A temperature-compensated pressure sensor correction method and system

Abstract

The application discloses a temperature compensation-based pressure sensor correction method and system, which comprises the following steps: acquiring corresponding multiple temperature data and pressure data of a pressure sensor in a time sequence, and constructing a multi-dimensional state parameter of the pressure sensor according to the multiple temperature data and pressure data; inputting the multi-dimensional state parameter into a linear pressure compensation model after feature conversion to obtain a high-order pressure compensation model, acquiring the coefficients of the high-order pressure compensation model, and obtaining original tensor data; decomposing the original tensor by using a tensor decomposition algorithm to obtain core tensor data and a factor matrix, and calculating a compensation pressure prediction value according to the core tensor data and the factor matrix; and constructing a sample set, adjusting the core tensor data and the factor matrix to calculate the error between the compensation pressure prediction value and an actual compensation pressure value in a multi-round test process, and optimizing the core tensor data and the factor matrix in a manner of minimizing the error.

Description

Technical Field

[0001] This invention relates to the field of pressure sensor technology, and in particular to a pressure sensor calibration method and system based on temperature compensation. Background Technology

[0002] Traditional temperature-compensated pressure sensors only consider the current temperature value, calculating the pressure compensation value by inputting the current temperature value as a linear function. However, this traditional pressure compensation method has the following technical problems: the linear pressure compensation model that only considers the current temperature ignores the influence of dynamic parameters such as the rate of temperature change, the rate of pressure change, and the temperature gradient on the pressure sensor response. Therefore, traditional linear pressure compensation based on temperature has significant errors in rapidly changing or non-uniform temperature fields. Furthermore, due to cost considerations, the embedded pressure calculation system of traditional pressure sensors is relatively simple. When multi-dimensional data exists, embedding multi-dimensional data to calculate pressure compensation results in high-order polynomials, leading to an explosion of parameters that makes calculation impossible. Therefore, how to reduce the amount of pressure compensation data calculation without affecting the accuracy of pressure compensation is a problem that needs to be solved. Summary of the Invention

[0003] One objective of this invention is to provide a pressure sensor calibration method and system based on temperature compensation. The method and system provide multidimensional dynamic parameters of temperature as calibration parameters for the pressure sensor. These multidimensional dynamic parameters include the rate of temperature change, temperature gradient, and real-time temperature. Therefore, this invention utilizes the multidimensional dynamic parameters of temperature to calibrate the pressure sensor values, enabling the pressure sensor to adapt to scenarios with rapid temperature changes and improving the pressure detection accuracy of the pressure sensor in extreme temperature environments.

[0004] Another objective of this invention is to provide a pressure sensor calibration method and system based on temperature compensation. The method and system provide tensor decomposition of dynamic parameters, including but not limited to temperature-related parameters, in a high-order linear pressure compensation function, decomposing the high-order tensor into a low-order core tensor, and using the low-order core tensor as the core parameter for pressure sensor compensation. Therefore, this invention utilizes tensor decomposition technology to use simple parameters as inputs to the pressure sensor embedded system from complex parameter types in the high-order linear function used for pressure compensation, thereby significantly reducing the amount of parameter calculation in the actual pressure calibration process.

[0005] Another objective of this invention is to provide a pressure sensor calibration method and system based on temperature compensation. When performing tensor decomposition on a high-order linear pressure compensation function, the method and system can employ, but are not limited to, the least squares method to decompose the core tensor to be obtained. The least squares method is calculated based on historical parameters, and the core tensor that satisfies the core relationship between pressure and temperature parameters is fitted by the least squares method. Therefore, this invention reduces the number of computational parameters while ensuring the accuracy of pressure compensation during the pressure compensation calculation process.

[0006] To achieve at least one of the above-mentioned objectives, the present invention further provides a pressure sensor calibration method based on temperature compensation, the method comprising: Multiple temperature and pressure data corresponding to the pressure sensor are acquired in a time series, and multidimensional state parameters of the pressure sensor are constructed based on the multiple temperature and pressure data. The multidimensional state parameters are transformed and then input into the linear pressure compensation model to obtain a higher-order pressure compensation model. The coefficients of the higher-order pressure compensation model are obtained to obtain the original tensor data. The original tensor is decomposed using a tensor decomposition algorithm to obtain core tensor data and a factor matrix. The predicted value of compensation pressure is then calculated based on the core tensor data and the factor matrix. A sample set is constructed using the predicted and actual compensation pressure values. During multiple rounds of testing, the error between the predicted and actual compensation pressure values ​​is calculated by adjusting the core tensor data and factor matrix. The core tensor data and factor matrix are then optimized by minimizing the error.

[0007] According to a preferred embodiment of the present invention, the multidimensional state parameters of the pressure sensor include the original pressure value, temperature value, temperature change rate, temperature gradient, and pressure change rate. The original pressure value, temperature value, temperature change rate, temperature gradient, and pressure change rate are normalized respectively to obtain corresponding type feature data. The feature data are input into the linear pressure compensation model to obtain the higher-order pressure compensation model.

[0008] According to another preferred embodiment of the present invention, the tensor decomposition algorithm includes: using the obtained high-order pressure compensation model coefficients as the original tensor for Tucker decomposition, obtaining core tensor data and factor matrices with parameter sizes smaller than the original tensor through principal component analysis, wherein each dimension of the multidimensional state parameters is calculated to obtain a corresponding factor matrix, and further calculating the modulus product of the core tensor and each dimension factor matrix to obtain a pressure compensation prediction value based on multidimensional temperature characteristics.

[0009] According to another preferred embodiment of the present invention, the actual applied pressure value and the displayed value of the pressure sensor are obtained based on the results of multiple rounds of testing. The actual compensated pressure value is calculated using the actual pressure value and the displayed value. The difference between the actual compensated pressure value and the pressure compensation predicted value is calculated. An objective function is constructed based on the difference between the actual compensated pressure value and the pressure compensation predicted value. The objective function includes minimizing the mean square error between the actual compensated pressure value and the pressure compensation predicted value. The core tensor data and the factor matrix are optimized based on the objective function.

[0010] According to another preferred embodiment of the present invention, a sinusoidal voltage signal of a specific frequency is generated in the bridge circuit of the pressure sensor, the phase and amplitude of the sinusoidal voltage signal are acquired, and the phase and amplitude ratio of different frequencies at the same detection time are calculated. Feature data is constructed based on the phase, amplitude ratio and frequency. The feature data of phase, amplitude ratio and frequency are integrated into the feature data of the original pressure value, temperature value, temperature change rate and pressure change rate to construct seven-dimensional feature data. The seven-dimensional feature data is subjected to tensor decomposition to obtain the corresponding core tensor data and factor matrix, and the core tensor data and factor matrix are optimized.

[0011] According to another preferred embodiment of the present invention, a heat conduction model of the pressure sensor is constructed to represent the temperature gradient of the pressure sensor in the thickness direction of the sensing area. The relationship of the heat conduction model in the frequency domain is established by using Fourier transform. Furthermore, a mapping relationship between the specific frequency response of the pressure sensor and the temperature gradient is constructed by using detection calibration. The frequency response is used to replace the actual detected temperature gradient.

[0012] According to another preferred embodiment of the present invention, the detection calibration method includes: using multiple temperature sensors to detect the true temperature gradient of the pressure sensor, and obtaining the phase and amplitude ratio of the pressure sensor bridge at a specific frequency, and using the phase and amplitude ratio as a feature transformation to construct a mapping relationship between the frequency and the frequency, wherein the mapping relationship is established by a neural network.

[0013] To achieve at least one of the above-mentioned objectives, the present invention further provides a temperature-compensated pressure sensor calibration system, wherein the system performs the above-mentioned temperature-compensated pressure sensor calibration method.

[0014] The present invention further provides a computer-readable storage medium storing a computer program that is executed by a processor to implement the above-described temperature-compensated pressure sensor calibration method. Attached Figure Description

[0015] Figure 1This invention presents a pressure sensor calibration method based on temperature compensation. Detailed Implementation

[0016] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0017] It is understood that the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0018] Please combine Figure 1 This invention discloses a pressure sensor calibration method and system based on temperature compensation, wherein the method mainly includes the following core steps: S01. Obtain multiple temperature and pressure data corresponding to the pressure sensor in the time series, and construct multi-dimensional state parameters of the pressure sensor based on the multiple temperature and pressure data; S02. After performing feature transformation on the multidimensional state parameters, input them into the linear pressure compensation model to obtain a higher-order pressure compensation model. Obtain the coefficients of the higher-order pressure compensation model to obtain the original tensor data. S03. Decompose the original tensor using a tensor decomposition algorithm to obtain core tensor data and factor matrix, and calculate the compensation pressure prediction value based on the core tensor data and factor matrix. S04. Construct a sample set by the predicted compensation pressure value and the actual compensation pressure value. During multiple rounds of testing, calculate the error between the predicted compensation pressure value and the actual compensation pressure value by adjusting the core tensor data and factor matrix, and optimize the core tensor data and factor matrix by minimizing the error.

[0019] Specifically, after collecting relevant temperature and pressure data, the present invention performs feature transformation to obtain the following state features: ,in This indicates the original pressure characteristics of the pressure sensor. Indicates temperature characteristics, This indicates the characteristic of the rate of change of pressure. Indicates the rate of temperature change. The temperature gradient characteristics are represented, and these characteristics are input into a linear pressure compensation model. After polynomial expansion, the following compensation function is obtained: ; in It is a five-dimensional primitive tensor that contains all the coefficients of the aforementioned polynomial expansion, wherein the size of the primitive tensor is ,in It is expressed as the power of the highest-order polynomial in the k-th dimension, when the corresponding When the value is 3, the corresponding original tensor size is 1024, resulting in a large number of parameters and too many input parameter features. Combined with high-order polynomial operations, this leads to an excessively large computational load, making it unsuitable for traditional embedded stress compensation calculation systems. Therefore, this invention further provides a tensor decomposition method to obtain a core tensor and factor matrix with a smaller number of parameters, thereby significantly reducing the number of parameters while maintaining the accuracy of the stress compensation values. It should be noted that the dimension of the original tensor described in this invention is not a limitation of this invention; this invention can also extend the dimension of the stress compensation calculation according to specific circumstances.

[0020] The tensor decomposition method includes: performing Tucker decomposition on the original tensor to obtain the following decomposition formula: ; in This is the core tensor, which can be obtained by randomly generating a tensor much smaller than the original, and then performing multiple rounds of optimization in subsequent processes. Let k denote the factor matrix. During initialization, a smaller matrix can be randomly generated. The superscript k of the factor matrix indicates the k-th dimension of the factor matrix. The multiplication sign and subscript k are symbolic. Let k denote the k-modulus product of the tensor and factor matrices. Substituting the above decomposition into the polynomial of the linear pressure compensation model, we obtain the following pressure compensation function through simplification: in , Let T represent the basis vectors of the polynomial, and T represent the transpose matrix. Therefore, the pressure compensation value in the multidimensional space is equal to the core tensor and the result of gradually shrinking the five low-dimensional vectors to obtain a scalar value. It should be noted that this invention utilizes the aforementioned core tensor and factor matrix, using the factor matrix to project the original high-order polynomial features into a low-dimensional space, while the core tensor describes the high-order interactions between these low-dimensional features.

[0021] Furthermore, the present invention also provides an optimization learning method for the core tensor and the corresponding factor matrix, specifically including: providing a multidimensional parameter set including each detection. And obtain the actual pressure value applied to the pressure sensor. And obtain the raw, uncompensated display value of the pressure sensor. The actual pressure compensation value is obtained by subtracting the original displayed value from the actual pressure value. The multidimensional parameter set and corresponding actual pressure compensation value As a calibration sample set By finding the optimal core tensor sum factor matrix This allows the multidimensional linear pressure compensation model to output the results. Close to the actual pressure compensation value The present invention further utilizes the objective function of minimizing the mean square error to optimize the output of the multidimensional linear pressure compensation model. Close to the actual pressure compensation value The specific formula for the objective function is as follows: ; Where N represents the number of samples tested, It is the Frobenius norm, which is used to control model complexity and prevent overfitting. This is the regularization coefficient. In the above optimization process, this is achieved by adjusting the core tensor... sum factor matrix After multiple rounds of modifications, the compensation model output results are obtained. By performing multiple rounds of learning using the aforementioned objective function, the optimal core tensor is obtained when the objective function is minimized and converges. sum factor matrix .

[0022] Furthermore, since the temperature gradient in this invention requires the acquisition of temperatures at multiple locations within the sensing area corresponding to the pressure sensor, and given the limited packaging space of the pressure sensor, multiple temperature sensors cannot be packaged into a confined space, this invention further provides an alternative solution for the actual measurement of the temperature gradient: A sinusoidal voltage signal of a specific frequency is provided across the bridge circuit of the pressure sensor, allowing the pressure sensor to obtain an AC excitation signal. Since the material's temperature characteristics affect the specific frequency AC excitation signal differently under different temperature gradients, this invention utilizes a thermodynamic model to construct a relationship model between the specific frequency response and the temperature gradient, which is used to replace actual temperature gradient detection. Specifically, the sensor's output phase lag φ is related to the thermal diffusion time constant. The relationship is as follows: , Due to the time constant The temperature gradient itself is related to the length of the heat diffusion path and the heat diffusion coefficient, while the temperature gradient is related to the heat flux density. The heat flux is related to the heat capacity and the rate of temperature change. Therefore, the corresponding temperature gradient can be solved by measuring the phase and amplitude at different frequencies.

[0023] This invention applies sinusoidal voltages of different frequencies to a pressure sensor via voltage scanning, and detects the amplitude and phase of the output signals of the sinusoidal voltages of different frequencies at the same time point, defining a specific frequency... and The detected phases are as follows: and The detected amplitudes are respectively and The amplitude ratio is / Further, the following feature data for phase and amplitude ratios are constructed: , ; in The phase feature data is used as a reference coefficient. Amplitude ratio characteristic data and the corresponding frequency characteristic data f c加入 Seven-dimensional feature data is constructed from the original feature data, wherein the corresponding frequency feature data f c To determine the frequency at which the phase lag reaches -45°, the following extended feature dataset is obtained: ,in , , Furthermore, the extended feature dataset is substituted into the previous linear pressure compensation model, and tensor decomposition is performed using the Tucker decomposition method. The tensor decomposition method has been described above and will not be repeated here.

[0024] It should be noted that the heat conduction equation in the frequency domain is constructed in this invention: ;in For the Fourier transform of the temperature field, in this formula This represents the one-dimensional coordinate value of the pressure sensor in the direction perpendicular to the sensing area, i.e., the thickness direction. To excite the angular frequency of the sinusoidal voltage, The imaginary unit, This indicates that the heat conduction equation introduces phase lag in the frequency domain. is the thermal diffusivity.

[0025] The relationship between the Fourier transform of the pressure sensor output signal and the temperature field distribution is as follows: ,in The Fourier transform of the pressure sensor output signal. The sensitivity coefficient distribution is related to the piezoresistive position, thus allowing us to obtain the amplitude-frequency and phase-frequency characteristics of the output signal in relation to the temperature gradient. Implicit relationships.

[0026] Furthermore, this invention measures the real temperature gradient through experiments and combines it with frequency response characteristics to construct a mapping relationship using, but not limited to, neural networks. The method of constructing the neural network mapping relationship is a mature existing technology, which will not be described in detail in this invention.

[0027] The processes described in the flowcharts above, as disclosed in the embodiments of this invention, can be implemented as computer software programs. Embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), the methods of this application are not limited to the aforementioned functions. It should be noted that the computer-readable medium described above can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wire segments, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical fibers, RF, etc., or any suitable combination thereof.

[0028] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0029] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and explained in the embodiments. Without departing from the stated principles, the implementation of the present invention may have any variations or modifications.

Claims

1. A temperature-compensation-based pressure sensor correction method, characterized by, The method includes: Multiple temperature and pressure data corresponding to the pressure sensor are acquired in a time series, and multidimensional state parameters of the pressure sensor are constructed based on the multiple temperature and pressure data. The multidimensional state parameters are transformed and then input into the linear pressure compensation model to obtain a higher-order pressure compensation model. The coefficients of the higher-order pressure compensation model are obtained to obtain the original tensor data. The original tensor is decomposed using a tensor decomposition algorithm to obtain core tensor data and a factor matrix. The predicted value of compensation pressure is then calculated based on the core tensor data and the factor matrix. A sample set is constructed using the predicted and actual compensation pressure values. During multiple rounds of testing, the error between the predicted and actual compensation pressure values ​​is calculated by adjusting the core tensor data and factor matrix. The core tensor data and factor matrix are then optimized by minimizing the error.

2. The pressure sensor calibration method based on temperature compensation according to claim 1, characterized in that, The multidimensional state parameters of the pressure sensor include the original pressure value, temperature value, temperature change rate, temperature gradient, and pressure change rate. The original pressure value, temperature value, temperature change rate, temperature gradient, and pressure change rate are normalized to obtain corresponding feature data. The feature data are then input into the linear pressure compensation model to obtain the higher-order pressure compensation model.

3. The pressure sensor calibration method based on temperature compensation according to claim 1, characterized in that, The tensor decomposition algorithm includes: using the obtained high-order pressure compensation model coefficients as the original tensor for Tucker decomposition; obtaining core tensor data and factor matrices with parameter sizes smaller than the original tensor through principal component analysis; calculating the corresponding factor matrix for each dimension of the multidimensional state parameters; further calculating the modulus product of the core tensor and each dimension factor matrix to obtain the pressure compensation prediction value based on multidimensional temperature characteristics.

4. The pressure sensor calibration method based on temperature compensation according to claim 1, characterized in that, The actual applied pressure value and the pressure sensor display value are obtained based on the results of multiple rounds of testing. The actual compensated pressure value is calculated using the actual pressure value and the display value. The difference between the actual compensated pressure value and the pressure compensation predicted value is calculated. An objective function is constructed based on the difference between the actual compensated pressure value and the pressure compensation predicted value. The objective function includes minimizing the mean square error between the actual compensated pressure value and the pressure compensation predicted value. The core tensor data and factor matrix are optimized based on the objective function.

5. The pressure sensor calibration method based on temperature compensation according to claim 1, characterized in that, A sinusoidal voltage signal of a specific frequency is generated in the bridge circuit of the pressure sensor. The phase and amplitude of the sinusoidal voltage signal are acquired, and the phase and amplitude ratios of different frequencies at the same detection time are calculated. Feature data is constructed based on the phase, amplitude ratio, and frequency. The feature data of phase, amplitude ratio, and frequency are integrated into the feature data of the original pressure value, temperature value, temperature change rate, and pressure change rate to construct seven-dimensional feature data. Tensor decomposition is performed on the seven-dimensional feature data to obtain the corresponding core tensor data and factor matrix, and the core tensor data and factor matrix are optimized.

6. The pressure sensor calibration method based on temperature compensation according to claim 5, characterized in that, A heat conduction model of the pressure sensor is constructed to represent the temperature gradient of the pressure sensor in the thickness direction of the sensing area. The relationship of the heat conduction model in the frequency domain is established by using Fourier transform. Furthermore, the mapping relationship between the specific frequency response of the pressure sensor and the temperature gradient is constructed by using detection calibration. The frequency response is used to replace the actual detected temperature gradient.

7. A pressure sensor calibration method based on temperature compensation according to claim 6, characterized in that, The detection calibration method includes: using multiple temperature sensors to detect the true temperature gradient of the pressure sensor, and obtaining the phase and amplitude ratio of the pressure sensor bridge at a specific frequency. Based on the phase and amplitude ratio, feature transformation is performed and used as mapping parameters to construct a mapping relationship between frequency and amplitude. The mapping relationship is established through a neural network.

8. A pressure sensor calibration system based on temperature compensation, characterized in that, The system performs a pressure sensor calibration method based on temperature compensation as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement a temperature-compensated pressure sensor calibration method according to any one of claims 1-7.

Images