedback values ​​at the corresponding time points of the characteristic observation window.

[0008] Preferably, when step S105 is executed, the compensation factor is superimposed as a feedback gain coefficient onto the control voltage of the drive circuit to compensate for the sensitivity drift of the controlled test object caused by the change in the elastic modulus of the internal sensitive component.

[0009] Preferably, a first-order hysteresis filter is used to filter out random environmental noise in the dynamic gain coefficient, and the update step size of the characteristic variable is consistent with the pulse period of the pulse pressure load.

[0010] Preferably, the initialization phase includes: performing 100 to 500 pulse cycles of preset intensity before loading test to obtain steady-state distribution data of feature variables, and setting the physical boundary of the standard response envelope based on three times the standard deviation range of the steady-state distribution data.

[0011] Preferably, the load excitation execution module includes a pulse pressure source module and a load sensing module, wherein the load sensing module is used to provide the raw load signal required to calculate the first derivative of the rising edge.

[0012] Preferably, the logic reconstruction includes: performing mean denoising on the sampled feature points and constructing a time series model reflecting performance evolution; the closed-loop control logic module identifies the nonlinear degradation trajectory of the controlled test object based on the predicted residual of the time series model.

[0013] Preferably, the system also includes a data storage module, which is connected to the closed-loop control logic module, for recording the historical values ​​of the dynamic correlation matrix, feature variables, and compensation factors.

[0014] Preferably, the closed-loop control logic module introduces an aging factor in the control loop. The aging factor is calculated by the closed-loop control logic module based on the number of cumulative pulse cycles executed and is used to correct the drive control commands.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In sensor pulse pressure fatigue testing, a phase-locked dynamic characteristic observation mechanism is constructed. By utilizing the pressure gradient identification logic during the rising edge of the pulse pressure, an observation window is selected within the drastic pressure range of each pulse cycle. By acquiring the pressure feedback value and sensor response signal within the window, a dynamic correlation matrix is ​​constructed, enabling the system to extract the essential performance parameters of the sensor from the dynamic process of high load changes. This breaks the mutually exclusive state of dynamic loading and drift monitoring on the time axis in traditional testing, avoids the physical monitoring blind spot caused by the separation of dynamic and static processing logic, and improves the transient response capability of failure feature capture.

[0016] 2. By extracting the dynamic gain coefficient of the response signal to pressure change within the observation window and inputting it into a first-order hysteresis filter, the degree of stiffness degradation of the sensor structure can be identified in real time. When the characteristic variable deviates from the standard response envelope established in the initialization stage, the closed-loop control system triggers drift compensation logic based on the volatility, dynamically adjusting the peak hold time or frequency of subsequent pulses to ensure that the sensor is under consistent strain energy conditions throughout the entire test cycle. This elevates the sensor performance evaluation from a single fault trigger judgment to a trend evolution analysis based on process parameters, enhancing the scientificity and reliability of fatigue life assessment.

[0017] 3. By utilizing the controller's high-speed sampling channel and floating-point arithmetic capabilities, and through logical reconstruction and phase locking of the sampling feature points, real-time decoupling of electrical signal disturbances caused by microscopic fatigue damage in materials can be achieved without introducing additional physical sensors. This signal flow logic reconstruction based on the algorithm level can capture the transient abnormal performance of sensor-sensitive components under extreme stress conditions online, effectively avoiding the risk of test system damage due to transient failure, and ensuring the continuity and safety of the online material performance testing process. Attached Figure Description

[0018] Figure 1 This is a flowchart of the closed-loop control of the pulse pressure fatigue test with phase locking and dynamic gain correction according to the present invention. Figure 2 This is a diagram of the closed-loop feedback hardware architecture of the system integrating the real-time control host and the physical test bench in this invention. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. This section is intended to elaborate on the present invention in detail to explain and illustrate the invention, and is not intended to limit the scope of protection of the present invention.

[0020] This invention provides a closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation, comprising a load excitation execution module, a signal feedback acquisition module, and a closed-loop control logic module. The signal feedback acquisition module is connected to the output of the controlled test object. The closed-loop control logic module is connected to both the load excitation execution module and the signal feedback acquisition module. The load excitation execution module includes a pulse pressure source module and a load sensing module. The load sensing module provides the raw load signal and executes a reference alignment procedure during system initialization. When the pulse pressure source module is in a stopped state, the closed-loop control logic module continuously acquires 1024 load sampling points and calculates their average static voltage as the system zero point. The first derivative of the load is then calculated in real-time. The original sampled value is subtracted from the average static voltage to eliminate zero-point drift interference at the actuator sensing end, thus ensuring that the opening time of the feature observation window is precisely aligned with the starting point of the load rising edge. The closed-loop control logic module selects the phase-locked feature observation window by calculating the first derivative of the load, extracts the dynamic gain coefficient reflecting the response characteristics of the controlled test object within the observation window, and generates feature variables. By comparing these variables with the standard response envelope and calculating the deviation, a compensation factor is calculated and superimposed on the drive command to achieve real-time correction of the output gain. During the fatigue failure process of the sensor, hysteresis and thermoelastic coupling phenomena occur. Failure characteristics often erupt at the rising edge of the pulse during periods of drastic pressure changes. To capture effective signal components under dynamic conditions, the closed-loop control logic module executes step S101, calculating the first derivative of the load output by the load excitation execution module. It is then compared with a preset gradient threshold, and a phase-locked feature observation window is selected at the rising edge of each cycle of the pulse pressure load.

[0021] Due to the influence of stress-strain transformation within the sensor material, transient drift signals under dynamic conditions are coupled with load noise. The closed-loop control logic module executes steps S102 to S103 to acquire the load feedback value and test feedback signal within the characteristic observation window. Based on the differential relationship between the load feedback value, the test feedback signal, and the time axis, a dynamic correlation matrix reflecting the response characteristics of the controlled test object is constructed. The closed-loop control logic module performs least-squares fitting on this dynamic correlation matrix to extract the dynamic gain coefficient of the test feedback signal to the load change. The sampling time interval of the feature observation window is set to Dynamic gain coefficient The calculation formula is as follows: ,in, This is the dynamic gain coefficient. The test feedback signal level value at the end of the feature observation window. The test feedback signal level value at the start of the feature observation window. The measured load feedback value corresponds to the end time of the feature observation window. The closed-loop control logic module uses the measured load feedback value corresponding to the start time of the characteristic observation window as the dynamic gain coefficient. The input is fed into a first-order hysteresis filter to obtain feature variables characterizing the degree of stiffness degradation of the controlled test object's structure. The first-order hysteresis filter is used to filter out random environmental noise, and the update step size of the feature variables is consistent with the pulse period of the pulsed pressure load. The closed-loop control logic module constructs a dynamic correlation matrix within the feature observation window. This matrix is ​​generated by real-time data acquisition. Each load feedback value With corresponding test feedback signal Composition, here The sampling point index is 1 to 1. The closed-loop control logic module extracts the dynamic gain coefficients by least-squares fitting of the dynamic correlation matrix. By constructing a system that includes load feedback values Design matrix and including test feedback signals The observed vectors are used to solve the slope term of the linear regression model using the normal equation method, thereby obtaining the dynamic gain coefficients that characterize the sensitivity of the controlled test object. The fitting process uses all sampled data within the feature observation window to reduce the interference of high-frequency pressure pulsation on signal acquisition, and eliminates the random error caused by calculating the slope using only the data at the start and end points of the window, so that the generated feature variables reflect the structural stiffness degradation state of the controlled test object along the pressure rise phase.

[0022] To address sensitivity drift caused by changes in the elastic modulus of sensitive components, the closed-loop control logic module executes steps S104 to S105. During the initialization phase, the system executes 100 to 500 pulse cycles of preset intensity to acquire steady-state distribution data of the characteristic variable. The physical boundary of the standard response envelope is set based on a range of three times the standard deviation of the steady-state distribution data. The closed-loop control logic module compares the characteristic variable with the standard response envelope. When the characteristic variable deviates from the standard response envelope and its volatility exceeds a preset threshold, the deviation of the characteristic variable relative to the standard response envelope is calculated. The closed-loop control logic module calculates a compensation factor based on the deviation and a preset mapping table, and adds it as a feedback gain coefficient to the control voltage of the drive loop. By correcting the output gain of the load excitation execution module, the controlled test object maintains consistent perception during fatigue testing. The physical boundary of the standard response envelope is based on the overall average value acquired during the initialization phase. vs. population standard deviation Confirmed, here The arithmetic mean of the characteristic variables in the first 500 cycles of preset intensity pulses. To correspond to the standard deviation, the physical boundary is It covers the distribution region of characteristic variables under steady-state conditions. The preset mapping table is established by offline calibration. By recording the driving voltage correction value required to make the output feedback of the controlled test object return to the center position of the standard response envelope under different stress bias states, the deviation and compensation factor are constructed. Interlinear mapping model, compensation factor , here To preset the dimensionless proportional gain coefficient, These are the measured values ​​of the characteristic variables for the current cycle. To initialize the baseline value and adapt the control command correction logic to the physical degradation rate of the controlled test object, the closed-loop control logic module also executes steps S201 to S202 to address the abnormal performance of the sensor under extreme stress conditions. This involves extracting sampling feature points using a high-speed sampling channel established by the feature observation window, performing phase locking and logic reconstruction on the sampling feature points, separating the signal disturbance caused by microscopic fatigue damage of the controlled test object, and performing mean denoising on the sampling feature points and constructing a time series model reflecting performance evolution. The closed-loop control logic module identifies the nonlinear degradation trajectory of the controlled test object based on the prediction residual of the time series model. It also monitors the amplitude and phase characteristics of the signal disturbance. When the signal median shift, peak attenuation, or waveform distortion exceeds a preset limit, an abnormal drift warning command is generated, and the current output state of the load excitation execution module is locked.

[0023] The closed-loop control logic module uses logic reconstruction to separate signal disturbances, identifies them through time series based on predicted residuals, and utilizes the dynamic gain coefficients stored in the preceding 20 pulse cycles. The mean is used as a prediction operator, and the predicted response level is calculated by combining it with the current period load feedback value. Specifically, the execution logic of this time series model is as follows: The closed-loop control logic module allocates a dynamic gain coefficient circular sliding buffer with a capacity of 20 groups in the RAM memory. In the pressure plateau segment of each pulse period, the arithmetic mean of all values ​​in the current buffer is extracted as the reference sensitivity prediction factor; the measured load feedback value collected in the current feature observation window is multiplied with the reference sensitivity prediction factor to generate the predicted response level sequence; the synchronously collected test feedback signal is then used to calculate the predicted response level. The measured level is subtracted from the predicted response level, and the absolute value of the difference is stored in the residual buffer. The closed-loop control logic module calls a 10th-order moving average filter operator to smooth the residual buffer, removing background random noise with a root mean square value less than 5.0 mV. The final output is the signal disturbance component caused by microscopic fatigue damage inside the controlled test object. The predicted response level is subtracted from the measured level of the real-time acquired test feedback signal to generate a residual sequence. The closed-loop control logic module averages and filters the residual sequence to extract the signal disturbance component caused by microscopic fatigue damage. When the mean of this component is continuous... When a pulse cycle exceeds 150% of the physical boundary of the standard response envelope, the controlled test object is determined to have entered the nonlinear degradation stage and an early warning command is triggered. By deconstructing the time-domain characteristics of the pressure change phase data, the inherent performance degradation characteristics of the sensor are identified online without introducing additional hardware sensors. To eliminate the logical coupling ambiguity of multiple positive terms during the drive command reconstruction process, the closed-loop control logic module executes the command integration procedure, which calculates the dimensionless compensation factor based on the deviation of the characteristic variables. And the aging factor retrieved based on the cumulative pulse cycle count. Unified mapping to the gain correction operator, driving voltage The calculation formula is set as follows: ,in, This is the control voltage command output to the drive circuit, in volts (V). The closed-loop control logic module determines the reference drive voltage based on the current test task. As a compensation factor, As an aging factor, by integrating transient sensitivity compensation and long-term structural aging compensation into a unified feedback voltage command in a linear superposition manner, the control system realizes closed-loop intervention of the drift of the sensing characteristics of the controlled test object.

[0024] Example 1: In an application scenario involving fatigue life assessment of aviation hydraulic sensors, the system operates at a pulse frequency of 10Hz and an ambient temperature of... to Under cyclically fluctuating operating conditions, the controlled test object faces the dual effects of thermoelastic drift and mechanical fatigue degradation. Furthermore, the sensitivity drift characteristics of sensitive components are masked by broadband load noise. The load excitation execution module applies periodic pulsed pressure loads to the controlled test object, and the closed-loop control logic module extracts the first derivative of the load in real time. , in response to the first derivative When the pressure exceeds a preset gradient threshold, the characteristic observation window is locked within the pressure rise edge interval of each pulse cycle, anchoring the data observation time to the abrupt phase change of pressure. The signal feedback acquisition module then uses this characteristic observation window to... The sampling time interval synchronously acquires the load feedback value. With test feedback signal The closed-loop control logic module uses the data within this observation window to construct a dynamic correlation matrix, and extracts the dynamic gain coefficients reflecting the transient response characteristics of the sensor through least-squares fitting. The dynamic gain coefficient The calculation formula is as follows: ,in, This is the dynamic gain coefficient. The test feedback signal level value at the end of the feature observation window. The test feedback signal level value at the start of the feature observation window. The measured load feedback value corresponds to the end time of the feature observation window. The closed-loop control logic module uses the measured load feedback value corresponding to the start time of the characteristic observation window to extract the dynamic gain coefficient within the characteristic observation window. A first-order hysteresis filter is input to filter out random environmental noise and generate characteristic variables, which characterize the evolution state of the internal elastic modulus of the controlled test object.

[0025] When a characteristic variable deviates from the standard response envelope due to the degradation of the sensitive structure of the controlled test object, the closed-loop control logic module calculates the deviation of the characteristic variable relative to the physical boundary of the standard response envelope, calculates a compensation factor based on the deviation and a preset mapping table, and adds it as a feedback gain to the control voltage command of the drive loop. The phase-locked characteristic observation window is the dynamic gain coefficient. The calculation provides data samples, and the compensation factor corrects the output gain of the load excitation execution module to offset the performance deviation caused by environmental temperature fluctuations and material degradation in real time. The output response of the controlled test object is maintained within the standard response envelope throughout the entire test cycle. The system converts the signal drift under dynamic conditions into quantifiable compensation variables, which solves the problem of difficulty in identifying the inherent performance degradation characteristics under dynamic loads, and keeps the strain energy input and response output consistent during the fatigue test process.

[0026] Example 2: In a test scenario where a piezoresistive pressure sensor is subjected to continuous loading using a high-frequency hydraulic pulse fatigue testing bench, the controlled test object is in an industrial environment with an electromagnetic pulse interference intensity of 20dB and an ambient temperature fluctuating at a rate of 1.5℃ / min. Because the amplitude of electromagnetic noise and thermal drift signals reaches 3% to 5% of the sensor's full-scale output during load intervals, existing sampling methods cannot distinguish between sensitivity attenuation caused by fatigue microcracks in the sensitive layer and spurious drift signals introduced by external environmental fluctuations. The test platform used for this verification includes a maximum output pressure of 100MPa, and... A hydraulic pulse generator with force control accuracy better than 0.05%FS and a synchronous signal feedback acquisition module with 1MHz analog bandwidth and 16-bit resolution are used. To balance the real-time performance of data acquisition with the computational load of the closed-loop control logic module, the sampling time interval of the characteristic observation window is set to 10μs. This parameter is selected based on the fact that the pressure response bandwidth of the controlled test object during the rising edge phase is distributed in the range of 5kHz to 20kHz. Setting the sampling frequency to 100kHz can meet the physical constraints of the Nyquist sampling theorem on signal reconstruction and avoid the dynamic gain coefficient caused by sparse sampling points. To address fitting distortion, the experiment included a control group without drift compensation, a comparative group with global mean compensation, and an experimental group using the phase-locked compensation method of this invention. Researchers actively superimposed Gaussian white noise with a root mean square value of 50mV at the input of the signal feedback acquisition module to simulate an engineering noise environment. The results were monitored by measuring the cumulative execution... The drive gain correction and sensitivity deviation after the next pulse cycle are shown in Table 1 below. The table records the recognition accuracy and compensation effect of signal drift for each group under different sampling conditions and interference intensities.

[0027] Table 1: Comparison of compensation effects under different experimental conditions

[0028] As shown in Table 1, when faced with noise and temperature drift interference of the same magnitude, the control group without phase locking, due to its feature extraction across the entire time domain, exhibits a higher dynamic gain coefficient. The calculation results are affected by the nonlinear coupling of zero-point drift during the load interval, causing the sensitivity deviation rate to rise to 8.45%. However, the experimental group of this invention concentrates the calculation data points on the pressure gradient through the feature observation window selected in step S101. The rapidly changing phase interval utilizes the physical monotonicity of the rising edge slope to achieve spatial separation between the effective signal and steady-state noise, keeping the sensitivity deviation rate stable within 0.22%. However, when the sampling time interval increases from 10 μs to boundary states exceeding 50 μs, the number of sample points within the characteristic observation window is insufficient to support the statistical convergence of least-squares fitting, resulting in a decrease in the dynamic gain coefficient. The nonlinear degradation trend confirms that the sampling accuracy range specified in the manual is the physical boundary to ensure the stable operation of the closed-loop intervention mechanism. The system converts the complex drift characteristics under dynamic conditions into a set of online-updated dynamic gain coefficients through phase-locked characteristic observation. The compensation factor reconstructs the command intensity of the load excitation execution module in real time according to the evolution trajectory of the coefficient, which solves the problem that the performance degradation characteristics under dynamic load are masked by the physical observation blind zone and load noise, and enables the intelligent sensing system to control the sensing error within 0.5% throughout the entire fatigue cycle.

[0029] Example 3: This example combines Figures 1 to 2 The following describes a closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation. Figure 1As shown, in step S101, the first derivative of the load output by the load excitation execution module is calculated and compared with a preset gradient threshold. A phase-locked feature observation window is selected at the rising edge of each cycle of the pulse pressure load. The process proceeds to step S102 to obtain the load feedback value and test feedback signal within the feature observation window. Based on the differential relationship between the two and the time axis, a dynamic correlation matrix reflecting the response characteristics of the controlled test object is constructed. Then, in step S103, least squares fitting is performed on the dynamic correlation matrix to extract the dynamic gain coefficient, which is then input into a first-order hysteresis filter to obtain the feature variable characterizing the degree of structural stiffness degradation of the controlled test object. In step S104, the feature variable is compared with the standard response envelope stored in the initialization stage. In response to the feature variable deviating from the standard response envelope and the volatility exceeding a preset threshold, the deviation of the feature variable is calculated. Finally, in step S105, a compensation factor is calculated based on the deviation and a preset mapping table, and then superimposed on the drive control command to correct the output gain of the load excitation execution module, thereby realizing the perception consistency closed-loop control of the controlled test object.

[0030] like Figure 2 As shown, the overall architecture mainly consists of three parts: the test site operation station, the real-time control host, and the physical test bench area. The test site operation station is a desktop computer or industrial tablet, which integrates human-machine interface and monitoring software as well as test parameter configuration modules. It is connected to the real-time control host, which is an industrial control computer or embedded controller, via Ethernet or LAN communication. The real-time control host has a layered deployment of a closed-loop control logic core program with a signal drift compensation algorithm, a high-speed data processing and storage engine, and a hardware I / O driver layer. The physical test bench area includes a pulse pressure generator, a pressure sensor under test, and a high-precision signal acquisition terminal. The real-time control host controls the pulse pressure generator, which includes a hydraulic or pneumatic pump station and valve group, through industrial bus control signals. This causes the generator to apply physical pulse pressure input to the pressure sensor under test installed in the pressure chamber. The pressure sensor under test outputs a sensor analog micro-signal to the high-precision signal acquisition terminal, which is an independent DAQ hardware module. The terminal feeds back the converted high-speed digital acquisition data stream to the real-time control host, thus forming a complete hardware closed-loop circuit.

[0031] Example 4: In a fatigue reliability assessment scenario involving variable frequency pulse loading of a common rail pressure sensor for a heavy-duty diesel engine, the system operates under conditions where the pulse frequency dynamically switches between 5Hz and 50Hz. Due to the nonlinear scaling of the pressure rise time caused by the change in load frequency, the fixed-phase feature observation window cannot stably capture the dynamic response characteristics of the controlled test object. To eliminate the influence of frequency changes on sampling timing, the closed-loop control logic module executes a preset gradient threshold. The dynamic calibration procedure involves the closed-loop control logic module obtaining the preset pulse width of the current pulse period through the load excitation execution module. The gradient triggering threshold is determined according to the following formula: ,in, This is a preset gradient threshold, in MPa / ms. The peak value of the pulse pressure load output by the load excitation execution module is expressed in MPa. The preset pulse width for the current load cycle, in milliseconds, is used by the closed-loop control logic module to calculate the first derivative of the original load signal in real time. ,when The system determines when a pressure surge phase has begun, and from that trigger moment, a duration of [duration to be specified] is initiated. The feature observation window.

[0032] Within the locked feature observation window, the closed-loop control logic module performs logic reconstruction to separate the signal disturbance caused by microscopic fatigue damage. The processing steps include: acquiring synchronously acquired data within the window. Group load feedback value and test feedback signals ; Utilizing the dynamic gain coefficient stored in the preceding cycle Calculate the predicted response level, i.e., perform the operation. ,in Index of sampling points within the observation window; calculate the predicted residual sequence. ,in For the first The prediction residuals of each sampling point For the first The measured level of the test feedback signal at each sampling point The corresponding predicted response level value; through the residual sequence The signal disturbance is obtained by performing a mean-averaging operation. If the mean of the signal disturbance exceeds 150% of the physical boundary of the standard response envelope for five consecutive pulse cycles, the closed-loop control logic module determines that a micro-crack has occurred inside the controlled test object and triggers an abnormal drift warning. The closed-loop control logic module then applies the aging factor. Introducing a drive command correction process, this aging factor The closed-loop control logic module determines the number of pulse cycles based on the accumulated pulse count. Retrieve the dimensionless coefficients determined by the preset mapping table, and execute the driving voltage. When calculating, the closed-loop control logic module adopts the following integrated correction model: ,in, This is the control voltage command output to the drive circuit, in volts (V). The reference drive voltage, in volts, is preset by the closed-loop control logic module based on the current test task. This is a dimensionless compensation factor determined based on the deviation of the characteristic variables. As an aging factor, taking a single experimental dataset as an example, when the system detects that the current characteristic variable deviates by 2.1% from the physical boundary of the standard response envelope, the closed-loop control logic module calculates a compensation factor. If the controlled test object has already executed at this time The aging factor retrieved by the closed-loop control logic module during the next pulse cycle. The total correction gain was determined to be 1.023 times. The closed-loop control logic module drives the load excitation execution module to increase the output intensity. Through this integrated correction model, the system achieves decoupling compensation for transient performance fluctuations and long-term structural aging, so that the test sensitivity deviation of the intelligent sensing system under variable frequency dynamic load is maintained within the industrial specification range of less than 0.35%.

[0033] Example 5: In the initial calibration scenario of the first loading deployment of the novel piezoresistive sensor, the system operates under controlled laboratory conditions. Before entering the fatigue loading program, the closed-loop control logic module executes the reference feature extraction procedure, and the drive load excitation execution module executes 500 pulse cycles with a target amplitude of 50% of the full-scale pressure. The signal feedback acquisition module synchronously acquires the dynamic gain coefficient of the sensor in the rising edge interval of each cycle. By constructing a benchmark database containing 500 sampled data and calculating its overall average, vs. population standard deviation The overall average The calculation formula is as follows: ,in, This represents the overall average value of the dynamic gain coefficient. This is the total number of loops, set to 500. For sample index, For the first The dynamic gain coefficient value corresponding to the next cycle is calculated by the closed-loop control logic module as the overall average value. The interval of three times the standard deviation above and below the standard response envelope is defined as the physical boundary of the standard response envelope. This discretized expression based on statistical distribution is used to determine the technical origin of the perceptual consistency judgment. When the system faces the condition of parameter calibration for sensitive components with different material properties, the closed-loop control logic module generates a preset mapping table by performing an offline strain energy step experiment. It then monitors the output feedback of the controlled test object using the controlled bias of the drive gain, records the feedback voltage correction value required to return the sensor output response to the center position of the standard response envelope under different deviation states, and establishes a system with the deviation as the input parameter and a compensation factor... The linear mapping model of the output result, where the compensation factor The calculation formula is as follows: ,in, As a compensation factor, The dimensionless proportional gain coefficient was determined experimentally. The measured value of the dynamic gain coefficient obtained through real-time observation. To determine the overall average value during the initialization phase, the closed-loop control logic module writes discrete data mapping pairs into a non-volatile memory to form a preset mapping table, thereby adapting the control command reconstruction logic to the physical degradation rate of the sensor.

[0034] Example 6: In a scenario where hardware adaptation is performed for piezoresistive sensors from different batches with varying response characteristics, the closed-loop control logic module acquires the original dynamic gain coefficient sequence containing white noise through the signal feedback acquisition module during the initial loading stage. A smoothing coefficient with an increment step size of 0.05 is used. Determine the time constant of the optimal first-order lag filter The smoothing coefficient With filter time constant The relationship follows the formula below: ,in, For smoothing coefficients, The sampling period is set to 10 μs. Let be the filter time constant in milliseconds (ms). The closed-loop control logic module compares the variance of the filtered characteristic variables with the response delay time under a step load, selecting the value that makes the variance of the characteristic variables less than 0.001 and the delay time less than 1 ms as the optimal smoothing coefficient. .

[0035] To address the singularity where the load difference is zero at the beginning of the pulsed pressure load, the closed-loop control logic module calculates the dynamic gain coefficient. The pre-set magnitude is bias constant And add them to the denominator. This non-zero constraint logic maintains the continuity of the closed-loop control decision path. When the system constructs an aging factor through the data storage module... When using the preset mapping table, the closed-loop control logic module obtains the mapping relationship between structural damage and signal drift by executing a multidimensional stress-accelerated degradation experiment. The experimenters selected five groups of controlled test objects with consistent physical specifications and performed the experiment under an overload stress of 1.2 times the rated pressure. The pulse cycle, the closed-loop control logic module at each interval The next loop records the steady-state offset of the characteristic variables from the standard response envelope, by accumulating the number of pulse loops. Perform a linear fitting operation with the steady-state offset to extract the aging constant that reflects the structural stiffness decay. And according to the aging constant Determine the aging factors corresponding to different cycle stages The values ​​are used to fill the lookup table during the execution of the drive voltage. In the correction calculation, the system indexes the corresponding aging factor in real time based on the total number of cycles that the currently controlled test object has run. This allows the output gain of the load excitation execution module to perform a monotonically increasing compensation action as the fatigue depth of the sensitive component increases, keeping the output level fluctuation of the controlled test object within 0.25% throughout the entire life evaluation cycle.

[0036] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation, characterized in that, It includes a load excitation execution module, a signal feedback acquisition module, and a closed-loop control logic module: The signal feedback acquisition module is connected to the output of the controlled test object; The closed-loop control logic module is connected to both the load excitation execution module and the signal feedback acquisition module. The closed-loop control logic module executes the following steps: Step S101: Calculate the first derivative of the load output by the load excitation execution module and compare it with a preset gradient threshold. Select a phase-locked feature observation window at the rising edge of each cycle of the pulse pressure load. Step S102: Obtain the load feedback value and test feedback signal within the feature observation window, and construct a dynamic correlation matrix reflecting the response characteristics of the controlled test object based on the load feedback value, test feedback signal, and the differential relationship of the time axis. Step S103: Perform least squares fitting on the dynamic correlation matrix to extract the dynamic gain coefficient of the test feedback signal to the load change, and input the dynamic gain coefficient into the first-order hysteresis filter to obtain the characteristic variable characterizing the degree of stiffness degradation of the controlled test object structure. Step S104: Compare the feature variable with the standard response envelope stored in the initialization phase. In response to the state where the feature variable deviates from the standard response envelope and the volatility exceeds a preset threshold, calculate the deviation of the feature variable relative to the standard response envelope. Step S105: Calculate the compensation factor based on the deviation and the preset image table, and superimpose the compensation factor into the drive control command of the load excitation execution module. By correcting the output gain of the load excitation execution module, the perception consistency closed-loop control of the controlled test object during the fatigue test process is realized.

2. The closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The closed-loop control logic module also performs the following steps: Step S201, extracting sampling feature points using the high-speed sampling channel established by the feature observation window, and performing phase locking and logic reconstruction on the sampling feature points to separate the signal disturbance caused by the micro-fatigue damage of the controlled test object; Step S202: Monitor the amplitude and phase characteristics of the signal disturbance. When the signal median shift, peak attenuation, or waveform distortion exceeds the preset limit, the closed-loop control logic module generates an abnormal drift warning command and locks the current output state of the load excitation execution module to avoid transient failure of the controlled test object.

3. The closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The sampling time interval of the feature observation window is set to 10μs, and the closed-loop control logic module calculates the dynamic gain coefficient within the feature observation window. The calculation formula is: ,in, as well as These represent the test feedback signal levels at the end and beginning of the feature observation window, respectively. as well as These are the measured load feedback values ​​at the corresponding time points of the characteristic observation window.

4. The closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, When step S105 is executed, the compensation factor is superimposed as a feedback gain coefficient onto the control voltage of the drive circuit to compensate for the sensitivity drift of the controlled test object caused by the change in the elastic modulus of the internal sensitive component.

5. The closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, A first-order hysteresis filter is used to filter out random environmental noise in the dynamic gain coefficient, and the update step size of the characteristic variable is consistent with the pulse period of the pulse pressure load.

6. The closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The initialization phase includes: executing 100 to 500 pulse cycles of preset intensity before loading the test to obtain steady-state distribution data of the feature variables, and setting the physical boundary of the standard response envelope based on three times the standard deviation of the steady-state distribution data.

7. A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The load excitation execution module includes a pulse pressure source module and a load sensing module. The load sensing module is used to provide the raw load signal required to calculate the first derivative of the rising edge.

8. A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 2, characterized in that, The logic reconstruction includes: performing mean denoising on the sampled feature points and constructing a time series model that reflects the performance evolution. The closed-loop control logic module identifies the nonlinear degradation trajectory of the controlled test object based on the predicted residual of the time series model.

9. A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The system also includes a data storage module, which is connected to the closed-loop control logic module to record the historical values ​​of the dynamic correlation matrix, feature variables, and compensation factors.

10. A closed-loop control system for sensor pulse pressure fatigue testing based on signal drift compensation according to claim 1, characterized in that, The closed-loop control logic module introduces an aging factor into the control loop. The aging factor is calculated by the closed-loop control logic module based on the number of cumulative pulse cycles executed and is used to correct the drive control commands.

Images