An Optimization Method for Low-Cost CO2 Sensor Calibration Solutions

Abstract

The present invention discloses an optimization method suitable for a low-cost CO2 sensor calibration scheme, which consists of sensor calibration and calibration scheme optimization. The calibration scheme optimization includes training set optimization of the calibration equation and optimization of the interval between two calibrations. The data 30 days after the training set is used as the test set, and the root mean square error between the calibration value of the low-cost CO2 sensor in the test set and the reference instrument is used as the standard. The training set with the shortest time length is selected in which the root mean square error meets the user's needs, and this is used as the basis for the optimal training set length. When the root mean square error is continuously greater than the observation requirement, it indicates that the reliability of the low-cost CO2 sensor can no longer meet the needs, that is, the test set data has exceeded the generalization ability of the calibration model, and the low-cost CO2 sensor needs to be calibrated again. The interval between two calibrations is the optimal calibration interval of the low-cost CO2 sensor. While ensuring the reliability of the low-cost CO2 sensor, the cost investment of the calibration work is reduced.

Description

Technical Field

[0001] The present invention relates to a method suitable for low-cost CO 2 The invention discloses an optimization method for a sensor calibration scheme, belonging to the technical field of environmental quality monitoring. Background Art

[0002] Low-cost CO 2 The sensor is small and inexpensive, breaking through the limitations of traditional reference instruments that are expensive and require high installation. It can be flexibly deployed in various environmental conditions and improve the spatial coverage of measurements to capture CO caused by specific emission sources. 2 However, these sensors require regular calibration to generate sufficiently reliable and accurate data. 2 The sensor lacks temperature and humidity control devices. After baseline calibration under laboratory conditions, the readings will still be biased due to changes in temperature, humidity and pressure under natural environmental conditions, requiring post-data calibration to optimize data quality. The field calibration method uses low-cost CO 2 The sensor is compared with the observation data of the high-precision reference instrument to establish the regression equation of the observation error on temperature, humidity and pressure. This method can not only correct the baseline drift caused by long-term use of the instrument, but also reduce the impact of fluctuations in environmental conditions on the observation error, providing better calibration results for real applications. Since the field calibration method reduces the investment in manpower and material resources required for traditional calibration, it is a good choice for low-cost CO 2 Sensors offer considerable market potential. However, the generalization ability of the above-mentioned field calibration model, that is, its adaptability to fresh samples, varies due to differences in the performance of the sensor itself and the environment of the application scenario. Therefore, the size of the calibration model training set and the interval between two calibrations are also different, which causes inconvenience to the application of the field calibration model. Summary of the invention

[0003] The present invention provides a method suitable for low-cost CO 2 The optimization method of the sensor calibration solution solves the problems disclosed in the background technology. 2 While improving the reliability of the sensor, it also further reduces the manpower and material resources input in the calibration work.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A low-cost CO 2 Optimization method of sensor calibration scheme:

[0006] Step A1: Low-cost CO 2The comparison observation data between the sensor and the high-precision reference instrument is used as a training set to establish the regression equation of the observation error with respect to the environmental factors;

[0007] Step A2: Use the data of several consecutive days after the training set as the test set;

[0008] Step A3, calculating the root mean square error of the calibrated low-cost sensor of the test set;

[0009] Step A4, determining whether the root mean square error of the calibrated low-cost sensor of the test set is less than the user's requirement on the error;

[0010] Step A5: If the root mean square error of the low-cost sensor after calibration of the test set is less than the user's error requirement, proceed to step 6; otherwise, n=n+1, return to step A1;

[0011] Step A6: Determine whether the environmental factors in the training set cover low-cost CO 2 Environmental factors of sensor application scenarios;

[0012] Step A7: If the environmental factors in the training set cover low-cost CO 2 Environmental factors of sensor application scenarios,

[0013] Then the output N1=n is the shortest number of consecutive days required for the training set data;

[0014] Otherwise, replace the low-cost CO 2 The comparison observation data between the sensor and the high-precision reference instrument is returned to step A1 as a training set.

[0015] Furthermore, the root mean square error of the low-cost sensor after calibration in step A3 is

[0016]

[0017] In the formula, C′ LCS For low-cost CO 2 Calibration value of the sensor, C REF is the reading of the reference instrument and n is the number of samples.

[0018] Furthermore, in step A2, the data of 30 consecutive days after the training set is used as the test set.

[0019] Furthermore, it also includes:

[0020] Step B1: Low-cost CO 2 The comparison observation data between the sensor and the high-precision reference instrument is used as a training set to establish the regression equation of the observation error with respect to the environmental factors;

[0021] Step B2: Use the low-cost CO for consecutive days after N1 2 The comparison observation data between the sensor and the high-precision reference instrument is used as the test set;

[0022] Step B3: Calculate the daily RMSE starting from the first day corresponding to the test set 2 ;

[0023] Step B4: Determine the RMSE after the nth day 2 Whether it is continuously greater than the user's requirement for error;

[0024] Step B5: If RMSE after the nth day 2 If it is continuously greater than the user's requirement for error, then n is the low-cost CO 2 The interval between two sensor calibrations goes to step B6, otherwise n=n+1 and returns to step B4;

[0025] Step B6: Determine whether the environmental factors in the training set cover low-cost CO 2 Environmental factors of sensor application scenarios;

[0026] Step B7: If the environmental factors in the training set cover low-cost CO 2 Environmental factors of sensor application scenarios,

[0027] Then the average value N2 of the multiple interval times output is the optimal time interval between two calibrations;

[0028] Otherwise, replace the low-cost CO for N1 days 2 The comparison observation data between the sensor and the high-precision reference instrument is returned to step B1 as a training set.

[0029] Accordingly, a computer-readable storage medium stores one or more programs: the one or more programs include instructions, and when the instructions are executed by a computing device, the computing device performs any of the methods described above.

[0030] Accordingly, a computing device includes:

[0031] One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the methods described above.

[0032] The beneficial effects achieved by the present invention are: 2 Optimization Method for Sensor Calibration Scheme for Low-Cost CO 2Sensor calibration provides accurate training set size and interval between two calibrations, ensuring low-cost CO 2 The reliability of the sensor is improved while reducing the maintenance cost of the instrument. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a schematic diagram of a zero point calibration device;

[0034] Figure 2 This is a schematic diagram of optimizing the number of days for training sets in the present invention;

[0035] Figure 3 Number of training set days and calibration effect index (RMSE 2 ) relationship diagram;

[0036] Figure 4 It is a schematic diagram of optimizing the interval between two calibrations in the present invention;

[0037] Figure 5 The daily root mean square error (RMSE) of the test set (1-120 days after the training set) 2 ) distribution map,;

[0038] Figure 6 The figure is a schematic diagram of the optimization method of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0040] The present invention aims to provide a low-cost CO 2 The optimization method of sensor calibration scheme can reduce the manpower and material input of calibration work as much as possible. This method consists of two aspects: sensor calibration and calibration scheme optimization, achieving high-quality and low-cost CO 2 Concentration observation.

[0041] Sensor calibration includes two steps: zero point calibration and environmental calibration. Zero point calibration is to convert low-cost CO 2 The sensor is placed in a self-developed calibration device to reduce the CO 2 The concentration is close to 0 μmol·mol -1 , adjust the sensor parameters to make its reading zero. Take out the low-cost CO 2 After the sensor is installed, it is compared with a high-precision reference instrument under real environmental conditions for environmental calibration. 2 The difference between the sensor and the reference instrument quantifies low-cost CO 2 The observation error of the sensor. 2The original readings of the sensor and environmental factors such as temperature, humidity, and air pressure are used as independent variables, with low-cost CO 2 The observation error of the sensor is taken as the dependent variable, and a regression equation is established to calibrate the low-cost CO caused by changes in environmental conditions. 2 The sensor's reading deviation.

[0042] The calibration scheme optimization includes the optimization of the training set of the calibration equation and the optimization of the interval between two calibrations. Figure 6 As shown, using low-cost CO 2 The sensor is compared with the high-precision reference instrument, and data of different time lengths are selected as training sets to obtain low-cost CO 2 The original readings of the sensor and environmental factors such as temperature, humidity, and air pressure are used as independent variables, with low-cost CO 2 The observation error of the sensor is taken as the dependent variable, and the regression equation is established. The data of 30 days after the training set is used as the test set, and the low-cost CO 2 The root mean square error between the sensor calibration value and the reference instrument is used as the standard. The training set with the shortest time length is selected from the training sets whose root mean square error meets the user's requirements, and this is used as the basis for the optimal training set length. The comparison data of the optimal training set length is selected as the training set, and the low-cost CO 2 The original readings of the sensor and environmental factors such as temperature, humidity, and air pressure are used as independent variables, with low-cost CO 2 The observation error of the sensor is taken as the dependent variable, and the regression equation is established. The comparison data of different time lengths after the training set is used as the test set, and the low-cost CO 2 The root mean square error between the calibration value of the sensor and the reference instrument is used as the standard. When the root mean square error is greater than the observation requirement, it indicates that the low-cost CO 2 The reliability of the sensor can no longer meet the needs, that is, the test set data has exceeded the generalization ability of the calibration model, and low-cost CO 2 The sensor is calibrated again, and the interval between the two calibrations is the low-cost CO 2 The optimal calibration interval for the sensor.

[0043] Example

[0044] The present invention is a set of suitable low-cost CO 2 The optimization method of the sensor calibration scheme, the specific implementation plan is as follows:

[0045] Step 1: Use a self-developed calibration device (such as Figure 1 ) for zero point calibration. The device uses soda lime, high purity CO 2 and N 2 Gas conditioning calibration device CO 2 The concentration of CO in the calibration device is tracked in real time by a high-precision measuring instrument.2 Reduce the CO concentration in the calibration device 2 concentration, when the concentration is close to 0 μmol·mol -1 When the low-cost sensor is in operation, click the calibration button to return the low-cost sensor's reading to 0.

[0046] Step 2: Choose an outdoor location with good ventilation and no emission sources nearby, and use a high-precision measuring instrument as a reference instrument to compare with the low-cost CO 2 The sensors were continuously compared and observed for more than half a year. 2 The environmental factors of sensor drift are monitored, including but not limited to temperature, humidity, air pressure, etc. 2 The sensor readings are compared with reference instruments to obtain low-cost CO 2 The sensor's observation error.

[0047] Step 3: Establish a calibration equation whose dependent variable is low-cost CO 2 Observation error of the sensor, the independent variable is low-cost CO 2 The original readings of the sensor and environmental factors such as temperature, humidity, and air pressure are used to filter out the root mean square error (RMSE) in regression equations such as linear regression, polynomial regression, support vector machine regression, decision tree regression, random forest regression, LASSO regression, Ridge regression, ElasticNet regression, and XGBoost regression. 1 )The smallest regression equation is used as the calibration equation:

[0048]

[0049] In the formula, C CAL Low-cost CO calculated for calibration equations 2 Sensor observation error, C OBS Low-cost CO for raw measurements 2 The observation error of the sensor, n is the number of samples.

[0050] Step 4: Calibration effect evaluation. Use the calibration equation to calibrate the low-cost CO 2 The sensor reading deviation will 2 The sensor's calibration values ​​are compared with the reference instrument and the root mean square error (RMSE) of the calibrated low-cost sensor is calculated. 2 ), as an indicator to evaluate the calibration effect.

[0051]

[0052] In the formula, C′ LCS For low-cost CO 2 Calibration value of the sensor, CREF is the reading of the reference instrument and n is the number of samples.

[0053] Step 5, such as Figure 2 As shown in the figure, the data from day 1 to n, day 1 to n+1, day 1 to n+2, etc. are used as training sets to establish calibration models for training sets of different days. The 30-day data after the training set is used as the test set, and the RMSE of the test set is calculated. 2 , as the evaluation index of the calibration effect, select RMSE from 2 The number of days in the training set is less than the observation requirement (the user's error requirement). Figure 3 shown.

[0054] Step 6: Change the comparison data and repeat step 5. Select RMSE from training sets of different days 2 The minimum number of training set days (N1) that is less than the observation requirement (user's error requirement) is taken as the optimal number of training set days.

[0055] Step 7: If the optimal number of days for the training set is N1, select N1 days of data as the training set and establish the calibration model. Use all data after the training set as the test set and calculate the daily root mean square error (RMSE) of the test set. 2 ). Find the RMSE 2 The time that lasts longer than the observation requirement is the time when the low-cost sensor needs to be calibrated again, and the time difference from the previous calibration is the interval between the two calibrations. Figure 4 , 5 shown.

[0056] Step 8: Replace the comparison data and repeat step 7. The average value of multiple interval times (N2) is taken as the optimal interval time between two calibrations.

[0057] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0058] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform a process suitable for low-cost CO 2 Optimization method for sensor calibration scheme.

[0059] A computing device includes one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including a program for executing a low-cost CO 2Instructions for optimization methods of sensor calibration schemes.

[0060] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0062] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0063] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0064] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. An optimization method for low-cost CO2 sensor calibration scheme, characterized by: Step A1, using the comparison observation data of the low-cost CO2 sensor and the high-precision reference instrument for n consecutive days as a training set to establish a regression equation of the observation error with respect to environmental factors; Step A2: Use the data of several consecutive days after the training set as the test set; Step A3, calculating the root mean square error of the calibrated low-cost sensor of the test set; Step A4, determining whether the root mean square error of the calibrated low-cost sensor of the test set is less than the user's requirement on the error; Step A5: If the root mean square error of the low-cost sensor after calibration of the test set is less than the user's error requirement, then proceed to step A6; otherwise, n=n+1, then return to step A1; Step A6: determine whether the environmental factors in the training set cover the environmental factors of the low-cost CO2 sensor application scenario; Step A7: If the environmental factors in the training set cover the environmental factors of the low-cost CO2 sensor application scenario, Then the output N1=n is the shortest number of consecutive days required for the training set data; Otherwise, replace the low-cost CO2 sensor with the high-precision reference instrument for n consecutive days of comparison observation data as the training set and return to step A1; The root mean square error of the low-cost sensor after calibration in step A3 In the formula, C L ' CS is the calibration value of the low-cost CO2 sensor, C REF is the reading of the reference instrument, n is the number of samples; Also includes: Step B1, using the comparison observation data of low-cost CO2 sensors and high-precision reference instruments for N1 consecutive days as a training set to establish a regression equation of observation error with respect to environmental factors; Step B2, using the comparison observation data of the low-cost CO2 sensor and the high-precision reference instrument for consecutive days after day N1 as the test set; Step B3, calculate the daily RMSE2 starting from the first day corresponding to the test set; Step B4, judging whether RMSE2 continues to be greater than the user's error requirement after the nth day; Step B5, if RMSE2 continues to be greater than the user's error requirement after the nth day, then n is the interval between two calibrations of the low-cost CO2 sensor, and the process goes to step B6; otherwise, n=n+1 and the process goes back to step B4; Step B6: determine whether the environmental factors in the training set cover the environmental factors of the low-cost CO2 sensor application scenario; Step B7: If the environmental factors in the training set cover the environmental factors of the low-cost CO2 sensor application scenario, Then the average value N2 of the multiple interval times output is the optimal time interval between two calibrations; Otherwise, replace the low-cost CO2 sensor with the high-precision reference instrument for N1 consecutive days of comparison observation data as the training set and return to step B1.

2. The optimization method for low-cost CO2 sensor calibration solution according to claim 1, characterized in that: In step A2, the data of 30 consecutive days after the training set is used as the test set.

3. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions which, when executed by a computing device, cause the computing device to perform any one of the methods according to claims 1 to 2.

4. A computing device, characterized in that include: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the methods according to claims 1 to 2.

Images