Compensation method of silicon resonance pressure sensor

Abstract

The invention provides a compensation method for a silicon resonance pressure sensor, and relates to the technical field of pressure measurement, and the method comprises the steps: collecting a frequency signal time sequence outputted by the silicon resonance pressure sensor, and carrying out the analysis to obtain a real-time pressure value; acquiring a predetermined compensation item, and performing multi-dimensional compensation feature collection to obtain multi-dimensional compensation information; performing traversal screening on the historical compensation database by taking the multi-dimensional compensation information as a screening constraint to obtain a target sensor; performing normalized comparison on the detected pressure value of the target sensor and a preset pressure value to obtain a pressure value deviation, and obtaining a multi-dimensional compensation coefficient; and performing compensation adjustment to obtain an effective pressure value. The technical problems that most compensation methods in the prior art are adjusted only for a single influence factor, the comprehensive influence of multi-dimensional factors is ignored, multiple error sources under the complex environment condition cannot be effectively handled, and then the precision and reliability of the silicon resonance pressure sensor are insufficient are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of pressure measurement, and in particular to a compensation method for a silicon resonant pressure sensor. Background Art

[0002] The silicon resonant pressure sensor is a miniaturized, low-power, high-precision pressure sensor specially developed for the meteorological field. Its core principle is to measure pressure by changing the resonant frequency of the silicon material with the pressure. However, since the silicon resonant pressure sensor may be affected by factors such as temperature, material aging, process variation, and environmental changes during use, there is a certain error in its output signal. Therefore, how to accurately compensate for the impact of these factors on the performance of the pressure sensor has become a key issue in improving its accuracy and reliability.

[0003] Existing silicon resonant pressure sensor compensation methods mainly solve sensor error problems through calibration and simplified compensation algorithms, but these methods usually ignore the combined impact of multi-dimensional factors. Most traditional compensation methods only adjust a single factor (such as temperature or pressure range) and cannot effectively deal with multiple error sources under complex environmental conditions, resulting in insufficient accuracy and reliability of silicon resonant pressure sensors. Summary of the invention

[0004] The present application provides a compensation method for a silicon resonant pressure sensor, aiming to solve the technical problem that most compensation methods in the prior art only adjust a single factor, ignoring the combined influence of multi-dimensional factors, resulting in the inability to effectively deal with multiple error sources under complex environmental conditions, thereby causing insufficient accuracy and reliability of the silicon resonant pressure sensor.

[0005] The present application discloses a compensation method for a silicon resonant pressure sensor, the method comprising: acquiring a frequency signal timing sequence output by the silicon resonant pressure sensor, and analyzing the frequency signal timing sequence to obtain a real-time pressure value at a real-time moment; obtaining a predetermined compensation item, and collecting multi-dimensional compensation features of the silicon resonant pressure sensor based on the predetermined compensation item to obtain multi-dimensional compensation information; traversing and screening a historical compensation database using the multi-dimensional compensation information as a screening constraint to obtain a target sensor; normalizing and comparing the pressure value deviation between the detected pressure value of the target sensor and the preset pressure value to obtain a multi-dimensional compensation coefficient; and compensating and adjusting the real-time pressure value using the multi-dimensional compensation coefficient as a weight to obtain an effective pressure value of the silicon resonant pressure sensor at the real-time moment.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By collecting the frequency signal timing in real time and analyzing it, the pressure value of the current sensor can be accurately obtained, which provides basic data for subsequent compensation and enables the system to respond quickly to pressure changes. By collecting multi-dimensional compensation features of the sensor based on predetermined compensation items, various factors affecting the performance of the sensor can be fully considered, which effectively integrates the performance of the sensor under different conditions, helps to compensate for errors caused by various environments or manufacturing processes, and provides more accurate compensation data. By screening the target sensor in the historical compensation database, the most matching compensation reference can be found based on the existing data. This process ensures that the known data of the historical sensor can be used to provide the current sensor with Compensation reference avoids complex modeling from scratch and improves the efficiency of sensor compensation; by comparing the detected pressure value of the target sensor with the preset pressure value, the pressure deviation can be accurately calculated, and normalized to obtain the multi-dimensional compensation coefficient, eliminating the deviation caused by sensor error, further improving the reliability of compensation, and ensuring the accuracy of the measurement results; using the multi-dimensional compensation coefficient as the weight, the real-time pressure value is adjusted to finally obtain the effective pressure value. This compensation adjustment can dynamically correct the sensor during the real-time measurement process, effectively eliminating various factors that affect the sensor measurement results, making the final effective pressure value more accurate and adaptable to changes in different environments.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of a compensation method for a silicon resonant pressure sensor is provided for an embodiment of the present application.

[0010] Figure 2 A schematic diagram of a real-time pressure value verification process in a compensation method of a silicon resonant pressure sensor is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0011] The embodiment of the present application provides a compensation method for a silicon resonant pressure sensor, which solves the technical problem that most compensation methods in the prior art only adjust a single factor and ignore the combined influence of multi-dimensional factors, resulting in the inability to effectively deal with multiple error sources under complex environmental conditions, thereby causing the silicon resonant pressure sensor to have insufficient accuracy and reliability.

[0012] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0013] like Figure 1 As shown, an embodiment of the present application provides a compensation method for a silicon resonant pressure sensor, the method comprising:

[0014] The frequency signal timing of the silicon resonant pressure sensor output is collected and obtained, and the frequency signal timing is analyzed to obtain the real-time pressure value at the real time.

[0015] The silicon resonant pressure sensor is a miniaturized, low-power, high-precision pressure sensor specially developed for the meteorological field. It uses a high-speed data acquisition card or a dedicated interface device to collect the frequency output of the silicon resonant pressure sensor in real time. The working principle of the silicon resonant pressure sensor is that the resonant frequency of the resonant beam changes according to the pressure change. The frequency change has a certain relationship with the pressure. The acquisition process should ensure the high precision and high frequency of the signal to avoid errors caused by too low sampling rate. The collected frequency signal sequence is analyzed in the time domain to extract relevant features of the signal, such as the maximum value, minimum value, average value, standard deviation and other statistics. The time domain features are used to preliminarily judge the change trend of the signal. Through the known mathematical relationship between frequency and pressure, the collected frequency signal is converted into a real-time pressure value. The linear function between pressure and frequency can be used for calculation, that is, the frequency change is in linear proportion to the pressure change, and the real-time pressure value is directly calculated based on the frequency deviation.

[0016] A predetermined compensation item is obtained, and multi-dimensional compensation characteristics of the silicon resonant pressure sensor are collected based on the predetermined compensation item to obtain multi-dimensional compensation information.

[0017] Obtain the predetermined compensation items, which are set according to the factors that affect the monitoring accuracy of the silicon resonant pressure sensor, including design rationality compensation, material property compensation, process adaptability compensation and environmental suitability compensation. Collect relevant compensation features for each compensation item. Specifically, for design rationality compensation, consider the modal coupling effect of the resonant beam and collect frequency response characteristics under different design schemes; for material property compensation, consider the effect of temperature change on the Young's modulus of silicon material and collect frequency data at different temperatures; for process adaptability compensation, collect the effect of errors occurring during the manufacturing process on the frequency signal, especially the changes in geometry or surface roughness occurring during micromachining; for environmental suitability compensation, record the changes in frequency signals under different environmental conditions, especially when temperature or humidity changes. Integrate the characteristic information of each compensation item collected to obtain multi-dimensional compensation information. This information set can describe in detail the various factors that affect the accuracy of the silicon resonant pressure sensor and provide support for subsequent compensation calculations.

[0018] The historical compensation database is traversed and screened using the multi-dimensional compensation information as a screening constraint to obtain a target sensor.

[0019] The historical compensation database contains compensation records of multiple known sensors, and each compensation record contains compensation data of multiple dimensions of the sensor, which can describe the compensation effect in different environments or usage scenarios. The multi-dimensional compensation information extracted from the current sensor is used as a constraint condition to filter the data in the historical compensation database. By calculating the similarity between the current multi-dimensional compensation information and each compensation data in the historical database, it is evaluated which historical compensation data is most suitable for the current sensor. For example, the similarity of two compensation feature sets is measured using methods such as Euclidean distance or cosine similarity. Through screening, the target sensor that best matches the current sensor is found. The historical compensation data of this target sensor can provide an effective compensation solution for the current sensor.

[0020] The pressure value deviation obtained by normalizing and comparing the pressure value detected by the target sensor and the preset pressure value is obtained to obtain a multi-dimensional compensation coefficient.

[0021] The historical compensation data of the target sensor contains the relationship with the corresponding pressure value. In this step, the actual detected pressure value of the target sensor in the historical data is extracted; the preset pressure value is a standardized reference pressure value, which represents the pressure value under ideal conditions. The detected pressure value of the target sensor is compared with the preset pressure value, and the deviation between them is calculated.

[0022] In order to eliminate the influence of different pressure ranges, units or dimensions, the deviation is normalized. For example, the minimum and maximum normalization method is used to compress the deviation to a fixed range, usually between 0 and 1, by subtracting the minimum value and dividing by the range. Through comparative analysis and normalization process, the multidimensional compensation coefficient is finally obtained, which is the adjustment factor for multiple compensation items (such as design, process, temperature, etc.).

[0023] The real-time pressure value is compensated and adjusted by taking the multi-dimensional compensation coefficient as a weight to obtain the effective pressure value of the silicon resonant pressure sensor at the real-time moment.

[0024] The real-time pressure value is calculated based on the output frequency signal of the silicon resonant pressure sensor. This pressure value may have deviations due to factors such as sensor design and environmental influence. The real-time pressure value is compensated and adjusted using the obtained multi-dimensional compensation coefficient. The specific operation is to apply the compensation coefficient to the calculation formula of the real-time pressure value, that is, to combine the compensation coefficient with the real-time pressure value in a weighted manner to calculate the final effective pressure value. For example, assuming that a certain compensation coefficient is 0.1, then the real-time pressure value is multiplied by (1+0.1) to achieve adjustment. The pressure value after compensation is the effective pressure value of the silicon resonant pressure sensor at the current moment. This value can provide more accurate pressure measurement results by correcting multiple factors such as design defects, material properties, and process influences.

[0025] Furthermore, if Figure 2 As shown, after acquiring the frequency signal sequence output by the silicon resonant pressure sensor and analyzing the frequency signal sequence to obtain the real-time pressure value at the real time, it also includes:

[0026] The time domain characteristic parameters of the frequency signal time series are collected; the frequency signal time series is subjected to Fourier transform to obtain a frequency signal spectrum; the frequency domain characteristic parameters of the frequency signal spectrum are collected; the frequency signal time series is subjected to modal decomposition to obtain a modal component set, wherein the modal component set includes a target component; the component characteristic parameters of the target component are collected, and a frequency parameter set is formed in coordination with the time domain characteristic parameters and the frequency domain characteristic parameters; the frequency parameter set is used as processing data of a pressure processing predictor, and a predicted pressure value is obtained through the pressure processing predictor; the real-time pressure value is verified through the predicted pressure value.

[0027] Perform time domain analysis on the frequency signal time series. Time domain analysis is the analysis of the changes in frequency signals over time, including analysis of the signal's amplitude, frequency, waveform, etc. Time domain analysis helps to understand the basic change trend of the signal. Through time domain analysis, time domain characteristic parameters are obtained, including: peak value, that is, the maximum amplitude value of the signal, reflecting the great change of the signal over time; mean, that is, the average value of the signal, representing the basic level of the signal in the time series; standard deviation, reflecting the signal volatility and degree of change; time domain period, that is, the information of the signal periodicity, which is related to the resonant frequency; maximum zero crossing point, that is, the number of zero crossing points in the change of the frequency signal, which can help determine the periodicity and stability of the signal.

[0028] Fourier transform is to convert time domain signal into frequency domain signal. Through Fourier transform, the original time series signal is decomposed into multiple different frequency components for analyzing the frequency characteristics of the signal. The collected frequency signal time series is subjected to Fourier transform. The result of Fourier transform is expressed as a spectrum, that is, the relationship between frequency and amplitude, which reveals the intensity distribution of the signal at each frequency. In practical applications, it is possible to identify which frequency components have an important influence on the response of the sensor, especially those related to the resonant frequency.

[0029] Frequency domain characteristic parameters are extracted from the frequency signal spectrum, including: main frequency, that is, the frequency component with the largest amplitude in the spectrum, which is usually directly related to the resonant frequency of the sensor; bandwidth, that is, the width of the signal frequency range, which reflects the frequency distribution range of the signal; frequency distribution, that is, the energy distribution of different frequency components, which helps to identify noise or interference frequency bands; harmonic components. In the spectrum, if there are multiple frequency components that are integer multiples of the fundamental frequency, these components are called harmonic components. By analyzing the harmonic components, it can be identified whether the signal is affected by nonlinear factors. Frequency domain characteristic parameters are used to accurately identify the resonant frequency, and can further analyze the frequency drift caused by temperature changes, material aging or process defects. By continuously monitoring these characteristics, the stability and accuracy of silicon resonant pressure sensors can be improved.

[0030] Modal decomposition is the process of decomposing a frequency signal into multiple modal components. Each mode represents a specific mode in the signal, such as different vibration modes or resonance modes. This method helps to identify and eliminate noise or irrelevant components in the signal. Modal decomposition methods include empirical mode decomposition, Hilbert-Huang transform, etc. Through these methods, different modal components can be extracted from the frequency signal.

[0031] After modal decomposition, a modal component set consisting of multiple modal components is obtained. Each modal component represents a vibration mode or resonance mode in the signal. In the modal component set, the component that has the greatest impact on the current sensor performance is selected as the target component. The target component is the modal component that is most relevant to the resonant frequency of the sensor. By selecting the target component, the key frequency components that affect the sensor performance can be more accurately extracted. The target component can reflect the main response of the sensor under specific conditions, thereby providing a basis for subsequent pressure calculation or compensation.

[0032] Extract the component characteristic parameters of the target component, including the amplitude, phase, frequency, and frequency offset of the target component. Combine the time domain characteristic parameters (such as mean, standard deviation, peak value, etc.) and frequency domain characteristic parameters (such as main frequency, bandwidth, harmonics, etc.) with the component characteristic parameters of the target component (such as amplitude, phase, etc.) to form a complete frequency parameter set. These characteristic parameters can reflect different levels of sensor response and provide comprehensive information for pressure prediction.

[0033] The pressure processing predictor is a machine learning model used to predict the pressure value of the sensor based on the input frequency parameter set. The predictor is obtained through supervised learning training. The specific training process is detailed in the subsequent steps. The obtained frequency parameter set is passed to the predictor as input data. Based on the frequency parameter set, the predictor predicts the relationship between the frequency parameter and the pressure value in the training data and outputs the predicted pressure value.

[0034] The real-time pressure value is verified by the predicted pressure value. Specifically, the predicted pressure value is compared with the real-time pressure value, the error between the two is calculated, and a tolerance range is set. If the error is less than the predetermined tolerance range, the real-time pressure value is considered to be accurate; if the error is large, further compensation of the real-time pressure value is required.

[0035] Furthermore, modal decomposition is performed on the frequency signal time series to obtain a modal component set, wherein the modal component set includes a target component, including:

[0036] Extract any component from the modal component set; calculate and obtain any information amount ratio between the arbitrary component and the frequency signal time series; arrange the arbitrary components in descending order based on the arbitrary information amount ratio to obtain a modal component descending list; take the first component in the modal component descending list as the target component.

[0037] Extract any component from the modal component set. Any component refers to any modal component randomly extracted from the component set as the current analysis object.

[0038] In signal processing, information volume is used to measure the contribution of a signal or a component to the overall signal. Information volume can be measured based on factors such as the energy distribution of the spectrum and the amplitude change. For any extracted modal component, the information volume can be calculated based on the following formula: ,in, The signal representing the modal component at time t, Characterize the amount of information of the modal component. Calculate the ratio of the amount of information of any modal component to the amount of information of the entire frequency signal to evaluate the importance of the modal component in the signal. The larger the information ratio, the greater the impact of the modal component on the signal and the higher its importance in pressure measurement.

[0039] According to the calculated arbitrary information ratio, the modal components are sorted. The component with a larger information ratio indicates a greater contribution to the signal, so it should be placed at the front during sorting. After sorting, a descending list of modal components is obtained. Each modal component in this list is sorted according to its contribution to the signal.

[0040] In the descending modal component list, the first component is the component with the largest information ratio, which means it is the component that has the most significant impact on the signal. The first component is taken as the target component, which represents the strongest vibration mode or resonance mode in the signal and is usually the main factor in the sensor response. Therefore, selecting this component as the target component helps to accurately describe the working state of the sensor and provide a reliable basis for subsequent pressure measurement and compensation.

[0041] Furthermore, using the frequency parameter set as processing data of a pressure processing predictor, and obtaining a predicted pressure value through the pressure processing predictor, comprises:

[0042] The historical frequency signal timing of the silicon resonant pressure sensor in a predetermined historical stage is obtained, and the terminal historical moment in the predetermined historical period corresponds to the historical pressure value; a historical frequency parameter set of the historical frequency signal timing is formed, and a training data set is formed with the historical pressure value; supervised learning is performed on the first data set in the training data set to obtain an initial predictor, and the initial predictor is verified by the second data set in the training data set; if the verification passes, the initial predictor is used as the pressure processing predictor.

[0043] Collect the historical frequency signal time series data of the silicon resonant pressure sensor in a predetermined historical period. The predetermined historical period is a specified time range involving long-term sampling data. The historical period is selected to ensure that the training of the prediction model has sufficient data support. Record the historical pressure value at the end of the predetermined historical period and pair it with the frequency signal data as label data. This historical pressure value and the corresponding frequency signal time series data together constitute a complete sample data.

[0044] A historical frequency parameter set is extracted from the acquired historical frequency signal time series, and the historical frequency parameter set corresponds to the frequency parameter set of the current frequency signal time series. The extracted historical frequency parameter set is paired with the corresponding historical pressure value to form a training data set, in which each sample consists of a frequency parameter set and a corresponding historical pressure value.

[0045] A part of the training data set is randomly selected as the first data set, and the remaining part is selected as the second data set. The ratio of the first data set to the second data set is usually 8:2. The first data set is used for supervised learning. Common supervised learning algorithms include linear regression, support vector machine, neural network, etc. Through these algorithms, the model is trained to enable it to predict historical pressure values ​​based on the frequency parameter set. After the training is completed, the initial predictor is obtained; the initial predictor is verified by the second data set. The purpose of the verification is to evaluate the performance of the predictor on new data and avoid model overfitting. Common evaluation indicators include mean square error, etc. According to the verification results, the performance of the initial predictor is evaluated. If the verification error is small, it means that the model has good prediction ability. If the verification error is large, it is necessary to adjust the model parameters or select other algorithms.

[0046] If the verification result of the second data set meets the preset performance requirements, for example, the error is less than the preset error threshold, it is considered that the initial predictor has good generalization ability and can make accurate pressure predictions on new data. The verified initial predictor is used as the formal pressure processing predictor, which can be used to process real-time frequency signal data and output real-time pressure values.

[0047] Furthermore, the predetermined compensation items include design rationality compensation, material property compensation, process adaptability compensation and environmental suitability compensation.

[0048] The predetermined compensation items are set according to the factors that affect the monitoring accuracy of the silicon resonant pressure sensor, including design rationality compensation, material property compensation, process adaptability compensation and environmental suitability compensation. Among them, the unreasonable factors that may exist in the sensor design process are compensated, such as asymmetric structure, design defects, etc. These factors will affect the measurement accuracy of the pressure sensor. Therefore, the design rationality compensation aims to make corresponding adjustments according to the design defects; due to the nonlinear characteristics, temperature sensitivity, elastic modulus changes, etc. of silicon materials, the material itself may cause performance deviations of the sensor, and the material property compensation aims to correct the sensor output according to these changes; minor errors in the manufacturing process (such as silicon bonding, surface treatment, etc.) may cause performance deviations of the sensor, and the process adaptability compensation is adjusted for these manufacturing errors to ensure the stability and accuracy of the sensor; the performance of the sensor may be affected by the external environment, such as temperature, humidity, electromagnetic interference, etc., and the environmental suitability compensation aims to cope with these environmental changes and maintain the accuracy and stability of the sensor in different environments.

[0049] These compensation items are common influencing factors when designing and manufacturing sensors, and can significantly affect the performance of silicon resonant pressure sensors. By introducing these predetermined compensation items, the output of the sensor can be corrected more accurately, allowing it to maintain high accuracy under different working conditions.

[0050] Furthermore, the historical compensation database is traversed and screened using the multi-dimensional compensation information as a screening constraint to obtain a target sensor, including:

[0051] Extracting a first historical data group from the historical compensation database, wherein the first historical data group includes first multidimensional compensation information of a first sensor; calculating a first similarity between the multidimensional compensation information and the first multidimensional compensation information; and if the first similarity reaches a predetermined similarity limit, using the first sensor as the target sensor.

[0052] The historical compensation database contains historical compensation records of multiple sensors, each record includes multi-dimensional compensation information of the sensor, and the information involves different types of compensation items, including design rationality compensation, process adaptability compensation, material property compensation, etc. A first historical data group is randomly extracted from the historical compensation database as the current analysis object, and the first historical data group includes first multi-dimensional compensation information of the first sensor.

[0053] By comparing the multidimensional compensation information of the current sensor with the first multidimensional compensation information in the first historical data group, a similarity value is calculated. The specific calculation process is carried out in subsequent steps. The calculation result is a first similarity value, which indicates the degree of similarity between the multidimensional compensation information of the current sensor and the historical sensor. The higher the similarity value, the more similar the two sensors are in compensation characteristics.

[0054] A predetermined similarity limit is set. If the similarity value exceeds this limit, it is considered that the compensation information of the current sensor is very similar to that of the historical sensor. The similarity limit is selected based on experience or is set by an optimization method (such as cross-validation). For example, it is set to 0.8. A similarity value exceeding 0.8 indicates that the two are very similar.

[0055] If the calculated first similarity exceeds the predetermined similarity limit, the historical sensor is used as the target sensor, and the compensation information of the target sensor can be used for the compensation of the current sensor. By selecting the target sensor that best matches the current sensor characteristics, the compensation characteristics in the historical data can be used to perform pressure compensation and error correction for the current sensor.

[0056] Furthermore, calculating a first similarity between the multi-dimensional compensation information and the first multi-dimensional compensation information includes:

[0057] Extract any item from the predetermined compensation items; match the arbitrary item in the multidimensional compensation information and the first multidimensional compensation information in turn to obtain an arbitrary compensation coefficient and a first arbitrary compensation coefficient respectively; calculate an arbitrary coefficient difference between the arbitrary compensation coefficient and the first arbitrary compensation coefficient; read a similar decision plan, and obtain a target coefficient difference in combination with the arbitrary coefficient difference; record the target coefficient difference as the first similarity.

[0058] The predetermined compensation items include multiple compensation factors, including design rationality compensation, material property compensation, process adaptability compensation, and environmental suitability compensation. A specific compensation item is randomly extracted from these predetermined compensation items as an item to be processed.

[0059] Each compensation item corresponds to a specific compensation coefficient. Based on the extracted arbitrary item, the multidimensional compensation information of the current sensor is matched with the first multidimensional compensation information of the historical sensor item by item, and the arbitrary compensation coefficient and the first arbitrary compensation coefficient are obtained through matching.

[0060] The difference between the arbitrary compensation coefficient and the first arbitrary compensation coefficient is calculated to obtain the arbitrary coefficient difference, which represents the difference between the two sensors in the compensation item. The larger the compensation coefficient difference, the greater the difference between the two sensors in the compensation item. This may be due to differences in sensor design, materials, processes, etc.

[0061] The similarity decision plan is a preset table that contains rules for taking different decisions based on different coefficient differences. It helps to determine how to adjust the compensation coefficient of the sensor based on the compensation coefficient difference. By reading the similarity decision plan, the calculated arbitrary coefficient difference is combined with the rules in the plan to obtain the target coefficient difference. The target coefficient difference reflects the difference in compensation items between the current sensor and the historical sensor.

[0062] The calculated target coefficient difference is used as the first similarity. This similarity value reflects the similarity between the current sensor and the historical sensor in compensation. The higher the similarity, the closer the compensation characteristics of the two are.

[0063] Further, similar decision plans are read and the target coefficient difference is obtained by combining the arbitrary coefficient difference, including:

[0064] Obtain a predetermined table, wherein the predetermined table includes 4 rows and 4 columns with a total of 16 cells; according to the similarity decision plan, fill the arbitrary coefficient difference into the first area of ​​the predetermined table; extract any cell in the second area of ​​the predetermined table, and calculate any initial filling value of the arbitrary cell; form an arbitrary reference cell set of the arbitrary cell, and compare to obtain a target cell in the arbitrary reference cell set; take the sum of the target filling value corresponding to the target cell and the arbitrary initial filling value, and update it to the arbitrary cell to form a table filling result; take the filling value of the predetermined cell in the table filling result as the target coefficient difference.

[0065] Get the reservation table. The reservation table is a two-dimensional table structure, which contains 4 rows and 4 columns with a total of 16 cells. This table is used to store the adjustment values ​​of the compensation coefficient differences. Each cell represents a specific compensation adjustment value. These adjustment values ​​are filled according to similar decision plans and compensation coefficient differences. The size of the table is 4x4, which is suitable for processing the relationship between multiple compensation items.

[0066] The similarity decision plan provides a guideline for determining how to adjust the compensation coefficient of the current sensor based on the compensation coefficient difference. The plan contains different rules, such as "if the coefficient difference is a certain value, fill a certain adjustment value in a specific position in the table", which is used to map the actual calculated compensation adjustment coefficient to the table for subsequent adjustment. The first area of ​​the predetermined table refers to a sub-area of ​​the table, such as the first row and first column of the table. According to the rules in the decision plan, any coefficient difference is filled into the cells of the first area.

[0067] The second area refers to another area in the table that is different from the first area. This area corresponds to other compensation items related to the difference in compensation coefficients or further refined decision rules. An arbitrary cell is selected from the second area for calculation. This cell corresponds to the adjustment value of a specific compensation item between the current sensor and the historical sensor.

[0068] According to the specific decision rules of the second area, the initial fill value of the cell is calculated. The initial fill value is usually determined based on the compensation coefficient difference, the similarity value and the rules specified in the decision plan. For example, assuming that the decision plan stipulates that if a certain compensation coefficient difference is large, the fill value of the cell will be higher, indicating that a larger compensation adjustment is required. The initial fill value indicates the compensation amplitude required for the current sensor under a specific compensation item. This value will be adjusted according to the working status and compensation requirements of the sensor.

[0069] The reference cell set is composed of a group of cells associated with the current cell, which represent similar compensation items, or areas in the table with similar characteristics to the current compensation item, including the left neighboring cell, the upper neighboring cell, and the upper left neighboring cell of the arbitrary cell. By comparing the cells in the reference cell set, the target cell that best matches the current cell is determined. The target cell is the compensation value that is most similar to the current compensation item. The target cell provides a reference fill value for determining the final adjustment value of the current cell.

[0070] Extract the target fill value corresponding to the target cell. This fill value represents the compensation adjustment value under the historical sensor or ideal state; any initial fill value reflects the initial adjustment of the current sensor under a specific compensation item. Add the target fill value to the any initial fill value to obtain the updated fill value. This process reflects that the current sensor needs to make certain adjustments based on the reference value of the target cell. Through the summation process, the current arbitrary cell in the table is updated to form a new fill value. This step makes the compensation coefficient adjustment of the current sensor closer to the compensation effect of the historical sensor.

[0071] The predetermined cell is a specific cell in the table. The fill value of this cell will eventually determine the target coefficient difference. It can be a cell related to the main compensation item of the current sensor, or a cell that needs to be focused on during the adjustment process. The fill value of the predetermined cell is extracted from the table filling result. This value indicates the final adjustment amount required for the current sensor under the specific compensation item after adjustment. The extracted fill value is used as the target coefficient difference. The target coefficient difference reflects the difference in compensation items between the current sensor and the historical sensor. The pressure measurement of the sensor can be adjusted through this difference.

[0072] Furthermore, the arbitrary reference cell set includes the left neighboring cell, the upper neighboring cell and the upper left neighboring cell of the arbitrary cell.

[0073] Any reference cell set includes the left neighbor cell, the upper neighbor cell and the upper-left neighbor cell of the arbitrary cell, wherein the left neighbor cell is the neighbor cell on the left side of the current arbitrary cell, the upper neighbor cell is the neighbor cell above the current arbitrary cell, and the upper-left neighbor cell is the cell at the upper left corner of the current arbitrary cell. By constructing this reference cell set, the information of the surrounding cells can be used to improve the accuracy of the current cell adjustment. The surrounding cells may have taken into account other similar compensation factors, thereby better providing a reference value for compensation adjustment for the current cell.

[0074] In summary, the compensation method of a silicon resonant pressure sensor provided in the embodiment of the present application has the following technical effects:

[0075] By collecting the frequency signal timing in real time and analyzing it, the pressure value of the current sensor can be accurately obtained, which provides basic data for subsequent compensation and enables the system to respond quickly to pressure changes. By collecting multi-dimensional compensation features of the sensor based on predetermined compensation items, various factors affecting the performance of the sensor can be fully considered, which effectively integrates the performance of the sensor under different conditions, helps to compensate for errors caused by various environments or manufacturing processes, and provides more accurate compensation data. By screening the target sensor in the historical compensation database, the most matching compensation reference can be found based on the existing data. This process ensures that the known data of the historical sensor can be used to provide the current sensor with Compensation reference avoids complex modeling from scratch and improves the efficiency of sensor compensation; by comparing the detected pressure value of the target sensor with the preset pressure value, the pressure deviation can be accurately calculated, and normalized to obtain the multi-dimensional compensation coefficient, eliminating the deviation caused by sensor error, further improving the reliability of compensation, and ensuring the accuracy of the measurement results; using the multi-dimensional compensation coefficient as the weight, the real-time pressure value is adjusted to finally obtain the effective pressure value. This compensation adjustment can dynamically correct the sensor during the real-time measurement process, effectively eliminating various factors that affect the sensor measurement results, making the final effective pressure value more accurate and adaptable to changes in different environments.

[0076] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A compensation method for a silicon resonant pressure sensor, characterized in that: The method comprises: Collecting the frequency signal timing of the silicon resonant pressure sensor output, and analyzing the frequency signal timing to obtain the real-time pressure value at the real time; Acquire a predetermined compensation item, and collect multi-dimensional compensation features of the silicon resonant pressure sensor based on the predetermined compensation item to obtain multi-dimensional compensation information; Using the multi-dimensional compensation information as a screening constraint, a historical compensation database is traversed and screened to obtain a target sensor; Normalizing and comparing the pressure value deviation obtained by the pressure value detected by the target sensor and the preset pressure value to obtain a multi-dimensional compensation coefficient; The real-time pressure value is compensated and adjusted by taking the multi-dimensional compensation coefficient as a weight to obtain the effective pressure value of the silicon resonant pressure sensor at the real-time moment.

2. The method according to claim 1, characterized in that: After acquiring the frequency signal sequence output by the silicon resonant pressure sensor and analyzing the frequency signal sequence to obtain the real-time pressure value at the real time, the method further includes: Acquiring time domain characteristic parameters of the frequency signal time series; Performing Fourier transform on the frequency signal time series to obtain a frequency signal spectrum; Acquiring frequency domain characteristic parameters of the frequency signal spectrum; Performing modal decomposition on the frequency signal time series to obtain a modal component set, wherein the modal component set includes a target component; Collecting component characteristic parameters of the target component, and forming a frequency parameter set in coordination with the time domain characteristic parameters and the frequency domain characteristic parameters; Using the frequency parameter set as processing data of a pressure processing predictor, and obtaining a predicted pressure value through the pressure processing predictor; The real-time pressure value is verified by the predicted pressure value.

3. The method according to claim 2, characterized in that: Performing modal decomposition on the frequency signal time series to obtain a modal component set, wherein the modal component set includes a target component, including: Extracting any component from the modal component set; Calculate and obtain the ratio of any information amount of the arbitrary component to the frequency signal time series; Arranging the arbitrary components in descending order based on the arbitrary information amount ratio to obtain a descending list of modal components; The first component in the descending list of modal components is taken as the target component.

4. The method according to claim 2, characterized in that: Using the frequency parameter set as processing data of a pressure processing predictor, and obtaining a predicted pressure value through the pressure processing predictor, comprises: Acquire a historical frequency signal time sequence of the silicon resonant pressure sensor in a predetermined historical stage, and the terminal historical moment in the predetermined historical period corresponds to a historical pressure value; assembling a historical frequency parameter set of the historical frequency signal time series and forming a training data set with the historical pressure value; Performing supervised learning on a first data set in the training data set to obtain an initial predictor, and verifying the initial predictor using a second data set in the training data set; If the verification is successful, the initial predictor is used as the pressure treatment predictor.

5. The method according to claim 1, characterized in that: The predetermined compensation items include design rationality compensation, material property compensation, process adaptability compensation and environmental suitability compensation.

6. The method according to claim 5, characterized in that: The historical compensation database is traversed and screened using the multi-dimensional compensation information as a screening constraint to obtain a target sensor, including: Extracting a first historical data group from the historical compensation database, wherein the first historical data group includes first multi-dimensional compensation information of a first sensor; Calculating and obtaining a first similarity between the multi-dimensional compensation information and the first multi-dimensional compensation information; If the first similarity reaches a predetermined similarity limit, the first sensor is used as the target sensor.

7. The method according to claim 6, characterized in that: Calculating a first similarity between the multidimensional compensation information and the first multidimensional compensation information includes: extracting any item from the predetermined compensation items; sequentially matching the arbitrary items in the multi-dimensional compensation information and the first multi-dimensional compensation information to obtain an arbitrary compensation coefficient and a first arbitrary compensation coefficient respectively; Calculating an arbitrary coefficient difference between the arbitrary compensation coefficient and the first arbitrary compensation coefficient; Read similar decision plans, and combine the arbitrary coefficient difference to obtain a target coefficient difference; The target coefficient difference is recorded as the first similarity.

8. The method according to claim 7, characterized in that: Read similar decision plans and combine the arbitrary coefficient difference to obtain the target coefficient difference, including: Obtaining a predetermined table, wherein the predetermined table includes 4 rows and 4 columns with a total of 16 cells; According to the similar decision plan, filling the arbitrary coefficient difference into the first area of ​​the predetermined table; Extracting any cell in the second area of ​​the predetermined table, and calculating any initial filling value of the any cell; Forming an arbitrary reference cell set of the arbitrary cell, and obtaining a target cell in the arbitrary reference cell set by comparison; Taking the sum of the target filling value corresponding to the target cell and the arbitrary initial filling value, and updating it to the arbitrary cell to form a table filling result; The filling value of a predetermined cell in the table filling result is taken as the target coefficient difference.

9. The method according to claim 8, characterized in that: The arbitrary reference cell set includes the left neighboring cell, the upper neighboring cell and the upper left neighboring cell of the arbitrary cell.

Images