Pressure sensor metering data calibration method and system

Abstract

The invention provides a calibration method and system for measurement data of a pressure sensor, and the method comprises the steps: obtaining an original output signal of the pressure sensor and synchronous collection data of environment parameters, and carrying out the preprocessing, and obtaining multi-source sensing data and dynamic environment characteristics; based on multi-source sensing data and dynamic environment characteristics, a multi-error coupled comprehensive error model is constructed, and a dynamic compensation coefficient is generated through online parameter identification; according to the dynamic compensation coefficient and a preset confidence evaluation strategy, dynamic compensation calibration is carried out on the sensing data, a high-precision pressure value is generated, and the high-precision pressure value is used for representing a real environment pressure measurement value. By adopting the method, the accuracy and the reliability of the tester in the calibration of the pressure sensor in a multi-error coexistence and complex environment can be improved, and the precision and the stability of the measurement data calibration of the pressure sensor are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of sensor calibration, and in particular relates to a calibration method and system for pressure sensor measurement data. Background Art

[0002] As a core technology in modern industry and scientific research, pressure sensor metrology data calibration is of irreplaceable importance in ensuring the reliability of measurement data and the stability of system performance. With the continuous expansion of high-precision application scenarios such as intelligent manufacturing, aerospace, and medical equipment, the calibration accuracy of pressure sensors directly determines the safety of equipment operation and the credibility of data, and its research value and application needs are becoming increasingly prominent. However, existing calibration methods often rely on a single error correction method or a simple traceability method, which makes it difficult to fully cope with multi-source error interference in complex environments, resulting in insufficient universality and stability of calibration results, especially in dynamic measurements or extreme conditions, where the error control effect is significantly reduced.

[0003] These limitations stem from the fact that several key technical factors in the calibration system have not been effectively resolved. First, the accuracy of the tester is limited by the combined effects of multiple error sources such as zero drift, linear error, and repeatability error, and the existing methods are not sufficiently refined in system compensation. Secondly, the evaluation of comprehensive measurement uncertainty lacks a unified statistical framework, which makes it difficult to quantify the credibility of the calibration results. In addition, the implementation of a multi-level traceability chain is often difficult to fully realize in actual operations due to equipment limitations, affecting the rigor of traceability. These unresolved technical factors make it difficult for the calibration system to balance high accuracy and high reliability when facing complex working conditions, which in turn raises the technical problem of how to improve the performance of the tester when multiple errors coexist.

[0004] However, the current calibration method for pressure sensor measurement data is insufficiently accurate and cannot meet the accuracy requirements of calibration under complex working conditions. Summary of the invention

[0005] Based on this, it is necessary to provide a pressure sensor measurement data calibration method and system to address the above technical problems, which can improve the accuracy and reliability of the tester in the pressure sensor calibration under multiple errors and complex environments, and improve the accuracy and stability of the pressure sensor measurement data calibration.

[0006] In a first aspect, the present application provides a method for calibrating measurement data of a pressure sensor, characterized by comprising: The original output signal of the pressure sensor and the synchronously collected data of the environmental parameters are obtained and preprocessed to obtain multi-source sensing data and dynamic environmental characteristics; Based on multi-source sensor data and dynamic environment characteristics, a comprehensive error model of multi-error coupling is constructed, and dynamic compensation coefficients are generated through online parameter identification; According to the dynamic compensation coefficient and the preset confidence assessment strategy, the sensor data is dynamically compensated and calibrated to generate high-precision pressure values, which are used to characterize the real environment pressure measurement values.

[0007] In a possible embodiment, the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters are obtained and preprocessed to obtain multi-source sensing data and dynamic environmental characteristics, including: The original output signal of the pressure sensor is subjected to wavelet threshold denoising to obtain multi-source sensing data, in which the threshold parameters are dynamically adjusted according to the environmental parameters; Perform first-order difference calculation on multi-source sensor data to generate dynamic pressure change rate; Based on environmental parameters and sensor working time, dynamic environmental characteristics are constructed. The dynamic environmental characteristics include temperature, time and dynamic pressure change rate.

[0008] In a possible embodiment, based on multi-source sensor data and dynamic environment characteristics, a comprehensive error model of multi-error coupling is constructed, and a dynamic compensation coefficient is generated through online parameter identification, including: A joint error model is constructed based on multi-source sensor data and dynamic environment characteristics. The input of the joint error model includes nonlinear sensor signal terms, temperature drift terms, and time drift terms. The recursive least square method is used to identify the temperature-dependent parameters and time-dependent parameters in the joint error model online to generate dynamic compensation coefficients. The nonlinear residual compensation term based on spectrum analysis is superimposed on the joint error model to generate an updated comprehensive error model, and the updated comprehensive error model is used to re-output the dynamic compensation coefficient.

[0009] In a possible embodiment, according to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value, including: Perform nonlinear compensation on multi-source sensor data based on dynamic compensation coefficients to generate initial calibration values; An adaptive Kalman filter is constructed to input the temperature change rate in the dynamic environmental characteristics into the state transfer matrix to track the dynamic error caused by temperature, and generate a filtered pressure value based on the initial calibration value as the observation value; Calculate the residual covariance between the filtered pressure value and the initial calibration value, and evaluate the confidence weight of the Kalman filter based on the residual covariance; The filtered pressure value and the initial calibration value are weightedly fused based on the confidence weight to generate a high-precision pressure value.

[0010] In a possible embodiment, the method further includes: Calculate the RMS error of the high-precision pressure value, which is used to evaluate the calibration accuracy; When the RMS error exceeds the preset threshold, the incremental learning mechanism is triggered to update the dynamic compensation coefficient; The updated dynamic compensation coefficients are recorded and the error counters are reset. The error counters are initialized and generated by the calibration process.

[0011] In a possible embodiment, a nonlinear residual compensation term based on spectrum analysis is superimposed on the joint error model to generate an updated comprehensive error model, including: Perform Fourier transform on historical error data to extract main spectral components. Historical error data is generated by calculating the residual between the calibrated pressure value and the reference pressure value. Determine the order of the nonlinear residual compensation term according to the energy distribution of the main spectral components; Based on the frequency and amplitude of the main spectral components, a sinusoidal function compensation term corresponding to the order is generated, and the sinusoidal function is compensated to the joint error model to generate an updated comprehensive error model.

[0012] In a possible embodiment, weighted fusion of the filtered pressure value and the initial calibration value is performed based on the confidence weight to generate a high-precision pressure value, including: Calculate the difference between the filtered pressure value and the initial calibration value to generate a residual sequence; Calculate the covariance matrix of the residual sequence to generate the residual covariance matrix; The confidence weights are calculated based on the diagonal elements of the inverse matrix of the residual covariance matrix. The confidence weights are inversely proportional to the covariance. The filtered pressure value and the initial calibration value are linearly weighted fused based on the confidence weight to generate a high-precision pressure value.

[0013] In a second aspect, the present application also provides a pressure sensor measurement data calibration system, characterized in that the system comprises: A data preprocessing module is used to obtain the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and perform preprocessing to obtain multi-source sensing data and dynamic environmental characteristics; Dynamic compensation module, which is used to build a comprehensive error model of multi-error coupling based on multi-source sensor data and dynamic environment characteristics, and generate dynamic compensation coefficients through online parameter identification; The calibration pressure value generation module is used to perform dynamic compensation calibration on the sensor data according to the dynamic compensation coefficient and the preset confidence evaluation strategy to generate a high-precision pressure value, and the high-precision pressure value is used to characterize the real environment pressure measurement value.

[0014] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned method for calibrating the measurement data of the pressure sensor when executing the computer program.

[0015] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the calibration method of the pressure sensor measurement data is implemented.

[0016] The above-mentioned pressure sensor measurement data calibration method and system obtains the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and performs preprocessing to obtain multi-source sensor data and dynamic environmental characteristics. Based on these data and characteristics, a comprehensive error model of multi-error coupling is constructed, and a dynamic compensation coefficient is generated through online parameter identification. According to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value for characterizing the real environmental pressure measurement value. The above-mentioned method can effectively solve the accuracy and reliability problems of pressure sensor calibration in the presence of multiple errors and complex environments, and improve the accuracy and stability of pressure sensor measurement data calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic flow chart of a method for calibrating measurement data of a pressure sensor provided in an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a pressure sensor measurement data calibration system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] First, a brief introduction is given to the terms involved in the embodiments of the present application.

[0021] Wavelet threshold denoising is a signal processing technology based on wavelet transform. It decomposes the signal into wavelet coefficients of different frequency components and applies threshold processing to these coefficients to remove noise components while retaining the useful information of the signal. This method uses the multi-resolution analysis characteristics of wavelet transform to effectively distinguish between signals and noise, and is particularly suitable for denoising of non-stationary signals.

[0022] Recursive Least Squares (RLS) is a real-time parameter estimation method used for online recursive estimation of parameters in dynamic systems. It is based on the least squares principle and minimizes the cumulative sum of squared errors by continuously updating the estimated values. This method uses a recursive formula and only requires the current measurement data and the estimated results of the previous moment to efficiently calculate the parameter estimates at the current moment, without reprocessing all historical data, thereby significantly improving the computational efficiency. It is particularly suitable for processing real-time data and dynamically changing system parameter estimation problems, and is widely used in adaptive filtering, system identification, signal processing and other fields.

[0023] The adaptive Kalman filter is an improved filtering algorithm based on the Kalman filter theory, which can automatically adjust the filtering parameters in a dynamic environment to adapt to changes in system characteristics. It dynamically adjusts the filter gain by monitoring the changes in the statistical characteristics of the system (such as the noise covariance matrix) in real time, so that the system state can be effectively estimated when the statistical characteristics of the noise are unknown or changing. This method combines the optimal estimation characteristics and adaptive mechanism of the Kalman filter, which can improve the robustness and accuracy of the filter in complex environments. It is widely used in target tracking, navigation positioning, signal processing, and robot control, especially in scenarios that require real-time processing and high-precision estimation.

[0024] According to the above explanation of terms, the implementation environment of a method for calibrating the measurement data of a pressure sensor provided in an embodiment of the present application is explained. Schematically, the implementation environment includes: a sensor, a terminal, a standard pressure generator and a processor. Among them, the sensor can be connected to the terminal through a signal acquisition module, the terminal communicates with the processor through a data bus, the standard pressure generator can be coupled to the sensor through a mechanical interface to provide a calibration pressure input, the processor integrates a storage module and an algorithm calculation unit to realize multi-source data fusion and dynamic compensation calibration; the sensor can include a pressure sensitive element, a piezoresistive pressure sensor, a temperature and humidity sensor, etc.; the processor can be a central processing unit, an artificial intelligence chip or a multi-core processor, which is not limited here.

[0025] In combination with the above-mentioned explanation of terms and implementation environment, the application scenarios of the embodiments of the present application are explained. The calibration method of pressure sensor measurement data provided in the embodiments of the present application can be applied to, but not limited to, the following scenarios: In the automotive engine management system, the accuracy of the fuel pressure sensor and coolant pressure sensor directly affects the engine's performance and fuel economy. Through this technical solution, these sensors can be calibrated in real time during vehicle operation, compensating for measurement errors caused by factors such as temperature changes and sensor aging, improving the long-term stability and measurement accuracy of the sensor, and thus optimizing the engine's operating efficiency and emission performance.

[0026] In industrial automation production lines, pressure sensors are often used to monitor hydraulic systems, pneumatic systems, and pipeline pressure. For example, in precision manufacturing processes, it is essential to measure the coolant pressure of processing equipment, the pressure of hydraulic fixtures, etc. with high precision. Through dynamic compensation calibration technology, the measurement error of the sensor can be adjusted in real time to ensure the stability of the production process and the consistency of product quality, while reducing equipment failures and production accidents caused by sensor errors.

[0027] Weather stations require high-precision pressure sensors to measure atmospheric pressure for weather forecasting and climate research. Due to the complexity of the meteorological environment (such as rapid changes in temperature, humidity, and air pressure), traditional calibration methods are difficult to meet the needs of high-precision measurements. Through the above technical solution, the pressure sensor can be calibrated dynamically in real time to ensure its measurement accuracy under different meteorological conditions, thereby improving the accuracy and reliability of meteorological data and providing higher quality data support for weather forecasting and climate research.

[0028] Illustratively, the method for calibrating pressure sensor measurement data provided in the embodiment of the present application can also be applied to other application scenarios. It is only used as an example here and is not limited to the specific application scenario.

[0029] In an exemplary embodiment, Figure 1 As shown, a method for calibrating pressure sensor measurement data is provided. This embodiment uses the method applied to a terminal in the aforementioned implementation environment as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. The following steps 101 to 103 are included: Step 101, obtaining the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and performing preprocessing to obtain multi-source sensing data and dynamic environmental characteristics.

[0030] Specifically, the original output signal is obtained through a high-precision pressure sensor, and the environmental parameter sensor is used to synchronously collect environmental parameter data, such as temperature, humidity, and air pressure. The output signals of these sensors are transmitted to the data acquisition module for preliminary signal conditioning and digital processing. Through preprocessing operations such as filtering, denoising, and data format conversion, high-quality multi-source sensor data and dynamic environmental characteristics are obtained, providing a reliable data basis for subsequent error modeling.

[0031] Step 102: construct a comprehensive error model of multi-error coupling based on multi-source sensor data and the dynamic environment characteristics, and generate dynamic compensation coefficients through online parameter identification.

[0032] Specifically, a comprehensive error model containing multiple error sources is constructed based on preprocessed multi-source sensor data and dynamic environmental characteristics. The model can reflect the complex coupling relationship between sensor measurement errors and environmental parameter changes. Through the online parameter identification algorithm, the deviation between the sensor output and the known reference value is analyzed in real time, the model parameters are dynamically adjusted, and the dynamic compensation coefficients that match the current environment and sensor status are generated. This process ensures the real-time and adaptability of the error model, and can effectively cope with environmental changes and sensor performance drift.

[0033] Step 103, according to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value, and the high-precision pressure value is used to represent the real environment pressure measurement value.

[0034] Specifically, the generated dynamic compensation coefficient is combined with a preset confidence assessment strategy to perform dynamic compensation calibration on the sensor data. The confidence assessment strategy is used to judge the reliability of the compensation coefficient and ensure the accuracy of the calibration process. The data after compensation calibration generates a high-precision pressure value, which can more accurately represent the pressure measurement value in the real environment.

[0035] The above-mentioned calibration method for pressure sensor measurement data obtains multi-source sensor data and dynamic environmental characteristics by obtaining the original output signal of the pressure sensor and the synchronously collected data of environmental parameters and preprocessing them. Based on these data and characteristics, a comprehensive error model of multi-error coupling is constructed, and a dynamic compensation coefficient is generated through online parameter identification. According to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value for characterizing the real environmental pressure measurement value. The above-mentioned method can effectively solve the accuracy and reliability problems of pressure sensor calibration in the coexistence of multiple errors and complex environments, and improve the accuracy and stability of pressure sensor measurement data calibration.

[0036] In a possible embodiment, the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters are obtained and preprocessed to obtain multi-source sensing data and dynamic environmental characteristics, including: Step 201 , performing wavelet threshold denoising processing on the original output signal of the pressure sensor to obtain multi-source sensing data, wherein the threshold parameters are dynamically adjusted according to the environmental parameters.

[0037] Specifically, the original output signal (such as voltage or current signal) of the pressure sensor is acquired in real time through the data acquisition terminal, and environmental parameters (such as temperature and humidity) are collected at the same time. The threshold parameters of the wavelet denoising algorithm are dynamically adjusted according to the current ambient temperature. For example, the threshold is increased in a high temperature environment to suppress thermal noise, and the threshold is reduced in a low temperature environment to retain signal details. The original signal is decomposed by multi-scale wavelet, and the high-frequency noise component is suppressed according to the adjusted threshold, and the denoised multi-source sensor data is reconstructed.

[0038] Step 202 , performing first-order difference calculation on multi-source sensor data to generate a dynamic pressure change rate.

[0039] For example, for the denoised multi-source sensor data, the pressure signals at adjacent time points are first-order differentially calculated with a fixed time window as the interval to calculate the dynamic pressure change rate. Specifically, the difference between the current pressure value and the previous pressure value is divided by the time interval to quantify the instantaneous change trend of the pressure signal.

[0040] Step 203: construct dynamic environmental characteristics based on environmental parameters and sensor working time, and the dynamic environmental characteristics include temperature, time and dynamic pressure change rate.

[0041] Exemplarily, the collected ambient temperature, sensor cumulative working time and dynamic pressure change rate are integrated to construct a dynamic environment feature vector. Specifically, using the timestamp as the index, the temperature data, working time data and dynamic pressure change rate are aligned in time series to generate a multidimensional feature matrix containing temperature-time-pressure change rate, and the dimensional difference is eliminated through standardization.

[0042] In a possible embodiment, based on multi-source sensor data and dynamic environment characteristics, a comprehensive error model of multi-error coupling is constructed, and a dynamic compensation coefficient is generated through online parameter identification, including: Step 301 : constructing a joint error model based on multi-source sensor data and dynamic environment characteristics. The input of the joint error model includes a nonlinear sensor signal term, a temperature drift term, and a time drift term.

[0043] For example, the nonlinear pressure signal in the multi-source sensing data, the temperature drift component and the time drift component in the dynamic environment characteristics can be used as input items to construct a joint error model. Specifically, the nonlinear sensing signal item is represented by the nonlinear fitting function of the pressure signal, the temperature drift item is described by the coupling polynomial of the temperature and pressure signal, and the time drift item is expressed by the exponential decay function of the sensor working time, forming a mathematical model of multi-error coupling to solve the problem of insufficient accuracy of the traditional single error compensation model under complex working conditions and improve the comprehensive suppression ability of the model against multi-source interference.

[0044] Step 302: Use the recursive least square method to perform online identification on the temperature-dependent parameters and time-dependent parameters in the joint error model to generate dynamic compensation coefficients.

[0045] Specifically, the multi-source sensor data and dynamic environmental characteristics collected in real time during the operation of the sensor are obtained and input into the joint error model. The temperature-dependent parameters and time-dependent parameters are iteratively updated using the recursive least squares method, and the compensation coefficients are dynamically adjusted to ensure that the model parameters are adaptively optimized as the environment changes, avoiding the high computational overhead of traditional batch processing algorithms, achieving low-latency real-time parameter identification, and making the compensation coefficients dynamically adapt to the error change trend in complex environments, thereby improving the robustness of the method.

[0046] Step 303 , superimposing the nonlinear residual compensation term based on spectrum analysis into the joint error model to generate an updated comprehensive error model, and the updated comprehensive error model is used to re-output the dynamic compensation coefficient.

[0047] Specifically, the residual data of the joint error model is spectrally analyzed to identify the main interference frequency bands. Nonlinear residual compensation terms in the form of sinusoidal functions are generated based on frequency domain characteristics and superimposed on the original model to form an updated comprehensive error model. By introducing nonlinear residual compensation terms, the accuracy of the error model is further optimized so that it can more comprehensively cover the sources of errors. The updated model recalculates the dynamic compensation coefficients and outputs them to achieve more accurate error compensation.

[0048] In a possible embodiment, according to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value, including: Step 401 : performing nonlinear compensation on multi-source sensor data based on dynamic compensation coefficients to generate initial calibration values.

[0049] Exemplarily, the dynamic compensation coefficient is input into a predefined nonlinear error correction model, and the pressure signal in the multi-source sensor data is corrected point by point to eliminate the inherent nonlinear error of the sensor and generate an initial calibration value. Specifically, the compensation coefficient is applied segmented according to the amplitude range of the pressure signal to ensure the linearity optimization within the full range, so as to improve the benchmark accuracy of static pressure measurement.

[0050] Step 402, construct an adaptive Kalman filter, input the temperature change rate in the dynamic environmental characteristics into the state transfer matrix to track the dynamic error caused by temperature, and generate a filtered pressure value based on the initial calibration value as the observation value.

[0051] Specifically, an adaptive Kalman filter is designed, and the temperature change rate in the dynamic environmental characteristics is used as the adjustment parameter of the state transfer matrix to adjust the dynamic response characteristics of the filter in real time. At the same time, the initial calibration value is input into the filter as the observation value, and the filtered pressure value is generated by fusing the state prediction and the observation value. The adaptive Kalman filter tracks the dynamic characteristics of the environment through the temperature change rate, effectively suppresses the pressure drift caused by temperature transients, and reduces the dynamic pressure tracking error.

[0052] Step 403, calculating the residual covariance between the filtered pressure value and the initial calibration value, and evaluating the confidence weight of the Kalman filter according to the residual covariance.

[0053] For example, the residual of the filtered pressure value and the initial calibration value can be calculated in real time, the covariance matrix of the residual sequence can be counted, and the confidence weight of the Kalman filter can be evaluated according to the covariance. Specifically, the smaller the covariance, the higher the consistency of the filtering result with the initial calibration value, and the higher the weight of the filtering value. The credibility of the filtering result is quantified by the residual covariance, which provides an objective basis for dynamic fusion and avoids the limitations of a single compensation method.

[0054] Step 404 , weighted fusion of the filtered pressure value and the initial calibration value is performed based on the confidence weight to generate a high-precision pressure value.

[0055] Specifically, the filtered pressure value and the initial calibration value are linearly weighted according to the confidence weight to generate the final high-precision pressure value. When the ambient temperature fluctuates violently, the weight of the filtered value is increased to enhance the dynamic tracking capability; when the environment is stable, the weight of the initial calibration value is increased to maintain static accuracy.

[0056] In a possible embodiment, the method may further include: Step 501 , calculating the root mean square error of the high-precision pressure value, where the root mean square error is used to evaluate the calibration accuracy.

[0057] For example, the high-precision pressure value and the reference pressure value provided by the standard pressure generator can be obtained, and the data points within a fixed time window can be used as samples to calculate the root mean square error between the two. Specifically, the square of the deviation of each sampling point is accumulated, the average value is taken and the square root is taken to quantify the current calibration accuracy. The root mean square error calculation can objectively reflect the overall deviation level of the calibration result, monitor the calibration performance in real time, and quickly identify the accuracy degradation caused by sudden environmental changes or sensor aging.

[0058] Step 502: When the root mean square error exceeds a preset threshold, an incremental learning mechanism is triggered to update the dynamic compensation coefficient.

[0059] Specifically, a root mean square error threshold (for example, 0.2% FS) can be preset. When the calculated value exceeds the threshold, the incremental learning mechanism is started to adjust the dynamic compensation coefficient based on the direction of the error gradient. For example, the temperature-dependent parameters and time-dependent parameters are fine-tuned through the back-propagation algorithm to iteratively update the compensation coefficient in the direction of reducing the residual. The incremental learning mechanism realizes self-optimization of the compensation coefficient through dynamic feedback, solves the performance degradation problem of traditional fixed parameter models in long-term operation, and ensures that the calibration accuracy continues to meet the requirements.

[0060] Step 503, record the updated dynamic compensation coefficient and reset the error counter, the error counter is initialized and generated through the calibration process.

[0061] Specifically, the updated dynamic compensation coefficient is stored in the non-volatile memory, and the error counter is reset to zero. The error counter is set to the default value when the system is initialized (, and reset to the initial state after each incremental learning is triggered. It is used to count the frequency of continuous over-threshold events. By recording the updated parameters and resetting the counter, frequent model oscillations caused by short-term interference are prevented, while retaining long-term learning results, improving system robustness and stability.

[0062] In a possible embodiment, a nonlinear residual compensation term based on spectrum analysis is superimposed on the joint error model to generate an updated comprehensive error model, including: Step 601, Fourier transform is performed on the historical error data to extract the main spectral components. The historical error data is generated by calculating the residual between the calibrated pressure value and the reference pressure value.

[0063] Specifically, the historical error data is Fourier transformed to extract the main spectral components. The historical error data is generated by calculating the residual between the calibrated pressure value and the reference pressure value. Through Fourier transform, the frequency band of periodic interference or high-frequency noise can be accurately located, providing data support for the subsequent compensation item design, solving the problem that traditional time domain analysis methods are insensitive to periodic errors.

[0064] Step 602: Determine the order of the nonlinear residual compensation term according to the energy distribution of the main spectral components.

[0065] Specifically, the energy cumulative distribution of the main spectral components is counted, and when the cumulative energy ratio reaches a preset threshold, the number of corresponding frequency bands is used as the order of the nonlinear residual compensation term. For example, if the energy ratio of the first three frequency bands reaches 95%, the compensation term order is set to 3. The compensation order is dynamically selected based on the energy distribution to avoid overfitting or undercompensation problems caused by fixed orders, and minimize computing resource consumption while ensuring accuracy.

[0066] Step 603: Generate a sinusoidal function compensation term corresponding to the order based on the frequency and amplitude of the main spectral components, and compensate the sinusoidal function into the joint error model to generate an updated comprehensive error model.

[0067] Specifically, based on the frequency and amplitude of the main spectral components, sinusoidal function compensation terms corresponding to the order are generated, and these sinusoidal function compensation terms are superimposed on the joint error model to generate an updated comprehensive error model. By introducing sinusoidal function compensation terms, the periodic components in the error signal can be compensated in a targeted manner, and the performance of the error model can be further optimized, so that it can more comprehensively reflect the actual error characteristics of the sensor.

[0068] In a possible embodiment, weighted fusion of the filtered pressure value and the initial calibration value based on the confidence weight to generate a high-precision pressure value may include: Step 701, performing difference calculation between the filtered pressure value and the initial calibration value to generate a residual sequence.

[0069] For example, by acquiring the filtered pressure value and the initial calibration value in real time, the difference between the two is calculated at each time point to generate a residual sequence. Specifically, the filtered pressure value at each sampling moment is subtracted from the initial calibration value at the corresponding moment to obtain the residual value and form a time series. The difference calculation directly reflects the deviation distribution between the filtering result and the initial calibration value, providing basic data for subsequent reliability evaluation and quickly identifying abnormal fluctuations or model mismatch problems.

[0070] Step 702: Calculate the covariance matrix of the residual sequence to generate a residual covariance matrix.

[0071] Specifically, the residual data within a fixed time window is used as samples, the covariance relationship of the residual values ​​at each moment is calculated, the residual fluctuation characteristics are quantified, and the covariance matrix characterizes the statistical distribution characteristics of the residual sequence, revealing the correlation between the filtering results and the initial calibration values, providing a mathematical basis for dynamic weight allocation.

[0072] Step 703: Calculate the confidence weight according to the diagonal elements of the inverse matrix of the residual covariance matrix, where the confidence weight is inversely proportional to the covariance.

[0073] Specifically, the diagonal elements of the inverse matrix of the residual covariance matrix are extracted to calculate the confidence weight at each time point. The smaller the covariance value, the larger the corresponding weight, and vice versa. For example, when the residual covariance is the minimum at a certain moment, the highest weight is given to the filtered pressure value, and the weight of the initial calibration value is reduced accordingly. The covariance-based weight allocation mechanism dynamically balances the credibility of the filtering result and the initial calibration value, suppresses the impact of the single path error on the final output, and improves the robustness of the fusion result.

[0074] Step 704 , performing linear weighted fusion on the filtered pressure value and the initial calibration value based on the confidence weight to generate a high-precision pressure value.

[0075] Specifically, the filtered pressure value and the initial calibration value are linearly weighted and summed according to the weight value to generate the final high-precision pressure value. The adaptive weighted fusion strategy takes into account both dynamic tracking capability and static calibration accuracy, reducing the comprehensive error of the output pressure value under complex working conditions to meet the needs of high-stability industrial measurement.

[0076] In summary, a method for calibrating pressure sensor measurement data provided in an embodiment of the present application obtains the synchronously collected original signal and environmental parameters of the pressure sensor, uses wavelet threshold denoising and dynamic pressure change rate calculation to generate multi-source sensor data and dynamic environment feature vectors, and quantifies the coupling relationship between temperature, time drift and nonlinear errors; constructs a comprehensive error model including nonlinear terms, temperature drift terms and time drift terms, uses recursive least squares method to identify dynamic compensation coefficients online, and superimposes periodic residual compensation terms based on spectrum analysis to achieve collaborative modeling and precise correction of multiple error sources; tracks dynamic environmental disturbances through an adaptive Kalman filter, combines the filtering results with the initial calibration value with confidence weights to balance static accuracy and dynamic response characteristics; introduces a closed-loop verification mechanism to calculate the output root mean square error, triggers incremental learning to optimize the compensation coefficient and resets the error counter to form long-term stability control. The above technical solution improves the accuracy and reliability of pressure sensor calibration under multiple error coexistence and complex working conditions through the synergistic effect of multi-source data fusion, dynamic parameter identification, residual spectrum compensation and closed-loop self-optimization technology, effectively suppresses the coupling effects of temperature drift, time drift and high-frequency interference, improves the accuracy and stability of pressure sensor calibration, and enhances the adaptability and reliability of the calibration process.

[0077] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0078] Based on the same inventive concept, the embodiment of the present application also provides a pressure sensor measurement data calibration system for implementing the pressure sensor measurement data calibration method involved above. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the pressure sensor measurement data calibration system provided below can refer to the limitations of the pressure sensor measurement data calibration method above, and will not be repeated here.

[0079] In an exemplary embodiment, Figure 2 As shown, a pressure sensor measurement data calibration system 10 is provided, comprising: The data preprocessing module 11 is used to obtain the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and perform preprocessing to obtain multi-source sensing data and dynamic environmental characteristics.

[0080] The dynamic compensation module 12 is used to construct a comprehensive error model of multi-error coupling based on multi-source sensor data and dynamic environment characteristics, and generate dynamic compensation coefficients through online parameter identification.

[0081] The calibration pressure value generation module 13 is used to perform dynamic compensation calibration on the sensor data according to the dynamic compensation coefficient and the preset confidence evaluation strategy to generate a high-precision pressure value, and the high-precision pressure value is used to represent the real environment pressure measurement value.

[0082] In a possible embodiment, the data preprocessing module 11 may include: The data denoising unit is used to perform wavelet threshold denoising on the original output signal of the pressure sensor to obtain multi-source sensing data, wherein the threshold parameters are dynamically adjusted according to the environmental parameters.

[0083] The dynamic pressure change calculation unit is used to perform first-order difference calculation on multi-source sensor data to generate a dynamic pressure change rate.

[0084] The dynamic environment feature construction unit is used to construct dynamic environment features based on environmental parameters and sensor working time. The dynamic environment features include temperature, time and dynamic pressure change rate.

[0085] In a possible embodiment, the dynamic compensation module 12 may include: The joint error model unit is used to construct a joint error model based on multi-source sensor data and dynamic environment characteristics. The input of the joint error model includes a nonlinear sensor signal term, a temperature drift term, and a time drift term.

[0086] The dynamic compensation unit is used to perform online identification of temperature-dependent parameters and time-dependent parameters in the joint error model using a recursive least squares method to generate dynamic compensation coefficients.

[0087] The integrated error model unit is used to superimpose the nonlinear residual compensation term based on spectrum analysis into the joint error model to generate an updated integrated error model, and the updated integrated error model is used to re-output the dynamic compensation coefficient.

[0088] In a possible embodiment, the calibration pressure value generating module 13 may include: The initial calibration unit is used to perform nonlinear compensation on multi-source sensor data based on dynamic compensation coefficients to generate initial calibration values.

[0089] The adaptive Kalman filter unit is used to construct an adaptive Kalman filter, input the temperature change rate in the dynamic environmental characteristics into the state transfer matrix to track the dynamic error caused by temperature, and generate a filtered pressure value based on the initial calibration value as the observation value.

[0090] The confidence weight calculation unit is used to calculate the residual covariance between the filtered pressure value and the initial calibration value, and evaluate the confidence weight of the Kalman filter according to the residual covariance.

[0091] The high-precision pressure value generating unit is used to perform weighted fusion on the filtered pressure value and the initial calibration value based on the confidence weight to generate a high-precision pressure value.

[0092] In a possible embodiment, the pressure sensor measurement data calibration system 10 may further include: The calibration accuracy evaluation module is used to calculate the root mean square error of the high-precision pressure value, and the root mean square error is used to evaluate the calibration accuracy.

[0093] The incremental learning module is used to trigger the incremental learning mechanism to update the dynamic compensation coefficient when the root mean square error exceeds a preset threshold.

[0094] The error counting module is used to record the updated dynamic compensation coefficient and reset the error counter, and the error counter is initialized and generated through the calibration process.

[0095] In a possible embodiment, the comprehensive error model unit may include: The main spectrum extraction subunit is used to perform Fourier transform on the historical error data to extract the main spectrum components. The historical error data is generated by calculating the residual between the calibrated pressure value and the reference pressure value.

[0096] The order determination subunit is used to determine the order of the nonlinear residual compensation term according to the energy distribution of the main spectral components.

[0097] The model updating subunit is used to generate a sinusoidal function compensation term corresponding to the order based on the frequency and amplitude of the main spectral components, and to compensate the sinusoidal function into the joint error model to generate an updated comprehensive error model.

[0098] In a possible embodiment, the high-precision pressure value generating unit may include: The difference calculation subunit is used to calculate the difference between the filtered pressure value and the initial calibration value to generate a residual sequence.

[0099] The covariance matrix calculation subunit is used to calculate the covariance matrix of the residual sequence and generate a residual covariance matrix.

[0100] The confidence weight subunit is used to calculate the confidence weight according to the diagonal elements of the inverse matrix of the residual covariance matrix, and the confidence weight is inversely proportional to the covariance.

[0101] The weighted fusion subunit is used to perform linear weighted fusion on the filtered pressure value and the initial calibration value based on the confidence weight to generate a high-precision pressure value.

[0102] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the aforementioned method for calibrating measurement data of a pressure sensor when executing the computer program.

[0103] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0104] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can refer to the partial description of the method embodiments. The device embodiments described above are only schematic, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0105] The above-mentioned embodiments only express several implementation methods of the embodiments of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the embodiments of the present application. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the embodiments of the present application, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. A method for calibrating pressure sensor measurement data, characterized in that: The method comprises: The original output signal of the pressure sensor and the synchronously collected data of the environmental parameters are obtained and preprocessed to obtain multi-source sensing data and dynamic environmental characteristics; Based on the multi-source sensor data and the dynamic environment characteristics, a comprehensive error model of multi-error coupling is constructed, and a dynamic compensation coefficient is generated through online parameter identification; According to the dynamic compensation coefficient and the preset confidence evaluation strategy, the sensor data is dynamically compensated and calibrated to generate a high-precision pressure value, and the high-precision pressure value is used to characterize the real environment pressure measurement value.

2. The method according to claim 1, characterized in that The method of obtaining the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and preprocessing them to obtain multi-source sensing data and dynamic environmental characteristics includes: Performing wavelet threshold denoising processing on the original output signal of the pressure sensor to obtain the multi-source sensing data, wherein the threshold parameter is dynamically adjusted according to the environmental parameter; Performing first-order difference calculation on the multi-source sensor data to generate a dynamic pressure change rate; Based on the environmental parameters and the sensor working time, the dynamic environmental characteristics are constructed, and the dynamic environmental characteristics include temperature, time and the dynamic pressure change rate.

3. The method according to claim 2, characterized in that The method of constructing a comprehensive error model of multi-error coupling based on the multi-source sensor data and the dynamic environment characteristics, and generating a dynamic compensation coefficient through online parameter identification, includes: Building a joint error model based on the multi-source sensor data and the dynamic environment characteristics, wherein the input of the joint error model includes a nonlinear sensor signal term, a temperature drift term, and a time drift term; Using a recursive least square method to perform online identification on the temperature-dependent parameters and the time-dependent parameters in the joint error model to generate the dynamic compensation coefficient; The nonlinear residual compensation term based on spectrum analysis is superimposed on the joint error model to generate an updated comprehensive error model, and the updated comprehensive error model is used to re-output the dynamic compensation coefficient.

4. The method according to claim 1, characterized in that The step of dynamically compensating and calibrating the sensor data according to the dynamic compensation coefficient and the preset confidence evaluation strategy to generate a high-precision pressure value includes: Performing nonlinear compensation on the multi-source sensing data based on the dynamic compensation coefficient to generate an initial calibration value; Constructing an adaptive Kalman filter, inputting the temperature change rate in the dynamic environmental feature into a state transfer matrix to track the dynamic error caused by temperature, and generating a filtered pressure value based on the initial calibration value as an observation value; calculating a residual covariance between the filtered pressure value and the initial calibration value, and evaluating a confidence weight of a Kalman filter according to the residual covariance; The filtered pressure value and the initial calibration value are weightedly fused based on the confidence weight to generate the high-precision pressure value.

5. The method according to claim 1, characterized in that Also includes: Calculating a root mean square error of the high-precision pressure value, wherein the root mean square error is used to evaluate calibration accuracy; When the root mean square error exceeds a preset threshold, an incremental learning mechanism is triggered to update the dynamic compensation coefficient; The updated dynamic compensation coefficient is recorded and an error counter is reset, wherein the error counter is initialized and generated through a calibration process.

6. The method according to claim 3, characterized in that: The nonlinear residual compensation term based on spectrum analysis is superimposed on the joint error model to generate an updated comprehensive error model, including: Performing Fourier transformation on the historical error data to extract main spectral components, the historical error data being generated by calculating the residual difference between the calibrated pressure value and the reference pressure value; Determining the order of the nonlinear residual compensation term according to the energy distribution of the main spectral components; Based on the frequency and amplitude of the main spectral components, a sinusoidal function compensation term corresponding to the order is generated, and the sinusoidal function is compensated to the joint error model to generate the updated comprehensive error model.

7. The method according to claim 4, characterized in that The filtered pressure value and the initial calibration value are weightedly integrated based on the confidence weight to generate the high-precision pressure value, including: Calculating the difference between the filtered pressure value and the initial calibration value to generate a residual sequence; Performing covariance matrix calculation on the residual sequence to generate a residual covariance matrix; Calculating the confidence weight according to the diagonal elements of the inverse matrix of the residual covariance matrix, wherein the confidence weight is inversely proportional to the covariance; The filtered pressure value and the initial calibration value are linearly weighted fused based on the confidence weight to generate the high-precision pressure value.

8. A pressure sensor measurement data calibration system, characterized in that: The system comprises: A data preprocessing module is used to obtain the original output signal of the pressure sensor and the synchronously collected data of the environmental parameters and perform preprocessing to obtain multi-source sensing data and dynamic environmental characteristics; A dynamic compensation module, used to construct a comprehensive error model of multi-error coupling based on the multi-source sensor data and the dynamic environment characteristics, and generate dynamic compensation coefficients through online parameter identification; The calibration pressure value generation module is used to perform dynamic compensation calibration on the sensor data according to the dynamic compensation coefficient and a preset confidence evaluation strategy to generate a high-precision pressure value, and the high-precision pressure value is used to characterize the real environment pressure measurement value.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images