Low-temperature environment LNG (Liquefied Natural Gas) metering sensor drift compensation method and system

Abstract

The invention discloses a low-temperature environment LNG (Liquefied Natural Gas) metering sensor drift compensation method and system, and particularly relates to the technical field of sensor drift compensation. Output data of a sensor are collected and processed, and a temperature compensation model of the sensor is established; in the working temperature range, digital filtering processing is carried out on data of displacement change of each measuring range point, and then analysis and mining are carried out; a data longitudinal interval in a temperature range is refined by using a table look-up method, so that the data interval is reduced. And calculating the G value according to the residual error obtained by the table look-up method and the numerical value change condition in each temperature section. If the G value is greater than a preset threshold value of the system, selecting a proper fitting method, such as a least square method or Legendre orthogonal polynomial fitting, according to the distribution of the G value and the data interval when curve fitting is used, so as to further optimize the temperature compensation model; and if the G value is smaller than the system preset threshold value, no subsequent operation is carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of sensor drift compensation, and more specifically, to a drift compensation method and system for an LNG metering sensor in a low-temperature environment. Background Art

[0002] During storage and transportation, liquefied natural gas is usually kept in an ultra-low temperature environment below -162°C, which places extremely high demands on the stability and accuracy of sensors. Drift in sensors is a relatively complex and unavoidable phenomenon. There are many factors that cause drift, such as sensor aging and poisoning, changes in temperature and humidity in the environment, and delays in data transmission.

[0003] Sensor drift refers to the phenomenon that the output signal of the sensor shifts or changes over time due to various factors, resulting in inaccurate measurement results or errors from the true value. Drift usually occurs gradually over time and may affect the measurement results and system performance. In the prior art, the accuracy is further compensated by using curve fitting on the basis of the two-dimensional table lookup method, which is prone to overfitting to a certain extent, making the originally accurate data even more deviated from the actual value; causing the sensor error to become larger. In order to solve the above-mentioned defects, a technical solution is now provided. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a drift compensation method and system for a low-temperature environment LNG metering sensor to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The drift compensation method of LNG metering sensor in low temperature environment includes the following steps:

[0007] Collect sensor data; process the data; establish a temperature compensation model for the sensor; analyze and mine the displacement change data of each range point within the working temperature range after digital filtering, and use the table lookup method to divide and narrow the vertical interval of the data within the temperature range;

[0008] Calculate G based on the residual error after the table lookup method and the value change in each temperature range, and determine whether it is necessary to use curve fitting to further compensate the temperature compensation model based on G;

[0009] If G is greater than the system preset threshold, when using the curve fitting method, the temperature compensation model is further compensated by using the least squares method or Legendre orthogonal polynomial fitting determined by the G value and data interval distribution; if G is less than the system preset threshold, no subsequent operation is performed.

[0010] In a preferred embodiment, the data acquisition needs to obtain the data changes of the strain gauge at different temperatures. During the design, a temperature sensing device is added near the resistance strain gauge, sensing circuit, and measurement circuit inside the sensor, and the real-time internal ambient temperature data and displacement data are uploaded simultaneously.

[0011] In a preferred embodiment, G is calculated by logistic regression based on the residual error after the table lookup method and the value change in each temperature range.

[0012] In a preferred embodiment, the method for obtaining the residual error after the table lookup method comprises the following steps:

[0013] Step A1: Construct an ideal reference data matrix; obtain the ideal output value matrix of the system under test ,in and is the discrete point of the input variable, covering all grid points of the compensation table;

[0014] Step A2: Collect the actual output data after compensation; measure the system to which the two-dimensional lookup table compensation has been applied, and record the output value matrix after compensation , ensuring that the test conditions are consistent with the ideal data acquisition conditions;

[0015] Step A3: Calculate the error value point by point; for each grid point , calculate the residual error: ; If the error needs to be normalized, it can be further converted into a percentage or a ratio relative to the full scale;

[0016] Step A4: Calculation of global error index; root mean square error: ; The root mean square error is used to represent the residual error after two-dimensional lookup table compensation.

[0017] In a preferred embodiment, the numerical value changes in each temperature segment are determined, and the proportion of continuous compensation values ​​between adjacent temperature segments is judged; the negative number of the proportion of continuous compensation values ​​between adjacent temperature segments represents the numerical value changes in each temperature segment.

[0018] In a preferred embodiment, when G is greater than a preset threshold of the system, it is necessary to use a curve fitting method to compensate the temperature compensation model based on a two-dimensional table lookup method, otherwise it is not necessary to use a curve fitting method to compensate the temperature compensation model.

[0019] In a preferred embodiment, G and data interval distribution are determined, and the selection coefficient is calculated by weighted summation.

[0020] In a preferred embodiment, the selection coefficient is determined and calculated, and when the value of the selection coefficient is greater than a system preset threshold, the least squares method is used for curve fitting, and when the value of the selection coefficient is less than the system preset threshold, the Legendre orthogonal polynomial fitting is used for curve fitting.

[0021] In a preferred embodiment, a temperature compensation model building module, a dynamic compensation decision module, and a curve fitting execution module;

[0022] The temperature compensation model construction module is used to divide the operating temperature range into base segments and sub-segments vertically, store the compensation voltages of the corresponding temperature segments, and construct a segmented data table through the reference voltage and compensation voltage formulas; the temperature drift amplitude is compressed to a smaller range through the table lookup method to reduce the complexity of subsequent fitting;

[0023] The dynamic compensation decision module is used to calculate G according to the residual error after the table lookup method and the value change in each temperature range, and to determine whether it is necessary to use curve fitting to further compensate the temperature compensation model according to G;

[0024] The curve fitting execution module is used to further compensate the temperature compensation model by using the least square method or Legendre orthogonal polynomial fitting through the G value and the data interval distribution when using the curve fitting method.

[0025] In a preferred embodiment, the dynamic compensation decision module includes a residual error evaluation module, and the residual error evaluation module is used to calculate the residual error point by point through an ideal reference data matrix and an actual output matrix.

[0026] Technical effects and advantages of the present invention:

[0027] The present invention collects sensor output data; processes the data; establishes a temperature compensation model of the sensor; performs digital filtering on the displacement change data of each range point within the working temperature range, performs analysis and mining, and uses a table lookup method to divide and reduce the vertical interval of the data within the temperature range; calculates G according to the residual error after the table lookup method and the value change in each temperature section, and determines whether it is necessary to use a curve fitting method to further compensate the temperature compensation model according to G; avoids overfitting, so that the originally accurate data becomes more out of touch with the actual value; if G is greater than a system preset threshold, when using the curve fitting method, determines to use the least square method or Legendre orthogonal polynomial fitting to further compensate the temperature compensation model according to the G value and the data interval distribution; the dynamic adaptive fitting method makes the fitted data more in line with the actual value; if G is less than the system preset threshold, no subsequent operation is performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0029] Figure 1 It is a schematic flow chart of a drift compensation method for a low-temperature environment LNG metering sensor of the present invention;

[0030] Figure 2 It is a structural schematic diagram of the drift compensation system of the LNG metering sensor in a low-temperature environment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The present invention collects sensor output data; processes the data; establishes a temperature compensation model of the sensor; performs digital filtering on the displacement change data of each range point within the working temperature range, performs analysis and mining, and uses a table lookup method to divide and reduce the vertical interval of the data within the temperature range; calculates G according to the residual error after the table lookup method and the value change in each temperature section, and determines whether it is necessary to use a curve fitting method to further compensate the temperature compensation model according to G; if G is greater than a system preset threshold, when using the curve fitting method, determines to use a least square method or Legendre orthogonal polynomial fitting to further compensate the temperature compensation model according to the G value and the data interval distribution; if G is less than the system preset threshold, no subsequent operation is performed.

[0033] Example 1

[0034] The present invention provides a drift compensation method for a low temperature environment LNG metering sensor, such as Figure 1 As shown, the following steps are included:

[0035] Collect sensor output data; process the data; establish a temperature compensation model for the sensor; analyze and mine the displacement change data of each range point within the working temperature range after digital filtering, and use the table lookup method to divide and narrow the vertical interval of the data within the temperature range;

[0036] Calculate G based on the residual error after the table lookup method and the value change in each temperature range, and determine whether it is necessary to use curve fitting to further compensate the temperature compensation model based on G;

[0037] If G is greater than the system preset threshold, when using the curve fitting method, the temperature compensation model is further compensated by using the least squares method or Legendre orthogonal polynomial fitting determined by the G value and data interval distribution; if G is less than the system preset threshold, no subsequent operation is performed.

[0038] Specific:

[0039] Establish a temperature compensation model for high-precision displacement sensors. Place the sensor in a slowly changing temperature field, and set the temperature change range of the temperature field to the product design operating temperature range. Use measurement analysis and correction software to collect displacement data at the zero-scale point, full-scale point, and multiple scale points in between (depending on the situation). After digital filtering, the displacement change data of each scale point within the operating temperature range is analyzed and mined. The table lookup method is used to separate and narrow the vertical interval of the data within the temperature range. Then, the data is linearly fitted with appropriate curve fitting to obtain regularly changing data, build a compensation model, and finally correct the measurement.

[0040] When studying the temperature characteristics of strain gauges, it was found that if the changing displacement data is not processed by digital filtering and table lookup, the data will fluctuate greatly, with no obvious regularity or linear relationship. The processed data amplitude will retain the original data characteristics, reduce data redundancy and data calculation, and easily obtain data with regularity and local linear relationships. When collecting data, it is necessary to obtain the data changes of strain gauges at different temperatures. During the design, temperature sensing devices are added near the resistance strain gauges, sensing circuits, and measurement circuits inside the sensor, and the real-time internal ambient temperature data and displacement data are uploaded simultaneously to facilitate the acquisition of data changes at different temperatures. Through data change analysis and mining, a digital expression or a curve that approximately expresses the data changes is obtained. When acquiring data, the displacement sensor is displaced to a certain position and fixed. The sampling frequency of the displacement sensor can be set to 10Hz. The baseline value of the displacement data of the sensor at this time is recorded. , the temperature is T_N, and 10 displacement data are collected every time the temperature rises by 0.1℃. The median average filtering method is used to find the data fitting value for every 0.1℃, and the output voltage data drift distribution is obtained. According to the measurement data analysis, as the temperature rises, the value of the cumulative displacement data increases. After digital filtering, a segmented data table is constructed, including the base segment temperature, sub-segment temperature and the corresponding output voltage. The data value collected every 0.1℃ is filled in the corresponding position. The processed data amplitude will retain the characteristics of the original data, avoid points with large data errors to a certain extent, and improve the compensation accuracy of high-precision displacement sensors.

[0041] The reference voltage represents the compensation voltage of the corresponding temperature range at the current temperature. It is obtained by subtracting the reference voltage value of the set temperature from the voltage value at the current temperature. When the temperature rises by 0.1°C, the current compensation voltage is increased on the basis of the reference voltage, and the result is: ;Wherein, VN is the reference value of a certain temperature range; Vi is the actual voltage at the current temperature; V_REF is the set temperature reference voltage; V_B is the compensation voltage for every 0.1°C change in the temperature range, that is, the offset voltage caused by temperature change.

[0042] The use of the table lookup method can effectively reduce the amplitude of the displacement result deviation caused by temperature drift, and compensate the data with a large transformation amplitude to a smaller range. The compensation accuracy depends on the reference voltage and the precise compensation of 0.1°C within the temperature range, forming a two-dimensional table lookup compensation method to improve the efficiency of the ordinary table lookup method.

[0043] At the same time, based on the two-dimensional table lookup method, the curve fitting method is used to make up for the deficiencies of the temperature compensation model in terms of flexibility and compensation accuracy, and to increase the reliability of temperature compensation for high-precision displacement sensors. However, in some cases, it is not necessary to further process the compensation accuracy by using curve fitting; excessive compensation will make the compensated value deviate further from the normal value; G is calculated based on the residual error after the table lookup method and the value changes in each temperature range, and G is used to determine whether it is necessary to use curve fitting to further compensate the temperature compensation model.

[0044] Furthermore, the steps of the method for obtaining the residual error after two-dimensional table lookup compensation are as follows:

[0045] Step A1: Construct an ideal reference data matrix; obtain the ideal output value matrix of the system under test through high-precision instruments or theoretical models ,in and For discrete points of input variables (such as sensor input voltage, temperature, etc.), cover all grid points of the compensation table.

[0046] Step A2: Collect the actual output data after compensation; measure the system to which the two-dimensional lookup table compensation has been applied, and record the output value matrix after compensation , ensure that the test conditions are consistent with the ideal data acquisition conditions (such as the same ambient temperature, sampling frequency, etc.).

[0047] Step A3: Calculate the error value point by point; for each grid point , calculate the residual error: ; If the error needs to be normalized, it can be further converted into a percentage or a ratio relative to the full scale.

[0048] Step A4: Calculation of global error index; root mean square error: ; The root mean square error is used to represent the residual error after two-dimensional lookup table compensation.

[0049] The two-dimensional table lookup method stores compensation voltage by temperature segment, but the actual temperature may span multiple segments, and segment compensation may cause discontinuity in compensation values ​​between adjacent temperature segments; determine the value change in each temperature segment, and judge the proportion of compensation value continuity between adjacent temperature segments; the lower the proportion of compensation value continuity between adjacent temperature segments, the more continuous functions should be fitted in each temperature segment to achieve accurate compensation when temperature changes, avoid segment jumps, and enhance compensation continuity. The negative number of the proportion of compensation value continuity between adjacent temperature segments represents the value change in each temperature segment. The larger the proportion of compensation value continuity between adjacent temperature segments, the smaller the value change in each temperature segment.

[0050] Furthermore, the residual error after the table lookup method and the value change in each temperature range are determined. First, the residual error after the table lookup method and the value change in each temperature range are normalized. Specifically, the following formula is used to normalize the data: Normalized to the range [0,1]: ;in is the normalized value, and The data The minimum and maximum values ​​of .

[0051] Then calculate G according to the logistic regression formula. The specific formula is as follows: ; Wherein wc represents the residual error after the table lookup method. The larger the residual error after the table lookup method, the larger G is, and vice versa. bh represents the numerical change in each temperature section. The larger the numerical change in each temperature section, the larger G is, and vice versa. α and β are the logistic regression coefficients of the residual error after the table lookup method and the numerical change in each temperature section, respectively; e is the natural base.

[0052] Determine the G value, and judge whether it is necessary to use curve fitting to further compensate the temperature compensation model based on G. When G is greater than the preset threshold of the system, it indicates that the temperature compensation model is insufficient in flexibility and compensation accuracy. It is necessary to use curve fitting on the basis of the two-dimensional table lookup method to make up for the deficiencies in flexibility and compensation accuracy of the temperature compensation model, and increase the reliability of temperature compensation for high-precision displacement sensors.

[0053] Furthermore, when using the curve fitting method, the selection coefficient is calculated by weighted summation through G and the data interval distribution, and the least squares method or Legendre orthogonal polynomial fitting is determined according to the value of the selection coefficient.

[0054] Specifically, determine G and data interval distribution through the following formula: F=a*G+b*fb; where F represents the selection coefficient; fb represents the data interval distribution; a and b are weight coefficients of G and data interval distribution, respectively, and both are greater than zero. Determine the selection coefficient. When the value of the selection coefficient is greater than the system preset threshold, use the least squares method to perform curve fitting. When the value of the selection coefficient is less than the system preset threshold, use Legendre orthogonal polynomial fitting to perform curve fitting.

[0055] The smaller the G value is, the smaller the range that needs to be fitted is. The least squares method is less effective in resisting overfitting than the Legendre orthogonal polynomial fitting, so the Legendre orthogonal polynomial fitting should be used when G is small. If the data is distributed in a fixed interval, the Legendre orthogonal polynomial fitting is used, because the Legendre orthogonal polynomial fitting is based on orthogonal basis functions, has high numerical stability, and is suitable for nonlinear high-precision approximation within a fixed interval.

[0056] On the basis of the two-dimensional table lookup method, the curve fitting method is used to make up for the deficiencies of the temperature compensation model in terms of flexibility and compensation accuracy, and to increase the reliability of temperature compensation for high-precision displacement sensors. There are two curve fitting methods. The least squares method is a mathematical optimization technique that performs regression analysis on related variables with nonlinear relationships based on scattered data to find the best regression function. Its goal is to find the parameters that make the fitting model as close as possible to the actual observation by minimizing the residual sum of squares between the observed data and the fitting curve, and to easily obtain unknown data. This method is simple in theory and small in calculation, and has been widely used in the fields of curve fitting, regression analysis, and data processing.

[0057] Legendre orthogonal polynomial fitting is a method for fitting the weight function on the interval [-1,1]. Orthogonal polynomial sequence. Its orthogonality is expressed as follows: if m is not equal to n, then ; If m is equal to n, then ;

[0058] The recursive formula is: , , .

[0059] In discrete data fitting, Legendre polynomials approximate functions through the least squares method. Their orthogonality can reduce the condition number of the equation system and improve numerical stability, making them particularly suitable for high-order fitting.

[0060] Example 2

[0061] The present invention provides a drift compensation system for LNG metering sensors in a low temperature environment, such as Figure 2 As shown, it includes the following modules: temperature compensation model building module, residual error evaluation module, dynamic compensation decision module, and curve fitting execution module;

[0062] The temperature compensation model construction module is used to divide the operating temperature range into base segments and sub-segments vertically, store the compensation voltages of the corresponding temperature segments, and construct a segmented data table through the reference voltage and compensation voltage formulas; the temperature drift amplitude is compressed to a smaller range through the table lookup method to reduce the complexity of subsequent fitting;

[0063] The residual error evaluation module calculates the residual error point by point through the ideal reference data matrix and the actual output matrix;

[0064] The dynamic compensation decision module is used to calculate G according to the residual error after the table lookup method and the value change in each temperature range, and to determine whether it is necessary to use curve fitting to further compensate the temperature compensation model according to G;

[0065] The curve fitting execution module is used to further compensate the temperature compensation model by using the least square method or Legendre orthogonal polynomial fitting through the G value and the data interval distribution when using the curve fitting method.

[0066] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0067] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0068] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0069] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0070] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A drift compensation method for LNG metering sensors in a low temperature environment, characterized in that: The following steps are involved: Collect sensor data; process the data; establish a temperature compensation model for the sensor; analyze and mine the displacement change data of each range point within the working temperature range after digital filtering, and use the table lookup method to divide and narrow the vertical interval of the data within the temperature range; Calculate G based on the residual error after the table lookup method and the value change in each temperature range, and determine whether it is necessary to use curve fitting to further compensate the temperature compensation model based on G; If G is greater than the system preset threshold, when using the curve fitting method, the temperature compensation model is further compensated by using the least squares method or Legendre orthogonal polynomial fitting determined by the G value and data interval distribution; if G is less than the system preset threshold, no subsequent operation is performed.

2. The drift compensation method for LNG metering sensor in low temperature environment according to claim 1 is characterized in that: The data acquisition requires obtaining the data changes of the strain gauges at different temperatures. During the design, a temperature sensing device is added near the resistance strain gauge, sensing circuit, and measurement circuit inside the sensor, and the real-time internal ambient temperature data and displacement data are uploaded simultaneously.

3. The drift compensation method for LNG metering sensor in low temperature environment according to claim 1 is characterized in that: Based on the residual error after the table lookup method and the value changes in each temperature range, G is calculated by logistic regression.

4. The drift compensation method for LNG metering sensor in low temperature environment according to claim 3 is characterized in that: The steps of obtaining the residual error after the table lookup method are as follows: Step A1: Construct an ideal reference data matrix; obtain the ideal output value matrix of the system under test ,in and is the discrete point of the input variable, covering all grid points of the compensation table; Step A2: Collect the actual output data after compensation; measure the system to which the two-dimensional lookup table compensation has been applied, and record the output value matrix after compensation , ensuring that the test conditions are consistent with the ideal data acquisition conditions; Step A3: Calculate the error value point by point; for each grid point , calculate the residual error: ; If the error needs to be normalized, it can be further converted into a percentage or a ratio relative to the full scale; Step A4: Calculation of global error index; root mean square error: ; The root mean square error is used to represent the residual error after two-dimensional lookup table compensation.

5. The drift compensation method for LNG metering sensor in low temperature environment according to claim 3 is characterized in that: The numerical value changes in each temperature segment are determined, and the continuous proportion of compensation values ​​between adjacent temperature segments is judged; the negative number of the continuous proportion of compensation values ​​between adjacent temperature segments represents the numerical value changes in each temperature segment.

6. The drift compensation method for LNG metering sensor in low temperature environment according to claim 3 is characterized in that: When G is greater than a preset threshold of the system, it is necessary to use a curve fitting method to compensate the temperature compensation model based on the two-dimensional table lookup method. Otherwise, it is not necessary to use a curve fitting method to compensate the temperature compensation model.

7. The drift compensation method for LNG metering sensor in low temperature environment according to claim 6 is characterized in that: Determine G, data interval distribution, and calculate the selection coefficient by weighted summation.

8. The drift compensation method for LNG metering sensor in low temperature environment according to claim 7, characterized in that: Determine and calculate the selection coefficient. When the value of the selection coefficient is greater than the system preset threshold, use the least squares method to perform curve fitting. When the value of the selection coefficient is less than the system preset threshold, use Legendre orthogonal polynomial fitting to perform curve fitting.

9. A drift compensation system for a low temperature environment LNG metering sensor, used to implement the drift compensation method for a low temperature environment LNG metering sensor according to any one of claims 1 to 8, characterized in that: It includes the following modules: temperature compensation model building module, dynamic compensation decision module, and curve fitting execution module; The temperature compensation model construction module is used to divide the operating temperature range into base segments and sub-segments vertically, store the compensation voltages of the corresponding temperature segments, and construct a segmented data table through the reference voltage and compensation voltage formulas; the temperature drift amplitude is compressed to a smaller range through the table lookup method to reduce the complexity of subsequent fitting; The dynamic compensation decision module is used to calculate G according to the residual error after the table lookup method and the value change in each temperature range, and to determine whether it is necessary to use curve fitting to further compensate the temperature compensation model according to G; The curve fitting execution module is used to further compensate the temperature compensation model by using the least square method or Legendre orthogonal polynomial fitting through the G value and the data interval distribution when using the curve fitting method.

10. The drift compensation system for LNG metering sensors in low temperature environments according to claim 9, characterized in that: The dynamic compensation decision module includes a residual error evaluation module, which is used to calculate the residual error point by point through an ideal reference data matrix and an actual output matrix.

Images