Error correction method and system based on test temperature sensor

Abstract

The invention discloses an error correction method and system based on a test temperature sensor, and relates to the field of temperature sensor correction. According to the error correction method and system based on the test temperature sensor, a fitting curve is generated based on errors of data samples of the temperature sensor, then an error compensation mechanism is established, fixed numerical value automatic compensation is carried out on the errors of different temperature points, dynamic compensation is added according to expressions of the error fitting curve at different temperatures, and the error correction accuracy is improved. And finally, on the basis of dynamic compensation, setting segmented temperature compensation, and introducing a weighting function at a partition boundary position to realize cross-temperature-zone compensation smooth transition, thereby realizing fixed compensation with the same average error and dynamic compensation with different average errors on the error of the temperature sensor for different temperature zones. Therefore, different error compensation can be carried out in different temperature zones, errors can be eliminated more accurately, and the requirement for ultrahigh precision of the temperature sensor in some special fields at present can be met.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature sensor correction, and in particular to an error correction method and system based on testing a temperature sensor. Background Art

[0002] A temperature sensor is an electronic component used for temperature monitoring. According to its principles, it can be divided into thermocouple temperature sensors, thermistor temperature sensors, resistance temperature sensors, infrared temperature sensors and semiconductor temperature sensors. Choosing a suitable temperature sensor according to the usage scenario can improve the accuracy of temperature detection.

[0003] However, in actual application, due to interference from manufacturing process, material properties and external environmental factors, the temperature sensor has a certain range of errors when in use. Reasonable use and product quality inspection can effectively eliminate certain errors, but the self-limiting error itself cannot be objectively eliminated. Therefore, it is usually necessary to use corresponding correction algorithm logic to compensate for it, so as to further reduce the error range.

[0004] A method for correcting nonlinear errors in a digital temperature sensor system is disclosed in a patent publication numbered CN112924054B. However, in the patent, the main core of error correction is to select multiple temperature points within the temperature measurement range for measurement, and then fit the error values ​​of the multiple temperature points, and then use the mean difference method for data processing to achieve error correction.

[0005] However, in actual applications, temperature sensors are affected by the characteristics of the materials they are made of, and the errors in different temperature zones are not expressed as functions of the same weight. In a specific temperature range, the error changes vary. Therefore, correction through the above scheme will cause the actual measurement error to have a large error in a specific temperature zone, which does not meet the current demand for ultra-high precision temperature sensors in some special fields. For this purpose, an error correction method and system based on a test temperature sensor are provided. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides an error correction method and system based on a test temperature sensor, which solves the problem that the correction method for the nonlinear error of the digital temperature sensor system disclosed in the existing patent CN112924054B has differences in error change performance within a specific temperature range. Therefore, correction through the above scheme will cause the actual measurement error to have a large error within a specific temperature zone, which does not meet the current demand for ultra-high precision temperature sensors in some special fields.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: an error correction method based on a test temperature sensor comprises the following steps;

[0008] S1, generating a fitting curve based on the error of the temperature sensor data sample;

[0009]

[0010] Where E is the error at each temperature point, αn, αn-1…, αn are polynomial coefficients, and n is the polynomial order;

[0011] S2. According to the fitting curve, an error compensation mechanism is established, and the error function E(T m ) dynamically adjusts the output of the temperature sensor, and the corrected temperature T c It is expressed as;

[0012] T c =T m +E(T m )

[0013] Substitute the error fitting polynomial into the equation to obtain;

[0014]

[0015] In this way, the errors at different temperature points can be automatically compensated with fixed values;

[0016] S3. Add dynamic compensation according to the performance of the error fitting curve at different temperatures;

[0017] The dynamic compensation formula is expressed as;

[0018] T c (t) = T m (t)+E(T m (t),t)

[0019] Among them, E(T m (t),t) is the error function that changes with time t;

[0020] After the introduction of dynamic compensation, a new error correction function is obtained;

[0021] Tc(t)=Tm(t)+(αn(t)×Tm(t)n+αn-1(t)×Tm(t)n- 1 +···+α1(t)×Tm(t)+α0(t))

[0022] Then, on the basis of dynamic compensation, segmented temperature compensation is set. According to the temperature sensitivity, sparse points are taken in the interval where the error changes smoothly, and dense points are taken in the interval where the error changes sharply. They are set as interval 1, interval 2...interval n respectively. For the kth temperature interval, an independent polynomial is fitted:

[0023] Ek(T)=αk, n Tn+αk, n-1 Tn- 1 +···+αk,0(k=1,2,…,k)

[0024] At the same time, in order to ensure smooth compensation transition between two adjacent temperature intervals, the weight function ω is introduced to define k (T), the boundary temperature between interval K and k+1 is T b , so the weight function is obtained as follows;

[0025] And ω k+1 (T) = 1-ω k (T)

[0026] When T is close to T b hour,

[0027] T C =ω k (T)·(T+E k (T))+ω k+1 (T)·(T+E k+1 (T))

[0028] S4. Verification and optimization.

[0029] Preferably, the specific implementation steps of S1 are:

[0030] 1. Data collection, at different temperatures T a Next, record the actual measured temperature value T of the temperature sensor m ;

[0031] 2. Calculation error E, E = T a -T m ;

[0032] 3. Use polynomial fitting error E and measurement value T m The relationship between them is obtained, thus obtaining the curve function shown in S1.

[0033] Preferably, during the test, n groups of temperature sensors are selected from the same batch for synchronous testing, and the average value is obtained. Under expected circumstances, the actual error curve should be a nearly smooth curve. If specific data appears, for the specific data of a single temperature sensor, it is judged to be a process problem of the temperature sensor, and this group of data is eliminated. For the specific data of multiple temperature sensors, it is judged to be a systematic error and retested after verifying the experimental environment.

[0034] Preferably, the data statistics in the test process follow the normal distribution principle, and in the specific data processing process, covariates are introduced and covariance analysis is used to eliminate the influence of uncertain factors.

[0035] Preferably, in the polynomial function fitting process in S2, the value range of the term order n is 2-4, according to the temperature difference T c The error variation range in each temperature zone can be adjusted by selecting appropriate orders. A high order can be used in a large range of variation, and a low order can be used in a small range of variation. Then, the optimal order can be determined through cross-validation to avoid overfitting.

[0036] Preferably, in S3, a high-order polynomial (such as third order) is used to compensate for nonlinear errors in the low temperature region, and a linear term is combined with an exponential decay term in the high temperature region to suppress the influence of material aging.

[0037] Preferably, in the verification process in S4, 20% of the sensor samples are reserved as a test set to calculate the mean absolute error after compensation;

[0038]

[0039] It is necessary to ensure that MAE ≤ 0.1.

[0040] Preferably, during the optimization process in S4, the entire test process needs to be continuously run for 100 hours between -50°C and 300°C to monitor the error drift after compensation.

[0041] An error correction system based on a test temperature sensor, comprising:

[0042] The data acquisition module obtains the actual temperature value of the test environment through the calibrated standard temperature sensor and reads the real-time data of the temperature sensor used for the test;

[0043] Data processing module, which cleans and stores the collected data;

[0044] Error dynamic model, which performs curve fitting based on the error at each temperature point to provide a model for the dynamic compensation mechanism;

[0045] The verification module performs repeated data verification on the error compensation mechanism and dynamic compensation of the final molding.

[0046] The present invention discloses an error correction method and system based on a test temperature sensor, which has the following beneficial effects:

[0047] 1. The error correction method based on the test temperature sensor generates a fitting curve based on the error of the temperature sensor data sample, and then establishes an error compensation mechanism according to the fitting curve, and dynamically adjusts the output of the temperature sensor through the error function, so as to automatically compensate the errors at different temperature points with fixed values, and then according to the performance of the error fitting curve at different temperatures, additional dynamic compensation is added, and finally, on the basis of dynamic compensation, segmented temperature compensation is set, and a weight function is introduced at the boundary of the partition to achieve a smooth transition of compensation across temperature zones, so as to achieve fixed compensation of the same mean difference and dynamic compensation of different mean differences for the errors of the temperature sensor in different temperature zones, so as to achieve different error compensation in different temperature zones, so as to eliminate errors more accurately and meet the current demand for ultra-high precision of temperature sensors in some special fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 This is a flow chart of an error correction method based on a test temperature sensor of the present invention;

[0050] Figure 2 This is a functional framework diagram of the error correction system based on the test temperature sensor of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are 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.

[0052] The embodiments of the present application provide an error correction method and system based on a test temperature sensor, thereby solving the problem that the method for correcting the nonlinear error of a digital temperature sensor system disclosed in the existing patent CN112924054B has differences in error variation within a specific temperature range. Therefore, correction through the above scheme will result in a larger error in the actual measurement error within a specific temperature zone, which does not meet the current demand for ultra-high precision temperature sensors in some special fields.

[0053] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0054] Embodiment 1

[0055] The embodiment of the present invention discloses an error correction method based on testing a temperature sensor.

[0056] According to the attached Figure 1-2 As shown, the following steps are included;

[0057] S1, generating a fitting curve based on the error of the temperature sensor data sample;

[0058]

[0059] Where E is the error at each temperature point, αn, αn-1…, αn are polynomial coefficients, and n is the polynomial order;

[0060] S2. According to the fitting curve, an error compensation mechanism is established, and the error function E(T m ) dynamically adjusts the output of the temperature sensor, and the corrected temperature T c It is expressed as;

[0061] T c =T m +E(T m )

[0062] Substitute the error fitting polynomial into the equation to obtain;

[0063]

[0064] In this way, the errors at different temperature points can be automatically compensated with fixed values;

[0065] S3. Add dynamic compensation according to the performance of the error fitting curve at different temperatures;

[0066] The dynamic compensation formula is expressed as;

[0067] T c (t) = T m (t)+E(Tm (t),t)

[0068] Among them, E(T m (t),t) is the error function that changes with time t;

[0069] After the introduction of dynamic compensation, a new error correction function is obtained;

[0070] Tc(t)=Tm(t)+(αn(t)×Tm(t)n+αn-1(t)×Tm(t)n- 1 +···+α1(t)×Tm(t)+α0(t))

[0071] Then, on the basis of dynamic compensation, segmented temperature compensation is set. According to the temperature sensitivity, sparse points are taken in the interval where the error changes smoothly, and dense points are taken in the interval where the error changes sharply. They are set as interval 1, interval 2...interval n respectively. For the kth temperature interval, an independent polynomial is fitted:

[0072] Ek(T)=αk, n Tn+αk, n-1 Tn- 1 +···+αk,0(k=1,2,…,k)

[0073] At the same time, in order to ensure smooth compensation transition between two adjacent temperature intervals, the weight function ω is introduced to define k (T), the boundary temperature between interval K and k+1 is T b , so the weight function is obtained as follows;

[0074] And ω k+1 (T) = 1-ω k (T)

[0075] When T is close to T b hour,

[0076] T C =ω k (T)·(T+E k (T))+ω k+1 (T)·(T+E k+1 (T))

[0077] S4. Verification and optimization.

[0078] Preferably, the specific implementation steps of S1 are:

[0079] 1. Data collection, at different temperatures T a Next, record the actual measured temperature value T of the temperature sensor m ;

[0080] 2. Calculation error E, E = T a -T m ;

[0081] 3. Use polynomial fitting error E and measurement value T m The relationship between them is obtained, thus obtaining the curve function shown in S1.

[0082] Preferably, during the test, n groups of temperature sensors are selected from the same batch for synchronous testing to ensure process consistency and obtain an average value. Under expected circumstances, the actual error curve should be a nearly smooth curve. If specific data appears, for the specific data of a single temperature sensor, it is judged to be a process problem of the temperature sensor, and this group of data is discarded. For the specific data of multiple temperature sensors, it is judged to be a systematic error, and retesting is performed after verifying the experimental environment. During the test, a constant temperature bath or high and low temperature box needs to be built, and the temperature range covers the sensor range to provide a reliable test environment and eliminate environmental errors.

[0083] Preferably, the data statistics in the test process follow the normal distribution principle, and in the specific data processing process, covariates are introduced and covariance analysis is used to eliminate the influence of uncertain factors.

[0084] Preferably, in the polynomial function fitting process in S2, the value range of the term order n is 2-4, according to the temperature difference T c The error variation range in each temperature zone can be adjusted by selecting appropriate orders. A high order can be used in a large range of variation, and a low order can be used in a small range of variation. Then, the optimal order can be determined through cross-validation to avoid overfitting.

[0085] If we assume that the order of the term is n = 2, it can be expressed as

[0086]

[0087] Then according to the above formula, we can get;

[0088] Tc(t)=Tm(t)+(α2(t)×Tm(t) 2 +α1(t)×Tm(t)+α0(t))

[0089] After introducing dynamic compensation, we can get:

[0090] Tc(t)=Tm(t)+(α2(t)×Tm(t) 2 +α1(t)×Tm(t)+α0(t))

[0091] Finally, by introducing the weight function, a smooth transition across temperature zones is achieved, thereby achieving fixed compensation with equal mean differences and dynamic compensation with different mean differences for the errors of temperature sensors in different temperature zones. This allows different error compensations to be performed in different temperature zones, thereby eliminating errors more accurately and meeting the current demand for ultra-high precision temperature sensors in some special fields.

[0092] Preferably, in S3, a high-order polynomial (such as third order) is used to compensate for nonlinear errors in the low temperature region, and a linear term is combined with an exponential decay term in the high temperature region to suppress the influence of material aging.

[0093] Preferably, in the verification process in S4, 20% of the sensor samples are reserved as a test set to calculate the mean absolute error after compensation;

[0094]

[0095] It is necessary to ensure that MAE ≤ 0.1.

[0096] Preferably, during the optimization process in S4, the entire test process needs to be continuously run for 100 hours between -50°C and 300°C to monitor the error drift after compensation.

[0097] Embodiment 2

[0098] See attached Figure 1-2 The present invention also discloses an error correction system based on a test temperature sensor, comprising:

[0099] The data acquisition module obtains the actual temperature value of the test environment through the calibrated standard temperature sensor and reads the real-time data of the temperature sensor used for the test;

[0100] Data processing module, which cleans and stores the collected data;

[0101] Error dynamic model, which performs curve fitting based on the error at each temperature point to provide a model for the dynamic compensation mechanism;

[0102] The verification module performs repeated data verification on the error compensation mechanism and dynamic compensation of the final molding.

[0103] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An error correction method based on a test temperature sensor, characterized in that: The steps include: S1, generating a fitting curve based on the error of the temperature sensor data sample; Where E is the error at each temperature point, αn, αn-1…, αn are polynomial coefficients, and n is the polynomial order; S2. According to the fitting curve, an error compensation mechanism is established, and the error function E(T m ) dynamically adjusts the output of the temperature sensor, and the corrected temperature T c It is expressed as; T c =T m +E(T m ) Substitute the error fitting polynomial into the equation to obtain; In this way, the errors at different temperature points can be automatically compensated with fixed values; S3. Add dynamic compensation according to the performance of the error fitting curve at different temperatures; The dynamic compensation formula is expressed as; T c (t)=T m (t)+E(T m (t),t) Among them, E(T m (t),t) is the error function that changes with time t; After the introduction of dynamic compensation, a new error correction function is obtained; Tc(t)=Tm(t)+(αn(t)×Tm(t)n+αn-1(t)×Tm(t)n- 1 +···+α1(t)×Tm(t)+α0(t)) Then, based on the dynamic compensation, segmented temperature compensation is set. According to the temperature sensitivity, sparse points are taken in the interval where the error changes smoothly, and dense points are taken in the interval where the error changes sharply. They are set as interval 1, interval 2...interval n respectively. For the kth temperature interval, an independent polynomial is fitted: Ek(T)=αk, n Tn+αk, n-1 Tn- 1 +···+αk,0(k=1,2,…,k) At the same time, in order to ensure smooth compensation transition between two adjacent temperature intervals, the weight function ω is introduced to define k (T), the boundary temperature between interval K and k+1 is T b , so the weight function is obtained as follows; and ω k+1 (T) = 1 - ω k (T) When T is close to T b hour, T C =ω k (T)·(T+E k (T))+ω k+1 (T)·(T+E k+1 (T)) S4. Verification and optimization.

2. The error correction method based on the test temperature sensor according to claim 1, characterized in that: The specific implementation steps of S1 are:

1. Data collection, at different temperatures T a Next, record the actual measured temperature value T of the temperature sensor m ; 2. Calculation error E, E = T a -T m ; 3. Use polynomial fitting error E and measurement value T m The relationship between them is obtained, thus obtaining the curve function shown in S1.

3. The error correction method based on the test temperature sensor according to claim 2, characterized in that: During the test, n groups of temperature sensors are selected from the same batch for synchronous testing and the average value is obtained. Under expected circumstances, the actual error curve should be a nearly smooth curve. If specific data appears, for the specific data of a single temperature sensor, it is judged to be a process problem of the temperature sensor, and this group of data is eliminated. For the specific data of multiple temperature sensors, it is judged to be a systematic error and retested after verifying the experimental environment.

4. The error correction method based on the test temperature sensor according to claim 3, characterized in that: The data statistics in the test process follow the normal distribution principle. In the process of specific data processing, covariates are introduced and covariance analysis is used to eliminate the influence of uncertain factors.

5. The error correction method based on testing temperature sensor according to claim 1, characterized in that: In the polynomial function fitting process in S2, the value range of the term order n is 2-4, according to the temperature difference T c The error variation range in each temperature zone can be adjusted by selecting appropriate orders. A high order can be used in a large range of variation, and a low order can be used in a small range of variation. Then, the optimal order can be determined through cross-validation to avoid overfitting.

6. The error correction method based on testing temperature sensor according to claim 1, characterized in that: In S3, a high-order polynomial (such as the third order) is used to compensate for nonlinear errors in the low-temperature region, and a linear term is combined with an exponential decay term in the high-temperature region to suppress the influence of material aging.

7. The error correction method based on testing temperature sensor according to claim 1, characterized in that: In the verification process in S4, 20% of the sensor samples are reserved as a test set to calculate the mean absolute error after compensation; It is necessary to ensure that MAE ≤ 0.

1.

8. The error correction method based on testing temperature sensor according to claim 1, characterized in that: During the optimization process in S4, the entire test process needs to be continuously run for 100 hours between -50°C and 300°C to monitor the error drift after compensation.

9. An error correction system based on a test temperature sensor, used to execute the error correction method based on a test temperature sensor according to any one of claims 1 to 8, characterized in that: include: The data acquisition module obtains the actual temperature value of the test environment through the calibrated standard temperature sensor and reads the real-time data of the temperature sensor used for the test; Data processing module, which cleans and stores the collected data; Error dynamic model, which performs curve fitting based on the error at each temperature point to provide a model for the dynamic compensation mechanism; The verification module performs repeated data verification on the error compensation mechanism and dynamic compensation of the final molding.

Images