Photometer detection temperature correction method and system based on machine learning

Abstract

The present application discloses a method and system for temperature correction of photometer detection based on machine learning, which belongs to the field of data processing technology. The present application fully grasps the working environment information of the photometer by obtaining atmospheric state inversion data, channel irradiation relationship data and temperature environment change data. The atmospheric state correction viewpoint determined based on the atmospheric state inversion data can weigh the measurement results according to the atmospheric state confidence coefficient to avoid measurement errors caused by unclear atmospheric state. The relationship correction strategy determined from the channel irradiation relationship data can accurately correct the channel irradiation disturbance under the temperature environment and ensure the accuracy of the channel irradiation measurement. The temperature-sensitive correlation strategy determined with the help of temperature environment change data can be adjusted for different types of temperature-sensitive disturbances. The temperature correction strategy integrates multiple factors to effectively eliminate the influence of temperature on the measurement of multi-spectral sunphotometer, making its application data in the fields of atmosphere, meteorology and environmental science more accurate and reliable.

Description

Technical Field

[0001] The present application belongs to the field of data processing technology, and specifically relates to a method and system for correcting temperature detected by a photometer based on machine learning. Background Art

[0002] In the application of multispectral sun photometers, temperature has a significant impact on their measurement accuracy. Traditional multispectral sun photometers have many problems during measurement due to the lack of an effective mechanism to comprehensively consider various factors. Changes in atmospheric conditions will interfere with measurement results, but existing technologies have difficulty accurately judging the extent of the impact of atmospheric conditions on measurements, and are unable to determine the confidence coefficient of atmospheric conditions, making it impossible to reasonably weigh measurement results. In terms of channel irradiance, temperature changes will cause detection disturbances, but there is no method to accurately determine correction rules based on channel irradiance relationships. Moreover, for types of temperature-sensitive disturbances, due to the lack of targeted strategies, it is difficult to effectively deal with the impact of different temperature changes on the photometer. This results in low measurement data accuracy and poor reliability when multispectral sun photometers are used in fields such as atmospheric optical properties research, meteorological monitoring, and environmental science. Summary of the Invention

[0003] The present application provides a method and system for correcting temperature detected by a photometer based on machine learning, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0004] The embodiment of the present application provides a photometer detection temperature correction method based on machine learning, which is applied to a photometer detection temperature correction system. The method comprises: obtaining an environmental detection data set of a multispectral solar photometer to be corrected, wherein the environmental detection data set comprises atmospheric state inversion data, channel irradiation relationship data and temperature environment change data, wherein the atmospheric state inversion data comprises an inversion cluster indicator variable associated with the atmospheric state corresponding to the multispectral solar photometer, the channel irradiation relationship data comprises linear data associated with the channel irradiation of the multispectral solar photometer, and the temperature environment change data comprises model data associated with temperature disturbance in the multispectral solar photometer; based on the atmospheric state inversion data of the multispectral solar photometer to be corrected, determining an atmospheric state correction viewpoint of the multispectral solar photometer to be corrected, wherein the atmospheric state correction viewpoint of the multispectral solar photometer to be corrected is used to indicate a confidence coefficient of the atmospheric state corresponding to the multispectral solar photometer to be corrected. number; based on the channel irradiance relationship data of the multispectral solar photometer to be corrected, determine the relationship correction strategy of the multispectral solar photometer to be corrected from several initial relationship correction strategies, the relationship correction strategy of the multispectral solar photometer to be corrected is used to indicate the channel irradiance correction rule corresponding to the detection disturbance of the multispectral solar photometer to be corrected in the temperature environment; based on the temperature environment change data of the multispectral solar photometer to be corrected, determine the temperature-sensitive association strategy of the multispectral solar photometer to be corrected from several initial temperature-sensitive association strategies, the temperature-sensitive association strategy of the multispectral solar photometer to be corrected is used to indicate the type of temperature-sensitive disturbance corresponding to the multispectral solar photometer to be corrected; generate the temperature correction strategy of the multispectral solar photometer to be corrected, the temperature correction strategy of the multispectral solar photometer to be corrected includes the atmospheric state correction viewpoint, relationship correction strategy and temperature-sensitive association strategy of the multispectral solar photometer to be corrected.

[0005] An embodiment of the present application provides a photometer detection temperature correction system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the above-mentioned method.

[0006] An embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.

[0007] The embodiment of the present application obtains an environmental detection data set including atmospheric state inversion data, channel irradiation relationship data and temperature environment change data to fully grasp the working environment information of the photometer. Based on the atmospheric state correction point of view determined by the atmospheric state inversion data, the measurement results can be weighed according to the atmospheric state confidence coefficient to avoid measurement errors caused by unclear atmospheric state. The relationship correction strategy determined from the channel irradiation relationship data can accurately correct the channel irradiation disturbance under the temperature environment and ensure the accuracy of the channel irradiation measurement. With the help of temperature environment change data, the temperature-sensitive correlation strategy is determined, which can be adjusted for different types of temperature-sensitive disturbances. The temperature correction strategy finally generated integrates multiple factors, effectively eliminates the influence of temperature on the measurement of the multi-spectral sun photometer, and makes its application data in the fields of atmosphere, meteorology and environmental science more accurate and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 This is a flowchart of a method for correcting temperature detected by a photometer based on machine learning provided in an embodiment of the present application.

[0009] Figure 2 This is a structural diagram of a photometer detection temperature correction system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0011] The terms "first," "second," and the like in this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein. Furthermore, the objects distinguished by "first," "second," and the like generally refer to a class and do not limit the number of objects; for example, the first object can be one or more. Furthermore, the term "and / or" in this application refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0012] Figure 1 A method for correcting photometer detection temperature based on machine learning is shown, which is applied to a photometer detection temperature correction system. The method includes S101-S105.

[0013] S101: The photometer detection temperature correction system obtains an environmental detection data set of a multispectral solar photometer to be corrected, wherein the environmental detection data set includes atmospheric state inversion data, channel irradiation relationship data and temperature environment change data. The atmospheric state inversion data includes an inversion cluster indicator variable associated with the atmospheric state corresponding to the multispectral solar photometer, the channel irradiation relationship data includes linear data associated with the channel irradiation of the multispectral solar photometer, and the temperature environment change data includes model data associated with temperature disturbance in the multispectral solar photometer.

[0014] S102: The photometer detection temperature correction system determines the atmospheric state correction viewpoint of the multispectral sun photometer to be corrected based on the atmospheric state inversion data of the multispectral sun photometer to be corrected, and the atmospheric state correction viewpoint of the multispectral sun photometer to be corrected is used to indicate the confidence coefficient of the atmospheric state corresponding to the multispectral sun photometer to be corrected.

[0015] S103: The photometer detection temperature correction system determines the relationship correction strategy of the multispectral solar photometer to be corrected from several initial relationship correction strategies based on the channel irradiation relationship data of the multispectral solar photometer to be corrected. The relationship correction strategy of the multispectral solar photometer to be corrected is used to indicate the channel irradiation correction rules corresponding to the detection disturbance of the multispectral solar photometer to be corrected in the temperature environment.

[0016] S104: The photometer detection temperature correction system determines a temperature-sensitive association strategy of the multispectral solar photometer to be corrected from several initial temperature-sensitive association strategies based on the temperature environment change data of the multispectral solar photometer to be corrected. The temperature-sensitive association strategy of the multispectral solar photometer to be corrected is used to indicate the type of temperature-sensitive disturbance corresponding to the multispectral solar photometer to be corrected.

[0017] S105: The photometer detection temperature correction system generates a temperature correction strategy for the multispectral sun photometer to be corrected. The temperature correction strategy for the multispectral sun photometer to be corrected includes an atmospheric state correction viewpoint, a relationship correction strategy, and a temperature-sensitive correlation strategy for the multispectral sun photometer to be corrected.

[0018] In the embodiment of the present application, the photometer detection temperature correction system is the executor of the temperature correction of the multi-spectral sun photometer, and the technical solution based on S101-S105 will be exemplarily introduced and explained below.

[0019] In S101 , the photometer detection temperature correction system first obtains an environmental detection data set of a multispectral sun photometer to be corrected. The environmental detection data set includes but is not limited to the following three types of data.

[0020] 1) Atmospheric state inversion data

[0021] The atmospheric state inversion cluster indicator variables associated with multispectral sun photometers in atmospheric state inversion data contain a wealth of important information. For example, in a specific measurement scenario, such as atmospheric monitoring in the central area of ​​a city, a multispectral sun photometer is used to measure atmospheric parameters such as aerosol optical depth (AOD) and single scattering albedo (SSA). Different cluster indicator variables may be used for AOD inversion. For example, when certain types of aerosols, such as fine particulate matter from industrial emissions, are present in the atmosphere, under certain meteorological conditions, they may form a cluster of aerosols with specific optical properties. If the AOD values ​​measured by the multispectral sun photometer in this area fall within a certain range, such as 0.3-0.5, this range can be used as part of the inversion cluster indicator variable. This reflects the characteristic performance of the multispectral sun photometer's AOD measurements under these specific atmospheric conditions.

[0022] Looking at single-scatter albedo, for example, it varies across seasons due to factors like vegetation cover and atmospheric humidity. In spring, when vegetation begins to grow and atmospheric water vapor levels are relatively moderate, the single-scatter albedo may range from 0.8 to 0.9. This numerical range correlates with the atmospheric state during spring and forms part of the atmospheric state inversion data, helping to more comprehensively describe the multispectral sun photometer's measurements under different atmospheric conditions.

[0023] 2) Channel irradiation relationship data

[0024] The linear data associated with channel irradiance in a multispectral sunphotometer within the channel irradiance relationship data is crucial for understanding how irradiance varies across different channels under the influence of temperature. For example, when measuring the 1020nm channel of the CW193 multispectral sunphotometer under temperature fluctuations, multiple measurements revealed that within a certain temperature range, the irradiance exhibited a near linear relationship with temperature. When the temperature increased from 20°C to 30°C, the irradiance increased from the initial value I1 to I2, maintaining a specific linear relationship. For example, for every 1°C increase in temperature, the irradiance increased by 0.005. This linear relationship data is a crucial component of the channel irradiance relationship data.

[0025] A similar situation exists for the 1640nm channel. For example, in another temperature range, such as between -5°C and 5°C, the irradiance increases with increasing temperature. When the temperature rises from -5°C to 0°C, the irradiance increases from J1 to J2. Measurement and analysis reveal a linear relationship: for every 1°C increase in temperature, the irradiance increases by 0.003. These linear data for different channels provide the basis for subsequent relationship correction strategies.

[0026] 3) Temperature environment change data

[0027] The model data associated with temperature perturbations in a multispectral sunphotometer within the temperature environment reflects the complex impact of temperature on photometric measurements. Taking the CW193 as an example, the detector response of the 1020nm channel is generally low when the temperature is below 25°C, and this response decreases and then increases with temperature throughout the day. For example, in the morning at 15°C, the detector response decreases by 10% relative to its standard response at 25°C. As the temperature rises to 20°C, the response recovers, but remains below the standard response. This temperature-dependent variation can be represented by a model.

[0028] The 1640nm channel exhibits a more pronounced underestimation of detector response at temperatures below 0°C. For example, at -5°C, the detector's response may be only 60% of its normal response at 10°C. At higher temperatures, similar to the 1020nm channel, it exhibits characteristics that are opposite to those observed at low temperatures. Model data related to temperature perturbations, such as the proportional relationship between detector response and standard values ​​at different temperatures, are crucial components of temperature environment variation data.

[0029] In S102, determining a correction viewpoint for the atmospheric state of the multispectral photometer to be corrected includes determining a correction viewpoint for the atmospheric state of the multispectral photometer to be corrected based on atmospheric state inversion data. This viewpoint is used to indicate a confidence coefficient for the atmospheric state corresponding to the multispectral photometer to be corrected. For example, in measuring atmospheric aerosol optical depth (AOD), if the AOD value measured by the multispectral photometer in a certain area is compared with data in a long-established AOD database for that area, for example, the long-term average AOD value for that area is around 0.4. Under specific meteorological conditions, the AOD value measured by the multispectral photometer is 0.41, and the fluctuation range of multiple measurements is within ±0.02. This indicates that the measurement result is very close to the long-term average value. In this case, the confidence coefficient for the atmospheric state can be set to a higher value, such as 0.9.

[0030] Conversely, if the measured AOD value is 0.6, which deviates significantly from the long-term average and fluctuates widely across multiple measurements (for example, within ±0.1), this means that there are interfering factors in the current atmospheric conditions that affect the accuracy of the measurement. Therefore, the confidence coefficient for the atmospheric conditions may be lower, such as 0.6. This confidence coefficient reflects the reliability of the multispectral sun photometer's measurement results under the current atmospheric conditions.

[0031] When studying the impact of atmospheric clouds on solar radiation, multispectral sunphotometers measure cloud optical thickness and other parameters. If the measured cloud optical thickness matches that inferred from satellite observations and ground-based meteorological station data—for example, if the difference is within 5%—then the confidence factor for the atmospheric state can be higher, such as 0.85. If the difference is larger, exceeding 15%, the confidence factor is lower, such as 0.5. This helps to comprehensively consider the impact of atmospheric conditions on measurement results when formulating temperature correction strategies.

[0032] Furthermore, S103 involves determining a relationship correction strategy for the multi-spectral sun photometer to be corrected. In the embodiment of the present application, a suitable strategy can be selected from the initial relationship correction strategy: based on the channel irradiance relationship data of the multi-spectral sun photometer to be corrected, a suitable strategy is determined from a number of initial relationship correction strategies. Taking the 1020nm channel of CW193 as an example, there are three types of initial relationship correction strategies: linear correction strategy, quadratic function correction strategy, and piecewise linear correction strategy. According to the channel irradiance relationship data obtained previously, when it is found that the irradiance of the 1020nm channel and the temperature in a larger temperature range (such as 15°C-35°C) show an obvious linear relationship, for example, for every 1°C increase in temperature, the irradiance increases by 0.005. In this case, a linear correction strategy can be selected.

[0033] For the 1640nm channel, if different relationships are present in different temperature ranges, for example, between -10°C and 0°C, irradiance and temperature exhibit a more complex relationship that, after analysis, is more appropriately described by a quadratic function, such as irradiance = aT² + bT + c (where a, b, and c are coefficients fitted from measured data), while a linear relationship is present between 0°C and 20°C, a segmented correction strategy can be adopted for different temperature ranges: a quadratic function-based correction strategy between -10°C and 0°C, and a linear correction strategy between 0°C and 20°C.

[0034] For example, in a channel of another multispectral sun photometer, extensive measurements revealed a strong correlation between irradiance and temperature squared across the entire measurement temperature range (-20°C to 40°C). For example, the proportionality coefficient between irradiance and temperature squared is 0.001. In this case, a correction strategy based on a quadratic function (correlated with the square of temperature) can be used to accurately correct for temperature-related disturbances in the channel's irradiance.

[0035] Next, S104 focuses on determining a temperature-sensitive correlation strategy for the multispectral sunphotometer to be corrected. Based on the temperature and environmental change data of the multispectral sunphotometer to be corrected, a suitable strategy is determined from several initial temperature-sensitive correlation strategies. Taking the 1020nm channel of the CW193 as an example, the detector response of this channel is generally low when the temperature is below 25°C, and it shows a trend of first decreasing and then increasing with temperature changes throughout the day. This is a specific type of temperature-sensitive disturbance. To address this situation, a temperature-sensitive correlation strategy can be developed. For example, when the temperature is below 20°C, a larger correction measure is adopted, multiplying the measured value by a correction factor. This correction factor can be determined based on actual measurement data, such as 1.1, to compensate for the underestimation of the detector response. Between 20°C and 25°C, a smaller correction measure with a correction factor of 1.05 is adopted.

[0036] For the 1640nm channel, the detector response will be more significantly underestimated when the temperature is below 0°C. To address this temperature-sensitive perturbation, a special temperature-sensitive correlation strategy can be developed. For example, when the temperature is below -5°C, a larger correction is used, with a correction factor of 1.2. Between -5°C and 0°C, the correction factor is 1.15. When the temperature is higher, based on its characteristics opposite to low temperatures, such as between 10°C and 30°C, the detector response increases with increasing temperature, and the increase is relatively stable. In this temperature range, it can be formulated that for every 5°C increase in temperature, the measured value is multiplied by a correction factor slightly less than 1, such as 0.98, to adjust for the overestimation that may occur due to the increase in temperature.

[0037] Finally, based on S105 , a temperature correction strategy for the multispectral sun photometer to be corrected may be generated. An exemplary technical approach may be to integrate various parts of the temperature correction strategy.

[0038] The photometer detection temperature correction system generates a temperature correction strategy for the multispectral sunphotometer to be corrected. This strategy incorporates the previously determined atmospheric state correction viewpoint, relationship correction strategy, and temperature-sensitive correlation strategy. For example, for a multispectral sunphotometer under specific atmospheric and temperature conditions, the atmospheric state correction viewpoint provides a confidence coefficient of 0.8 (indicating a certain degree of reliability in the measurement results under these atmospheric conditions), the relationship correction strategy for the 1020nm channel is a linear correction strategy (increasing irradiance by 0.005 for every 1°C increase in temperature), and the temperature-sensitive correlation strategy uses a correction coefficient of 1.1 for temperatures below 20°C and a correction coefficient of 1.05 for temperatures between 20°C and 25°C.

[0039] In the actual temperature correction process, the reliability weight of the overall measurement result is first determined based on the confidence coefficient from the atmospheric state correction perspective. If the confidence coefficient is high, such as close to 1, adjustments to atmospheric state-related factors can be relatively reduced in subsequent temperature corrections. Next, the channel irradiance is corrected according to the relationship correction strategy. For example, for the 1020nm channel, the irradiance is adjusted based on temperature changes and linear relationships. Finally, the detector response is adjusted according to the temperature-sensitive correlation strategy, for example, multiplying the measured value by a correction factor of 1.1 when the temperature is below 20°C.

[0040] In meteorological monitoring, for example, multispectral sunphotometers are used to monitor the absorption of solar radiation by atmospheric water vapor. The atmospheric state correction viewpoint indicates a confidence coefficient of 0.75 under current meteorological conditions. The relationship correction strategy for a particular channel is a quadratic function correction strategy (determined based on channel irradiance relationship data), and the temperature-sensitive correlation strategy adjusts the correction coefficient within a specific temperature range. This comprehensive temperature correction strategy allows for more accurate measurement of the true absorption of solar radiation by water vapor content, eliminating temperature interference. In environmental science research, such as studying the effects of atmospheric pollutants on solar radiation scattering, this temperature correction strategy ensures the accuracy of multispectral sunphotometer measurement data, enabling more precise analysis of the relationship between atmospheric pollutants and solar radiation.

[0041] Another example is a long-term atmospheric optical properties research project involving measurements in multiple regions and different seasons. In different measurement scenarios, the temperature correction strategy of the multispectral sunphotometer will vary depending on the prevailing atmospheric state, channel irradiance relationship, and temperature and environmental changes. In coastal areas in summer, the atmospheric state correction perspective may have a higher confidence coefficient due to the relative stability of the atmosphere. The relationship correction strategy may adopt a linear correction strategy based on the channel irradiance relationship data, and the temperature-sensitive correlation strategy may be adjusted according to the temperature variation range and detector response characteristics. In industrial cities in winter, the confidence coefficient of the atmospheric state correction perspective may be lower due to factors such as the high concentration of atmospheric pollutants. The relationship correction strategy and temperature-sensitive correlation strategy will also be adjusted accordingly according to the local temperature and channel irradiance conditions to ensure the accuracy and reliability of the multispectral sunphotometer's measurement data in different environments.

[0042] It can be seen that the photometer detection temperature correction system, through S101-S105, comprehensively considers multiple factors such as the atmospheric state, channel irradiation relationship, and temperature environment changes of the multispectral sunphotometer, and accurately generates a temperature correction strategy. This is of great significance for improving the measurement accuracy and reliability of multispectral sunphotometers in the fields of atmospheric optical properties research, meteorological monitoring, environmental science, etc.

[0043] In summary, the embodiments of the present application can significantly improve the accuracy and reliability of multi-spectral sun photometer measurements: by obtaining an environmental detection data set containing atmospheric state inversion data, channel irradiance relationship data, and temperature environment change data, the working environment information of the photometer can be fully grasped. Based on the atmospheric state correction point of view determined by the atmospheric state inversion data, the measurement results can be weighed according to the atmospheric state confidence coefficient to avoid measurement errors caused by unclear atmospheric state. The relationship correction strategy determined from the channel irradiance relationship data can accurately correct the channel irradiance disturbance under the temperature environment and ensure the accuracy of the channel irradiance measurement. With the help of temperature environment change data, the temperature-sensitive correlation strategy is determined, which can be adjusted for different types of temperature-sensitive disturbances. The temperature correction strategy finally generated integrates multiple factors, effectively eliminating the influence of temperature on the measurement of the multi-spectral sun photometer, making its application data in many fields such as atmospheric optical properties research, meteorological monitoring, and environmental science more accurate and reliable, and improving the quality of related research and monitoring.

[0044] In some optional embodiments, the relationship correction strategy of the multispectral solar photometer to be corrected is determined from several initial relationship correction strategies based on the channel irradiation relationship data of the multispectral solar photometer to be corrected, including: calculating the commonality matching weights between the channel irradiation relationship data of the multispectral solar photometer to be corrected and the several initial relationship correction strategies; and determining the initial relationship correction strategy with the largest commonality matching weight with the channel irradiation relationship data of the multispectral solar photometer to be corrected as the relationship correction strategy of the multispectral solar photometer to be corrected.

[0045] In the embodiment of the present application, the step of determining a relationship correction strategy from a plurality of initial relationship correction strategies based on the channel irradiance relationship data of the multispectral sun photometer to be corrected is crucial in the temperature correction process.

[0046] First, regarding channel irradiance relationship data, taking the CW193 multispectral sun photometer as an example, its different channels, such as the 1020nm and 1640nm channels, have different irradiance relationships. For the 1020nm channel, within a specific temperature range, for example, from 15°C to 30°C, measurements found that its irradiance exhibits a certain pattern of change as the temperature increases. When the temperature rises from 15°C to 20°C, the irradiance increases from the initial value I1 to I2. When the temperature continues to rise to 25°C, the irradiance changes to I3, and from 25°C to 30°C, it changes to I4. There is an inherent linear relationship or other complex relationship between these irradiance values ​​at different temperatures. The 1640nm channel also exhibits its own unique irradiance relationship across different temperature ranges. For example, within the -5°C to 10°C range, its irradiance changes in a pattern: when the temperature rises from -5°C to 0°C, the irradiance changes from J1 to J2, from 0°C to 5°C to J3, and from 5°C to 10°C to J4. These channel irradiance relationship data form the basis for subsequent relationship correction strategies.

[0047] Next, consider several initial relationship correction strategies. For example, there are three initial relationship correction strategies: linear correction strategy, quadratic function correction strategy, and piecewise linear correction strategy. For the linear correction strategy, there is a linear relationship between the channel irradiance relationship and temperature, such as the change in irradiance is in a fixed proportion to the change in temperature. The quadratic function correction strategy assumes that the channel irradiance relationship and temperature satisfy a quadratic function relationship, which may be similar to the relationship of irradiance = aT² + bT + c. The piecewise linear correction strategy divides the temperature range into different temperature intervals and uses different linear relationships to describe the channel irradiance relationship in different temperature intervals.

[0048] The commonality matching weight is then calculated. Taking the 1020nm channel as an example, the commonality matching weight between the channel irradiance relationship data and the linear correction strategy is calculated. For example, through relevant measurement data and algorithm analysis, if the measured irradiance change and the irradiance change predicted by the linear correction strategy are relatively close at multiple temperature points within the temperature range of 15°C to 30°C, for example, the difference between the irradiance measured at 18°C ​​and the irradiance predicted by the linear correction strategy is within 0.01, the difference at 22°C is within 0.008, and the difference at 28°C is within 0.005. By combining the conditions at multiple temperature points, a weight value representing the degree of commonality matching with the linear correction strategy can be calculated, such as 0.8 (indicating a high degree of matching).

[0049] Similarly, the commonality match weight between the 1020nm channel irradiance relationship data and the quadratic function correction strategy was calculated. Within the temperature range of 15°C to 30°C, there was a significant discrepancy between the actual measured irradiance variation and the irradiance variation predicted by the quadratic function correction strategy. For example, at 16°C, the difference between the measured and predicted values ​​reached 0.05, at 20°C the difference was 0.04, and at 25°C the difference was 0.03. Taking these differences into account, the commonality match weight with the quadratic function correction strategy was calculated to be 0.3 (indicating a low degree of match).

[0050] The commonality matching weight between the 1020nm channel irradiation relationship data and the piecewise linear correction strategy was then calculated. For example, the piecewise linear correction strategy divides the temperature range from 15°C to 30°C into three intervals: 15°C-20°C, 20°C-25°C, and 25°C-30°C. Within the 15°C-20°C interval, there is a certain deviation between the measured and predicted values ​​at certain temperature points. The deviation is relatively small within the 20°C-25°C interval, and increases significantly within the 25°C-30°C interval. Taking into account the situation across the entire temperature range, the commonality matching weight with the piecewise linear correction strategy is calculated to be 0.6.

[0051] By comparing the three commonality matching weights of 0.8, 0.3, and 0.6, it can be seen that the linear correction strategy has the largest commonality matching weight with the irradiation relationship data of the 1020nm channel. Therefore, the linear correction strategy is determined as the relationship correction strategy for the 1020nm channel.

[0052] The same calculation method was used for the 1640nm channel. For example, within the temperature range of -5°C to 10°C, the commonality matching weights for the three relationship correction strategies were calculated. The commonality matching weight calculated with the linear correction strategy was 0.5, the commonality matching weight calculated with the quadratic function correction strategy was 0.7, and the commonality matching weight calculated with the piecewise linear correction strategy was 0.4. Through comparison, the quadratic function correction strategy was determined to be the relationship correction strategy for the 1640nm channel.

[0053] In the practical application of multispectral sunphotometers, different channels require precise calibration strategies to ensure measurement accuracy under varying temperature environments. By calculating commonality matching weights to determine the relationship calibration strategy, we can more accurately adapt to the irradiance relationship characteristics of different channels. Using an inappropriate relationship calibration strategy can lead to increased measurement deviation. For example, if a quadratic function calibration strategy is mistakenly used for the 1020nm channel, rather than a linear calibration strategy, subsequent measurements may result in significant deviations from the actual atmospheric parameters (such as aerosol optical depth). For example, when a linear calibration strategy is correctly used, the error between the measured aerosol optical depth and the actual value is within 0.02, while when a quadratic function calibration strategy is incorrectly used, the error may exceed 0.1 (the values ​​here are merely illustrative). Similarly, for the 1640nm channel, if the relationship calibration strategy is not accurately determined, measurement results can be adversely affected, compromising the accuracy of research in atmospheric optical properties, meteorological monitoring, and environmental science.

[0054] This design calculates the common matching weights between channel irradiance relationship data and different initial relationship correction strategies, enabling the precise selection of the most appropriate relationship correction strategy. This helps reduce measurement errors caused by improper relationship correction strategy selection. For example, in measurements of different channels, measurement errors can be reduced from potentially large values ​​(e.g., greater than 0.1) to smaller values ​​(e.g., within 0.02), thereby improving the data accuracy and reliability of multispectral sun photometers in various related research fields.

[0055] Under some other optional technical ideas, the method also includes: obtaining channel irradiation relationship data corresponding to several reference multispectral solar photometers, wherein the reference multispectral solar photometer refers to a multispectral solar photometer that has completed the correction operation; clustering the channel irradiation relationship data corresponding to the several reference multispectral solar photometers to obtain at least one channel irradiation cluster data cluster, each channel irradiation cluster data cluster including channel irradiation relationship data of at least one reference multispectral solar photometer; for each channel irradiation cluster data cluster, based on the channel irradiation relationship data of the reference multispectral solar photometer included in the channel irradiation cluster data cluster, determining the relationship correction strategy corresponding to the channel irradiation cluster data cluster.

[0056] It can be understood that in the initial step of obtaining channel irradiance relationship data corresponding to several reference multispectral photometers, since the reference multispectral photometers have already undergone calibration, the accuracy and reliability of their data are crucial for subsequent operations. For example, consider a set of 10 reference multispectral photometers, labeled Photometer 1, Photometer 2, Photometer 3, and so on. Taking Photometer 1 as an example, it has multiple channels. First, consider the 1020nm channel. Over a wide temperature range, such as from 5°C to 35°C, the channel irradiance relationship data exhibits a series of complex yet orderly changes. At 5°C, its channel irradiance value is P1. When the temperature rises to 10°C, this value becomes P2. There is a difference between P2 and P1 of perhaps 0.03 (the actual value will vary depending on the specific photometer characteristics and measurement environment). Next, when the temperature rises to 15°C, the irradiance value becomes P3, and the difference between P3 and P2 changes again, for example, to 0.04. As the temperature continues to rise to 20℃, 25℃, 30℃ and 35℃, the irradiation values ​​become P4, P5, P6 and P7 respectively. The differences in irradiation values ​​at adjacent temperatures fluctuate within a certain range. These fluctuation ranges reflect the irradiation relationship characteristics of the channel in this temperature range.

[0057] Looking at the 1640nm channel of photometer 1, within another temperature range, for example, -10°C to 10°C, the variation pattern of its irradiance values ​​is also unique. At -10°C, the irradiance value is Q1. When the temperature rises to -5°C, it becomes Q2, with the difference between Q2 and Q1 being 0.05. As the temperature continues to rise to 0°C, 5°C, and 10°C, the irradiance values ​​become Q3, Q4, and Q5, respectively. The difference in irradiance values ​​at each adjacent temperature also varies within a certain range, reflecting the unique irradiance relationship pattern of this channel within this temperature range.

[0058] Other reference multispectral sunphotometers, such as Photometers 2 through 10, also have their own irradiance relationship data patterns for different channels in different temperature ranges. For example, Photometer 2's 1020nm channel has an irradiance value of R1 at 12°C, changing to R2 at 16°C, with the R2-R1 difference being 0.025. At 20°C, it's R3, with the R3-R2 difference being 0.03, and so on. For the 1640nm channel, over the -5°C to 8°C temperature range, it has an irradiance value of S1 at -5°C, changing to S2 at 0°C, with the S2-S1 difference being 0.04, and S3 at 4°C, with the S3-S2 difference being 0.035, and so on. This rich channel-to-channel irradiance relationship data forms the foundational data pool for subsequent operations.

[0059] Next, the channel irradiance relationship data corresponding to these reference multispectral sun photometers is clustered. This process aims to group similar data together based on their inherent characteristics, thereby generating at least one channel irradiance cluster. Clustering is based on many factors, not just simple numerical similarity but also complex factors such as temperature range characteristics, irradiance value trends, and the relative change ratio of irradiance values ​​at different temperatures.

[0060] Taking the 1020nm channel as an example, we carefully analyze the channel irradiance relationship data for Photometers 1, 3, 5, and 7. Within a specific temperature range, say 18°C ​​to 26°C, the channel irradiance values ​​of these photometers show strikingly similar trends as the temperature rises. At 18°C, the irradiance value of Photometer 1 is P4, and that of Photometer 3 is T1, with a difference of 0.01. When the temperature rises to 20°C, the irradiance value of Photometer 1 changes to P5, and that of Photometer 3 to T2. The difference between T2-T1 and P5-P4 is within 0.005. As the temperature continues to rise to 22°C, 24°C, and 26°C, the difference between the changes in irradiance values ​​of the two photometers remains within a very small range, such as within 0.003. Similarly, for photometers 5 and 7, within this temperature range, their irradiance values ​​show very similar trends and amounts of change compared to those of photometers 1 and 3. For example, at 22°C, the irradiance value of photometer 5 differs by 0.008 from that of photometer 1, and the irradiance value of photometer 7 differs by 0.006 from that of photometer 3. Based on these data features, the irradiance relationship data for the 1020nm channel of photometers 1, 3, 5, and 7 within the temperature range of 18°C ​​to 26°C can be grouped into a single channel irradiance cluster.

[0061] Considering the 1640nm channel, consider the channel irradiance relationship data for Photometers 2, 4, 6, and 8. Within the temperature range of -3°C to 6°C, they exhibit similar irradiance relationship characteristics. At -3°C, the irradiance value for Photometer 2 is S1, and that for Photometer 4 is U1, with a difference of 0.015. When the temperature rises to 0°C, the irradiance value for Photometer 2 changes to S2, and that for Photometer 4 to U2. The differences between U2-U1 and S2-S1 are within 0.008. As the temperature continues to rise to 3°C and 6°C, the irradiance values ​​for these photometers exhibit very similar patterns of change, with the difference in irradiance values ​​at adjacent temperatures within 0.005. Therefore, the irradiance relationship data for the 1640nm channel of these four photometers within the temperature range of -3°C to 6°C can form another channel irradiance cluster.

[0062] For each channel irradiance cluster data cluster, determining the relationship correction strategy corresponding to the channel irradiance cluster data cluster based on the channel irradiance relationship data of the reference multispectral sun photometer included in the channel irradiance cluster data cluster is the key link of the entire technical solution.

[0063] Taking the 1020nm channel irradiance cluster data for photometers 1, 3, 5, and 7 as an example, the temperature range for this cluster is 18°C ​​to 26°C. A closer analysis of the channel irradiance relationship data reveals that the channel irradiance values ​​for these photometers exhibit a nearly linear growth pattern with increasing temperature. From 18°C ​​to 20°C, the irradiance value for photometer 1 increases from P4 to P5 by ΔP1; the irradiance value for photometer 3 increases from T1 to T2 by ΔT1. The ratio of ΔT1 to ΔP1 is close to 1 (for example, 0.98, indicating a very close linear relationship). Similarly, when the temperature rises from 20°C to 22°C, the proportional relationship between the irradiance increases for each photometer is also very close to 1. For example, the ratio of the increase in photometer 5 to that in photometer 1 is 0.99, and the ratio of the increase in photometer 7 to that in photometer 3 is 1.01. This characteristic of the stable and close-to-1 ratio of the irradiance increase between photometers at different temperature intervals strongly suggests that the relationship correction strategy for this channel irradiance cluster data cluster should be a linear correction strategy.

[0064] Consider the 1640nm channel irradiance cluster data cluster, which includes photometers 2, 4, 6, and 8. The corresponding temperature range is -3°C to 6°C. Analysis of the channel irradiance relationship data within this cluster reveals that the relationship between irradiance values ​​and temperature is not a simple linear one. At -3°C, the irradiance values ​​of each photometer can be used to calculate certain relevant parameters, such as coefficient a1, which exhibits a quadratic function-like relationship. When the temperature rises to 0°C, the coefficient a2 calculated based on the new irradiance value exhibits a nonlinear relationship with a1, which exhibits some characteristics of a quadratic function. When the temperature continues to rise to 3°C and 6°C, the coefficients calculated based on the irradiance values ​​also follow a similar quadratic function pattern. Based on these characteristics of the channel irradiance relationship data, it can be determined that the relationship correction strategy for this channel irradiance cluster data cluster is a quadratic function.

[0065] This cluster-based relationship correction strategy is crucial for the practical application of multispectral sunphotometers. In real-world measurement scenarios, channel irradiance relationship data from different multispectral sunphotometers can vary significantly due to factors such as manufacturing process, operating environment, and the aging of internal components. If a fixed relationship correction strategy is used instead of clustering, the correction effect is likely to be poor, leading to significant measurement errors.

[0066] For example, for the 1020nm channel, if clustering is not performed, a group of photometers that should use a linear correction strategy will be calibrated using a quadratic function correction strategy. For example, in a specific atmospheric environment, when the linear correction strategy is accurately used, the error between the measured aerosol optical density and the actual value is within 0.02. However, due to the incorrect use of the quadratic function correction strategy, the error between the measured aerosol optical density and the actual value may increase to more than 0.1. This is because the calibration mode of the quadratic function correction strategy does not match the actual channel irradiance relationship of the group of photometers, resulting in over-correction or under-correction of the irradiance value during the calculation process, which seriously affects the accuracy of the measurement results.

[0067] Similarly, for the 1640nm channel, similar issues can arise if accurate clustering and a suitable relationship correction strategy are not implemented. For example, calibrating a group of photometers suitable for a quadratic function correction strategy using a linear correction strategy can lead to significant deviations in the measurement results when measuring the impact of certain atmospheric components on solar radiation. For example, when a quadratic function correction strategy is correctly implemented, the error between the measured parameter related to solar radiation and the actual value is within 0.03. However, when a linear correction strategy is incorrectly implemented, the error may increase to over 0.15.

[0068] This cluster-based relationship correction strategy allows for precise correction of different types of channel irradiance relationship data, significantly improving measurement accuracy and reliability. This is crucial for the application of multispectral sunphotometers in numerous fields, including atmospheric optical properties research, meteorological monitoring, and environmental science. In atmospheric optical properties research, accurate measurement data can help scientists more precisely analyze the absorption and scattering mechanisms of solar radiation by various atmospheric components, leading to a deeper understanding of the changing patterns of atmospheric optical properties. In meteorological monitoring, precise multispectral sunphotometer measurement data can improve the accuracy of atmospheric condition monitoring, enabling more accurate predictions of weather changes and monitoring of the diffusion of atmospheric pollutants. In environmental science, reliable measurement data can help assess the impact of human activities on the environment, such as analyzing changes in the atmospheric environment caused by industrial emissions.

[0069] In this way, by clustering the channel irradiance relationship data of the reference multispectral sunphotometer and then determining the relationship correction strategy based on the clustering, the accuracy of the relationship correction strategy is significantly improved. This improvement in accuracy helps to significantly reduce measurement errors caused by improper correction strategies. For example, for certain channels, the measurement error that could originally be as high as 0.1 or more can be reduced to within 0.02. This has a great promoting effect on improving the data accuracy and reliability of multispectral sunphotometers in the application of atmospheric optical properties research, meteorological monitoring, environmental science and other fields.

[0070] Based on the above technical ideas, the channel irradiance relationship data corresponding to the several reference multispectral sun photometers are clustered to obtain at least one channel irradiance cluster data cluster, including: determining the channel irradiance correlation between every two reference multispectral sun photometers in the several reference multispectral sun photometers, the channel irradiance correlation is used to indicate the commonality matching weight between the channel irradiance relationship data corresponding to the two reference multispectral sun photometers; based on the channel irradiance correlation, the detected thermal values ​​of the several reference multispectral sun photometers and the first cluster indicator variable, determining A first cluster decision feature is determined, where the first cluster decision feature is used to adjust the channel irradiance correlation between the reference multispectral photometers belonging to the same channel irradiance cluster data cluster, and the first cluster indicator variable refers to the cycle round of the clustering process; based on the first cluster decision feature and the second cluster decision feature, the channel irradiation relationship data corresponding to the plurality of reference multispectral photometers are clustered to obtain the at least one channel irradiation cluster data cluster, and the second cluster decision feature is used to adjust the detected thermal value of the reference multispectral photometer included in each channel irradiation cluster data cluster.

[0071] Furthermore, the channel irradiation relationship data corresponding to the several reference multispectral sun photometers are clustered based on the first cluster decision feature and the second cluster decision feature to obtain the at least one channel irradiation cluster data cluster, including: in the u-th cycle, the channel irradiation relationship data corresponding to the several reference multispectral sun photometers are clustered based on the first cluster decision feature and the second cluster decision feature to obtain at least one initial channel irradiation cluster data cluster; wherein u represents the first cluster indicator variable, and the default value of u is 1; calculating the channel irradiation quality characteristic value of the first channel irradiation cluster data cluster in the at least one initial channel irradiation cluster data cluster, the first channel irradiation cluster data cluster refers to the set with the largest detected thermal values ​​of the reference multispectral sun photometers included in the at least one initial channel irradiation cluster data cluster, and the channel irradiation quality characteristic value is used to characterize the u-th cycle. A noise correction weight of a channel irradiation cluster data cluster; on the basis that the channel irradiation quality characteristic value is greater than a first set characteristic value, determining the at least one initial channel irradiation cluster data cluster as the at least one channel irradiation cluster data cluster; on the basis that the channel irradiation quality characteristic value is less than the first set characteristic value, and u is less than a second set characteristic value, setting u=u+1, and again executing the step of clustering the channel irradiation relationship data corresponding to the several reference multispectral sun photometers based on the first cluster decision feature and the second cluster decision feature to obtain at least one initial channel irradiation cluster data cluster; on the basis that the channel irradiation quality characteristic value is less than the first set characteristic value, and u is equal to the second set characteristic value, determining the at least one initial channel irradiation cluster data cluster obtained in a cycle process with the largest channel irradiation quality characteristic value as the at least one channel irradiation cluster data cluster.

[0072] In the above technical solution, the focus is first on determining the channel irradiance correlation between each pair of reference multispectral photometers. Consider a large set of reference multispectral photometers, say 20, labeled A1 through A20. Taking the 1020nm channel as an example, the channel irradiance relationship data for photometers A1 and A2 is examined over a wide temperature range, such as 10°C to 40°C. At 10°C, the channel irradiance value for A1 is set to X1. When the temperature rises to 15°C, its value changes to X2, X3 at 20°C, X4 at 25°C, X5 at 30°C, X6 at 35°C, and X7 at 40°C. For photometer A2, the channel irradiance values ​​at the same temperature are Y1, Y2, Y3, Y4, Y5, Y6, and Y7, respectively.

[0073] To determine the channel irradiance correlation between them, or the commonality matching weight, multiple factors must be considered. At 10°C, the difference between X1 and Y1 might be 0.02, at 15°C the difference between X2 and Y2 might be 0.015, at 20°C the difference between X3 and Y3 might be 0.01, at 25°C the difference between X4 and Y4 might be 0.008, at 30°C the difference between X5 and Y5 might be 0.005, at 35°C the difference between X6 and Y6 might be 0.003, and at 40°C the difference between X7 and Y7 might be 0.002. These differences reflect the degree of deviation between the two irradiance values ​​at different temperatures. However, these differences alone cannot fully determine the commonality matching weight; factors such as the trend of the differences over the entire temperature range must also be considered. For example, if the difference gradually decreases from 0.02 to 0.002, showing a trend of gradual convergence, this may indicate a high commonality matching weight. For example, after complex calculations, it is found that the commonality matching weight between A1 and A2 in the 1020nm channel is 0.9.

[0074] Following the same method, this analysis is performed for all possible pairwise combinations of photometers, such as A1 and A3, A1 and A4, and A2 and A3, to determine the channel irradiance correlation between each pair of reference multispectral sunphotometers in the 1020nm channel. Furthermore, since the conditions may vary across different channels, a similar process needs to be repeated for the 1640nm channel. For example, for A5 and A6 in the 1640nm channel, within the temperature range of -5°C to 15°C, the channel irradiance value for A5 is Z1 at -5°C, Z2 at 0°C, Z3 at 5°C, Z4 at 10°C, and Z5 at 15°C. The channel irradiance values ​​for A6 at the same temperatures are W1, W2, W3, W4, and W5, respectively. By analyzing the differences between Z1 and W1, Z2 and W2, and other differences at different temperatures, as well as the trends in these differences, the commonality matching weight for A5 and A6 in the 1640nm channel is calculated, for example, 0.85.

[0075] Next, the decision-making characteristics for the first cluster are determined based on the channel irradiance correlation, the detection thermal values ​​of several reference multispectral sunphotometers, and the first cluster indicator variable. The detection thermal value is closely related to the characteristics of the multispectral sunphotometer itself, reflecting the comprehensive status of the photometer during the detection process. For example, for photometer A10, its detection thermal value is M1, which can be affected by various factors, such as the detector sensitivity within the photometer, the instrument's heat dissipation performance, and the collaborative working state of each channel.

[0076] For example, the initial value of the first cluster indicator variable u is 1. For example, the channel irradiance correlation (commonality matching weight) between A1 and A2 in the 1020nm channel has been previously determined to be 0.9. The detected thermal value of A1 is N1, and the detected thermal value of A2 is N2. The first cluster indicator variable u = 1. Then, a value for the first cluster decision feature is calculated. For example, this value is 0.7. The important role of this first cluster decision feature is to adjust the channel irradiance correlation between reference multispectral sun photometers belonging to the same channel irradiance cluster data cluster.

[0077] Then, based on the first and second clustering decision features, the channel irradiance relationship data corresponding to several reference multispectral sunphotometers are clustered to obtain at least one channel irradiance cluster data cluster. During the uth round (u is initially 1), clustering is performed based on the first and second clustering decision features. For example, in the first round, if the first clustering decision feature is 0.7 and the second clustering decision feature is 0.6, the irradiance relationship data for the 1020nm channel of 20 reference multispectral sunphotometers (A1-A20) are clustered.

[0078] Based on these decision features, A1, A3, A5, and A9 might be grouped into one initial channel irradiance cluster data cluster, as the channel irradiance correlations between these photometers exhibit a certain degree of similarity, modulated by the first and second cluster decision features. A2, A4, A6, and A8 are then grouped into another initial channel irradiance cluster data cluster, and so on, resulting in at least one initial channel irradiance cluster data cluster. Similarly, a similar clustering operation is performed for the 1640nm channel.

[0079] After obtaining at least one initial channel irradiation cluster data cluster, the channel irradiation quality characteristic value of the first channel irradiation cluster data cluster within the at least one initial channel irradiation cluster data cluster is calculated. The first channel irradiation cluster data cluster refers to the set of reference multispectral sun photometer detected thermal values ​​included in the at least one initial channel irradiation cluster data cluster with the largest total detected thermal values. For example, among the initial channel irradiation cluster data clusters obtained above, the initial channel irradiation cluster data cluster containing A1, A3, A5, and A9 has the largest total detected thermal values, and this is considered the first channel irradiation cluster data cluster.

[0080] The channel irradiation quality eigenvalue is used to characterize the noise correction weight for the first channel irradiation cluster data cluster. This value is calculated based on multiple factors within the cluster. For example, within this cluster, the detected thermal value of A1 is N1, the detected thermal value of A3 is N3, the detected thermal value of A5 is N5, and the detected thermal value of A9 is N9. Furthermore, the channel irradiation relationship data between photometers within the cluster must be considered, such as the distribution of channel irradiation values ​​for each photometer at different temperatures. Then, through feature operations (such as full connection processing), the channel irradiation quality eigenvalue of this first channel irradiation cluster data cluster is obtained as 0.8.

[0081] If the channel irradiation quality characteristic value is greater than the first set characteristic value (for example, the first set characteristic value is 0.75), at least one initial channel irradiation cluster data cluster is determined as at least one channel irradiation cluster data cluster. In the previous example, since the calculated channel irradiation quality characteristic value of 0.8 is greater than 0.75, the initial channel irradiation cluster data cluster obtained by the previous division is directly determined as the final channel irradiation cluster data cluster.

[0082] If the channel irradiance quality characteristic value is less than the first set characteristic value, and u is less than the second set characteristic value (for example, the second set characteristic value is 3), u is set to u+1, and the step of clustering the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers based on the first clustering decision feature and the second clustering decision feature is performed again to obtain at least one initial channel irradiance cluster data cluster. For example, if the calculated channel irradiance quality characteristic value is 0.7 less than 0.75, and u=1 is less than 3 at this time, then u is set to 2 and the clustering operation is repeated.

[0083] In the second cycle, since the value of u changes to 2, the values ​​of the first and second cluster decision features may change based on the new value of u and the previous data. Then, based on the new first and second cluster decision features, the irradiance relationship data for the 1020nm channel of the 20 reference multispectral sun photometers (A1-A20) are clustered again to obtain new initial channel irradiance cluster data clusters. For example, this time A1, A4, A7, and A10 might be assigned to a new initial channel irradiance cluster data cluster, A2, A5, A8, and A11 to another new initial channel irradiance cluster data cluster, and so on. The channel irradiance quality characteristic values ​​are then calculated again for the new first channel irradiance cluster data cluster (the cluster with the largest new detected thermal value).

[0084] If the channel irradiation quality characteristic value is less than the first set characteristic value and u is equal to the second set characteristic value, the at least one initial channel irradiation cluster data cluster obtained in the cycle with the largest channel irradiation quality characteristic value is determined as the at least one channel irradiation cluster data cluster. For example, if after three cycles (u ranges from 1 to 3), the channel irradiation quality characteristic values ​​obtained in the three cycles are 0.7, 0.65, and 0.72 respectively, since the 0.72 obtained in the third cycle is the largest, the at least one initial channel irradiation cluster data cluster obtained in the third cycle is determined as the final channel irradiation cluster data cluster.

[0085] This clustering method is crucial in the practical application of multispectral sunphotometers. For example, inaccurate clustering can severely impact subsequent measurements when measuring atmospheric aerosol optical depth (AOD). For example, in the 1020nm channel, if the A1, A3, and A5 photometers are correctly clustered, their channel irradiance data will exhibit similar patterns. Therefore, a specific calibration strategy (such as linear calibration) should be employed. With correct clustering and calibration, the error between the measured AOD and the actual value should be within 0.02.

[0086] However, if the clustering is incorrect, for example, if A1 and A3 are grouped together with photometers with significantly different channel irradiance relationship data patterns (such as A7 and A8), this can lead to deviations in the cluster-based relationship correction strategy. An inappropriate correction strategy (such as a quadratic function correction strategy) may be mistakenly adopted, and the error between the measured aerosol optical depth and the actual value may increase to more than 0.1.

[0087] A similar situation exists for measurements in the 1640nm channel. For example, when measuring the absorption of solar radiation by atmospheric water vapor, improper clustering can affect measurement accuracy, even if the photometer, which should be using a specific calibration strategy, is incorrectly clustered. For example, when clustering is correct, the error between the measured parameter related to water vapor absorption and the actual value is within 0.03. However, when clustering is incorrect, the error may increase to over 0.15.

[0088] The technical effect of the above technical solution lies in rigorously determining the channel irradiance correlation, carefully calculating the clustering decision characteristics, and then determining the final clustering result based on the channel irradiance quality characteristic values ​​after multiple cycles of clustering. This process greatly improves the accuracy of clustering, thus laying a solid foundation for the subsequent precise determination of relationship correction strategies. By avoiding clustering errors, the measurement errors caused by improper clustering can be significantly reduced. For example, some measurement errors can be reduced from a possible 0.15 or more to within 0.02, effectively improving the data accuracy and reliability of multispectral sunphotometers in related applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0089] Based on the above technical ideas, the channel irradiation cluster data cluster includes the channel irradiation relationship data of the reference multispectral solar photometer, and the relationship correction strategy corresponding to the channel irradiation cluster data cluster is determined, including: mining the channel irradiation relationship quantitative knowledge corresponding to the channel irradiation relationship data of each reference multispectral solar photometer included in the channel irradiation cluster data cluster, and the channel irradiation relationship quantitative knowledge is used to characterize the relationship state information of the channel irradiation relationship data; based on the channel irradiation relationship quantitative knowledge corresponding to the channel irradiation relationship data of each reference multispectral solar photometer, the cluster description knowledge corresponding to the channel irradiation cluster data cluster is determined, and the cluster description knowledge corresponding to the channel irradiation cluster data cluster is used to characterize the temperature-noise relationship state information between the channel irradiation relationship data of each reference multispectral solar photometer; based on the cluster description knowledge corresponding to the channel irradiation cluster data cluster, the relationship correction strategy corresponding to the channel irradiation cluster data cluster is determined.

[0090] In an embodiment of the present application, based on the technical ideas described above, the following describes in detail how to determine the corresponding relationship correction strategy based on the channel irradiance relationship data of the reference multispectral sun photometer contained in the channel irradiance cluster data cluster.

[0091] First, we mine quantitative knowledge about the channel irradiance relationships corresponding to the channel irradiance relationship data of each reference multispectral photometer included in the channel irradiance cluster data cluster. For example, a channel irradiance cluster data cluster consisting of five reference multispectral photometers (labeled as Photometer 1, Photometer 2, Photometer 3, Photometer 4, and Photometer 5) for the 1020nm channel is used. For Photometer 1, its channel irradiance relationship data is examined within the temperature range of 15°C to 30°C. At 15°C, its channel irradiance value is A1; at 20°C, it is A2; at 25°C, it is A3; and at 30°C, it is A4. These values ​​exhibit certain relationships, for example, A2 - A1 = 0.03, A3 - A2 = 0.02, and A4 - A3 = 0.01, reflecting the magnitude of the channel irradiance changes with increasing temperature. At the same time, we can also examine the proportional relationship between irradiance values ​​at adjacent temperatures, such as (A2-A1) / A1=0.1, (A3-A2) / A2=0.067, etc. These values ​​and their relationships together constitute part of the quantitative knowledge of the channel irradiance relationship of the 1020nm channel of photometer 1 within this temperature range.

[0092] For photometer 2, within the same temperature range of 15°C to 30°C, the channel irradiance value at 15°C is B1, at 20°C it is B2, at 25°C it is B3, and at 30°C it is B4. The differences in the channel irradiance values ​​are B2-B1 = 0.025, B3-B2 = 0.018, and B4-B3 = 0.012. The proportional changes in irradiance values ​​at adjacent temperatures are (B2-B1) / B1 = 0.083, (B3-B2) / B2 = 0.072, and so on. These data provide quantitative information on the channel irradiance relationship of photometer 2 within this channel and temperature range.

[0093] Similarly, photometers 3, 4, and 5 were analyzed in the same manner, yielding quantitative knowledge about their respective channel irradiance relationships within the 15°C-30°C temperature range for the 1020nm channel. This quantitative knowledge comprehensively describes the relationship between the channel irradiance data for each photometer within a specific channel and temperature range, including information such as the absolute change in irradiance value and the relative change ratio.

[0094] Next, based on the channel irradiance relationship quantitative knowledge corresponding to the channel irradiance relationship data of each reference multispectral sun photometer, the cluster description knowledge corresponding to the channel irradiance cluster data cluster is determined. Continuing with the aforementioned 1020nm channel irradiance cluster data cluster containing five photometers as an example, the channel irradiance relationship quantitative knowledge of these five photometers is comprehensively considered. In terms of the amplitude of change in channel irradiance values, the amplitude of change in channel irradiance values ​​of Photometer 1, Photometer 2, and Photometer 3 with increasing temperature is relatively close. For example, within the temperature range of 20°C-25°C, the difference in the amplitude of change in their channel irradiance values ​​is within 0.005. In terms of the proportional relationship of irradiance value changes, the difference in the proportional change of irradiance values ​​of these three photometers at adjacent temperatures is also within a certain range, such as within 0.01.

[0095] However, photometers 4 and 5 differ from the first three photometers in certain aspects. For example, within the temperature range of 15°C to 20°C, the amplitude of change in the channel irradiance values ​​of photometer 4 is slightly larger than that of the first three photometers, with a difference of between 0.01 and 0.02. Within the temperature range of 25°C to 30°C, the amplitude of change in the irradiance values ​​of photometer 5 differs by approximately 0.015 from that of the first three photometers. These comparisons and analyses of the quantitative knowledge about the channel irradiance relationships between the various photometers constitute the cluster description knowledge corresponding to the channel irradiance cluster data clusters. This information characterizes the temperature-noise relationship state information between the channel irradiance relationship data of the various reference multispectral sun photometers. This temperature-noise relationship state information can be understood as the interference experienced by the channel irradiance relationship data under temperature fluctuations and the correlation between them. For example, if the amplitude of change in the channel irradiance values ​​of a particular photometer differs significantly from that of the other photometers, it may indicate that it is affected differently by temperature noise or other interfering factors, and this difference is reflected in the cluster description knowledge.

[0096] Finally, based on the cluster description knowledge corresponding to the channel irradiation cluster data clusters, a relationship correction strategy for the corresponding channel irradiation cluster data clusters is determined. Still taking the 1020nm channel cluster data cluster as an example, based on the cluster description knowledge, since photometers 1, 2, and 3 have high similarity in terms of channel irradiation relationship quantitative knowledge, a relationship correction strategy, such as a linear correction strategy, can be considered for the subset of these three photometers. This is because their channel irradiation values ​​exhibit a relatively regular linear characteristic with temperature. As can be seen from the previous quantitative knowledge analysis, the variation amplitude and proportional relationship of their irradiation values ​​at adjacent temperatures are relatively stable.

[0097] As for photometer 4 and photometer 5, since they have certain differences from the other three photometers, special considerations are required when determining the relationship correction strategy. For example, a correction strategy with a compensation coefficient can be adopted for the large variation range of photometer 4 in the temperature range of 15°C-20°C. For example, in this temperature range, a compensation coefficient of 1.1 is determined based on the degree of difference between it and other photometers, so that when performing relationship correction, its channel irradiance value can be appropriately adjusted. For the special case of photometer 5 in the temperature range of 25°C-30°C, a similar correction strategy for its special case can also be determined, such as adjusting the calculation method of its irradiance value in this temperature range to match the correction logic of the entire cluster data cluster.

[0098] This method of determining the relationship correction strategy is of great significance for the accurate measurement of multispectral sun photometers. Taking the measurement of aerosol optical depth in the atmosphere as an example, if the relationship correction strategy is inaccurate, it will lead to increased measurement errors. For example, when the relationship correction strategy is correctly determined, for the measurement of the 1020nm channel, the error between the measured aerosol optical depth and the actual value is within 0.02. However, if the relationship correction strategy is not correctly determined based on the cluster description knowledge, for example, if the photometer that should use a compensation coefficient correction strategy is processed according to the linear correction strategy, the measurement error may increase to more than 0.1.

[0099] The same is true for measurements of the 1640nm channel. For example, when measuring the absorption of solar radiation by water vapor in the atmosphere, if the relationship correction strategy is inappropriate, the error between the measured parameter related to water vapor absorption and the actual value may be within 0.03 when correctly corrected, but may increase to more than 0.15 when incorrectly corrected. This shows that accurately determining the relationship correction strategy based on the cluster description knowledge corresponding to the channel irradiance cluster data cluster is crucial to improving the accuracy and reliability of multispectral sun photometer measurements.

[0100] This design allows for the mining of quantitative knowledge about channel irradiance relationships, determining cluster description knowledge, and ultimately accurately determining relationship correction strategies. This process helps improve the accuracy of relationship correction strategies, thereby reducing measurement errors caused by inappropriate correction strategies. For example, it can reduce measurement errors in certain channels from a potential value exceeding 0.15 to within 0.02, improving the data accuracy and reliability of multispectral sun photometers in applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0101] In a preferred embodiment, the temperature-sensitive association strategy of the multispectral sunphotometer to be corrected is determined from several initial temperature-sensitive association strategies based on the temperature environment change data of the multispectral sunphotometer to be corrected, including: calculating the commonality matching weights between the temperature environment change data of the multispectral sunphotometer to be corrected and the several initial temperature-sensitive association strategies; and determining the initial temperature-sensitive association strategy with the largest commonality matching weight with the temperature environment change data of the multispectral sunphotometer to be corrected as the temperature-sensitive association strategy of the multispectral sunphotometer to be corrected.

[0102] It is understandable that the core invention point of determining a temperature-sensitive correlation strategy from a number of initial temperature-sensitive correlation strategies based on the temperature environment change data of the multispectral sunphotometer to be corrected is introduced as follows.

[0103] First, the temperature and environmental variation data of the multispectral sunphotometer to be corrected is a key element. Taking a specific multispectral sunphotometer as an example, its temperature and environmental variation data covers the instrument's various performance characteristics at different temperatures. For example, within the temperature range of -10°C to 30°C, at -10°C, the detector response value of a particular channel of the photometer is A1. When the temperature rises to 0°C, the response value changes to A2, with a certain difference between A2 and A1, for example, 0.05. As the temperature continues to rise to 10°C, the response value changes to A3, with a difference of 0.03 between A3 and A2. At 20°C, the response value changes to A4, with a difference of 0.02 between A4 and A3. At 30°C, the response value is A5, with a difference of 0.01 between A5 and A4. These changes in response values ​​at different temperatures, as well as the relationship between these differences, are important components of the temperature and environmental variation data. Temperature and environmental variation data may also include changes in the instrument's noise level at different temperatures. For example, at -10°C, the noise level is B1, and at 0°C it is B2. There is a proportional relationship between B2 and B1, for example, B2 / B1=1.2. This means that as the temperature rises from -10°C to 0°C, the noise level increases by a ratio of 0.2. As the temperature rises to 10°C, 20°C, and 30°C, the noise levels become B3, B4, and B5, respectively. These levels have similar proportional relationships with the noise levels at the previous temperatures. These proportional relationships are also part of the temperature environment change data.

[0104] Next, consider several initial temperature-sensitive association strategies. For example, there are three initial temperature-sensitive association strategies, labeled Temperature-Sensitive Association Strategy 1, Temperature-Sensitive Association Strategy 2, and Temperature-Sensitive Association Strategy 3. Temperature-Sensitive Association Strategy 1 is designed for situations where the detector response drops sharply and the noise level rises rapidly at low temperatures. It includes a series of adjustment parameters and rules. For example, when the detector response drops by more than a certain value (e.g., 0.08) and the noise level rises by more than 1.5 at low temperatures, the calculation of subsequent measurements is adjusted in a specific manner. This adjustment may involve multiplying by a coefficient or adding a compensation value. Temperature-Sensitive Association Strategy 2 is designed for situations where the detector response changes relatively steadily during temperature changes, but the noise level fluctuates significantly. For example, when the difference in detector response at different temperatures is between 0.01 and 0.03 and the noise level fluctuates by a ratio between 1.1 and 1.3, a different specific measurement adjustment rule is used. Temperature-sensitive correlation strategy 3 is designed to address the situation where detector response values ​​may fluctuate abnormally at high temperatures. When the temperature is higher than a certain threshold (for example, 25°C) and the fluctuation range of the detector response value exceeds 0.05, the measurement value is corrected according to specific rules.

[0105] Next, the commonality matching weights between the temperature and environmental change data of the multispectral sunphotometer to be corrected and several initial temperature-sensitive association strategies were calculated. For Temperature-Sensitive Association Strategy 1, the temperature and environmental change data of the multispectral sunphotometer mentioned above were compared. The change in detector response values ​​showed a decrease of 0.05 within the temperature range of -10°C to 0°C. While this did not exceed the 0.08 set in Temperature-Sensitive Association Strategy 1, there was still a certain degree of similarity. The change in noise level showed an increase of 0.2 between -10°C and 0°C, partially exceeding the 1.5 set in Temperature-Sensitive Association Strategy 1. Taking these factors into consideration, the commonality matching weight between the temperature and environmental change data of the multispectral sunphotometer to be corrected and Temperature-Sensitive Association Strategy 1 was 0.4.

[0106] For Temperature-Sensitive Correlation Strategy 2, the detector response value variation remained generally between 0.01 and 0.03 across the entire -10°C to 30°C temperature range, meeting some of the requirements for Temperature-Sensitive Correlation Strategy 2. Regarding noise level fluctuation, the noise level fluctuation ratio varied across the -10°C to 30°C temperature range, but some ranges met the 1.1-1.3 range for Temperature-Sensitive Correlation Strategy 2. Based on these findings, the calculated commonality matching weight between the temperature environment change data of the multispectral sunphotometer to be corrected and Temperature-Sensitive Correlation Strategy 2 was 0.6.

[0107] For Temperature-Sensitive Correlation Strategy 3, the detector response at low and medium temperatures is not relevant to the design of Temperature-Sensitive Correlation Strategy 3 to address abnormal detector response fluctuations at high temperatures. Only when the temperature rises to 30°C does the detector response fluctuate by 0.01, which is far below the 0.05 threshold set in Temperature-Sensitive Correlation Strategy 3. Taking all these factors into consideration, the calculated commonality matching weight between the temperature environment change data of the multispectral sunphotometer to be corrected and Temperature-Sensitive Correlation Strategy 3 is 0.2.

[0108] Finally, the initial temperature-sensitive association strategy with the highest commonality matching weight with the temperature environment change data of the multispectral sunphotometer to be corrected is selected as the temperature-sensitive association strategy for the multispectral sunphotometer to be corrected. Among the previously calculated commonality matching weights, Temperature-Sensitive Association Strategy 2 has the highest commonality matching weight of 0.6, so Temperature-Sensitive Association Strategy 2 is selected as the temperature-sensitive association strategy for the multispectral sunphotometer to be corrected.

[0109] It can be seen that the above method of determining the temperature-sensitive correlation strategy is of great significance to the accuracy of multispectral sun photometer measurements. If the temperature-sensitive correlation strategy is not accurately determined, it may lead to large deviations in the measurement results. For example, when measuring the optical depth of aerosols in the atmosphere, if the temperature-sensitive correlation strategy 1 is mistakenly adopted instead of the correct temperature-sensitive correlation strategy 2, when the temperature varies between -10°C and 30°C, the error between the measured optical depth of aerosols and the actual value may increase from less than 0.03 under the correct strategy to more than 0.1, because the temperature-sensitive correlation strategy 1 is not fully suitable for the temperature environment changes of the photometer. Similarly, when measuring the absorption of solar radiation by water vapor in the atmosphere, if an inappropriate temperature-sensitive correlation strategy is adopted, the error in the measurement results will also increase, affecting the accurate measurement of atmospheric environment-related parameters.

[0110] The technical effect of the above embodiment is to accurately determine the temperature-sensitive association strategy by calculating the common matching weight between temperature environment change data and the temperature-sensitive association strategy. This helps reduce measurement errors caused by improper selection of the temperature-sensitive association strategy. For example, it can reduce certain measurement errors from a possible error of more than 0.1 to within 0.03, thereby improving the data accuracy and reliability of multispectral sun photometers in applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0111] In another preferred embodiment, the method also includes: obtaining temperature environment change data corresponding to several reference multispectral solar photometers, wherein the reference multispectral solar photometer refers to a multispectral solar photometer that has completed a correction operation; clustering the temperature environment change data corresponding to the several reference multispectral solar photometers to obtain at least one temperature environment cluster set, each temperature environment cluster set including the temperature environment change data of at least one reference multispectral solar photometer; for each temperature environment cluster set, based on the temperature environment change data of the reference multispectral solar photometer included in the temperature environment cluster set, determining the temperature-sensitive association strategy corresponding to the temperature environment cluster set.

[0112] In this preferred embodiment, the entire process begins by first acquiring temperature and environmental change data corresponding to several reference multispectral photometers. These reference multispectral photometers have already undergone calibration, and their temperature and environmental change data is highly valuable. For example, there are 10 reference multispectral photometers, labeled Photometer 1 through Photometer 10. For Photometer 1, its temperature and environmental change data exhibits a specific pattern within the temperature range of -20°C to 40°C. When the temperature is -20°C, the detector response value of a certain channel is A1. When the temperature rises to -10°C, the response value changes to A2, and the difference between A2 and A1 is 0.03. When the temperature continues to rise to 0°C, the response value changes to A3, and the difference between A3 and A2 is 0.02. At 10°C, the response value is A4, and the difference between A4 and A3 is 0.01. Meanwhile, at -20°C, the noise level is B1, and at -10°C it is B2, with the ratio of B2 to B1 being 1.1 (meaning B2 is 1.1 times B1). At 0°C, the noise level is B3, with the ratio of B3 to B2 being 1.05. For photometer 2, within the same temperature range, its detector response and noise level also vary. At -20°C, the response is C1, but at -10°C it becomes C2, with the difference between C2 and C1 being 0.025. At 0°C, it is C3, with the difference between C3 and C2 being 0.018. The noise level is D1 at -20°C and D2 at -10°C, with the ratio of D2 to D1 being 1.08. Similarly, photometers 3 through 10 have similar data on how the detector response and noise levels change at different temperatures, which together constitute their respective temperature environment variation data.

[0113] Next, the temperature environment change data corresponding to several reference multispectral sunphotometers are clustered to obtain at least one temperature environment cluster set. Clustering is based on multiple factors, including the changing trends of detector response values ​​and the changing patterns of noise levels. For example, for Photometer 1, Photometer 3, and Photometer 5, the response values ​​of these sensors are relatively similar within the temperature range of -20°C to 0°C. The response value differences for Photometer 1 from -20°C to -10°C are 0.03, and from -10°C to 0°C are 0.02. The response value differences for Photometer 3 from -20°C to -10°C are 0.028, and from -10°C to 0°C are 0.019. The response value differences for Photometer 5 from -20°C to -10°C are 0.032, and from -10°C to 0°C are 0.021. In terms of noise levels, the three photometers also have similar proportional relationships for noise levels within the temperature range of -20°C to 0°C. Therefore, the temperature environment change data for Photometers 1, 3, and 5 can be grouped into one temperature environment cluster. Looking at Photometers 2, 4, and 6, they share another similarity in the patterns of change in detector response values ​​and noise levels. For example, within the temperature range of 0°C to 20°C, the difference in their response value changes and the proportional relationship between their noise levels are similar. Therefore, the temperature environment change data for these three photometers can be grouped into another temperature environment cluster. In this way, the temperature environment change data for the ten reference multispectral sunphotometers can be clustered to obtain at least one temperature environment cluster.

[0114] Then, for each temperature environment cluster, a temperature-sensitive association strategy is determined based on the temperature environment change data from the reference multispectral sun photometer in the cluster. For example, for the temperature environment cluster containing photometers 1, 3, and 5, the detector response values ​​in this cluster exhibit relatively stable variations within the -20°C to 0°C temperature range, with smaller variations and a gradual flattening within the 0°C to 20°C temperature range. The noise level exhibits little proportional variation within the -20°C to 0°C temperature range and a slowly decreasing trend within the 0°C to 20°C temperature range. Based on these temperature environment change data characteristics, a temperature-sensitive association strategy can be determined. This strategy can be adjusted to address variations in detector response values. For example, within the -20°C to 0°C temperature range, when the detector response value variation is between 0.02 and 0.03, a small correction factor, such as 1.02 (determined based on the data characteristics within the cluster), is applied and multiplied when calculating the measured value. In the temperature range of 0°C to 20°C, since the response value changes more slowly and more gradually, a smaller correction factor, such as 1.01, is used. Regarding noise level, in the temperature range of -20°C to 0°C, when the noise level ratio is between 1.05 and 1.1, a specific noise processing method is used. In the temperature range of 0°C to 20°C, based on the slowly decreasing noise level, a different noise processing method is used.

[0115] Taking the temperature environment cluster containing photometers 2, 4, and 6 as an example, in this cluster, the detector response values ​​show a trend of first slowly rising and then stabilizing within the temperature range of 0°C to 20°C, with small fluctuations within the temperature range of 20°C to 40°C. The noise level also shows a small upward trend within the temperature range of 0°C to 20°C and remains relatively stable within the temperature range of 20°C to 40°C. Based on these temperature environment change data characteristics, another temperature-sensitive correlation strategy can be determined. In the temperature range of 0°C to 20°C, when the detector response values ​​first slowly rise and then stabilize, a correction coefficient, such as 1.03, is used when the response value increases by between 0.01 and 0.02. In the temperature range of 20°C to 40°C, when the detector response value fluctuates by between 0.005 and 0.01, another correction coefficient, such as 1.015, is used. Regarding the noise level, in the temperature range of 0℃ to 20℃, when the noise level increases by 0.05-0.1, a specific noise suppression strategy is adopted. In the temperature range of 20℃ to 40℃, since the noise level is relatively stable, a strategy to maintain measurement stability is adopted.

[0116] It is understandable that if such accurate clustering and the determination of the temperature-sensitive association strategy based on the clustering are not performed, large deviations in the measurement results may occur. For example, when measuring the optical thickness of aerosols in the atmosphere, if the clustering and temperature-sensitive association strategy are not correctly determined, for the situation where photometers 1, 3, and 5 are located, if the temperature-sensitive association strategy for photometers 2, 4, and 6 is incorrectly adopted, the error between the measured optical thickness of aerosols and the actual value may increase from within 0.02 under the original correct strategy to more than 0.1. Similarly, when measuring the absorption of solar radiation by water vapor in the atmosphere, if an appropriate temperature-sensitive association strategy is not adopted, the error in the measurement results will also increase, affecting the accurate measurement of atmospheric environment-related parameters.

[0117] The core innovation of the above-mentioned embodiment lies in clustering the temperature and environmental change data from a reference multispectral sunphotometer and then determining a temperature-sensitive association strategy based on the clustering. This helps improve the accuracy of the temperature-sensitive association strategy, thereby reducing measurement errors caused by improper selection of the temperature-sensitive association strategy. For example, it can reduce certain measurement errors from a potential error of more than 0.1 to within 0.02, thereby improving the data accuracy and reliability of multispectral sunphotometers in applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0118] In another preferred embodiment, the method further includes: obtaining atmospheric state inversion data of at least one atmospheric state, the atmospheric state inversion data being used to characterize the correction record of the reference multispectral sun photometer corresponding to the atmospheric state, the atmospheric state inversion data including a plurality of atmospheric state inversion events; for each atmospheric state, based on the plurality of atmospheric state inversion events included in the atmospheric state inversion data of the atmospheric state, identifying event evaluation values ​​corresponding to the plurality of atmospheric state inversion events, the event evaluation values ​​corresponding to the atmospheric state inversion events being used to indicate the confidence coefficient of the atmospheric state determined based on the atmospheric state inversion event; determining a scene evaluation value of the atmospheric state based on the event evaluation values ​​corresponding to the plurality of atmospheric state inversion events, the scene evaluation value being used to indicate the confidence coefficient of the atmospheric state.

[0119] Furthermore, the atmospheric state inversion data based on the atmospheric state includes several atmospheric state inversion events, and event evaluation values ​​corresponding to the several atmospheric state inversion events are identified, including: on the basis that the atmospheric state inversion event is an abnormal fluctuation event, the event evaluation value corresponding to the atmospheric state inversion event is calculated by a first regression algorithm, and the inverse gradient of the linear weight value of the first regression algorithm has a mapping relationship with the temporal characteristics of the atmospheric state inversion event; on the basis that the atmospheric state inversion event is a non-abnormal fluctuation event, the event evaluation value corresponding to the atmospheric state inversion event is calculated by a second regression algorithm, and the second regression algorithm is a least squares algorithm.

[0120] In the above preferred embodiment, first, atmospheric state inversion data of at least one atmospheric state are obtained. These atmospheric state inversion data are used to characterize the correction record of the reference multispectral sun photometer corresponding to the atmospheric state, and contain several atmospheric state inversion events. Taking a specific atmospheric state as an example, for example, there are 5 atmospheric state inversion events in this atmospheric state, respectively marked as event 1, event 2, event 3, event 4 and event 5. These events are records related to the atmospheric state generated during the measurement and correction process of the reference multispectral sun photometer. For example, under the atmospheric state, for the measurement of a certain channel, event 1 may record the difference between the detector response value measured at a certain moment and the expected value, and event 2 may record the deviation of the noise level measured at another moment from the normal range, etc.

[0121] Next, for each atmospheric state, the event evaluation values ​​corresponding to several atmospheric state inversion events contained in the atmospheric state inversion data are identified. Different algorithms are used for calculation depending on whether the atmospheric state inversion event is an abnormal fluctuation event or a non-abnormal fluctuation event.

[0122] Let's first consider the case where the atmospheric state inversion event is an abnormal fluctuation. For example, Event 1 is an abnormal fluctuation event. This may be caused by a sudden, significant fluctuation in the detector response of a channel of a multispectral sun photometer under certain unique meteorological conditions. The event evaluation value corresponding to this event is calculated using the first regression algorithm. The inverse gradient of the linear weight value of the first regression algorithm is mapped to the temporal characteristics of the atmospheric state inversion event. For example, in Event 1, the detector response value suddenly dropped by 0.05 (indicating the magnitude of the abnormal fluctuation) within a short period of time, and this fluctuation occurred at a specific time. The first regression algorithm is used to calculate the abnormal fluctuation based on its magnitude, time of occurrence, and its relationship to other relevant factors. If this abnormal fluctuation occurs early in the measurement, it may have a significant impact on the overall measurement. Therefore, the event evaluation value calculated using the first regression algorithm may be lower, for example, 0.3 (indicating a low confidence level). If this abnormal fluctuation occurs later in the measurement and subsequent measurements show a certain recovery trend, the event evaluation value calculated using the algorithm may be higher, for example, 0.5.

[0123] Let's consider the case where the atmospheric state inversion event is a non-anomalous fluctuation event. Taking Event 2 as an example, this is a non-anomalous fluctuation event, likely a small, normal change in the detector response under normal meteorological conditions. In this case, the event evaluation value is calculated using the second regression algorithm, the least squares algorithm. For example, the change in the detector response value in Event 2 over a period of time is 0.01, which is within the normal fluctuation range. The least squares algorithm comprehensively considers this change and the fit with other normal measurement data to calculate the event evaluation value for Event 2. For example, the calculated event evaluation value is 0.8 (indicating a high confidence level), indicating that this non-anomalous fluctuation event had a minimal impact on the atmospheric state measurement and a high overall reliability.

[0124] Then, based on the event evaluation values ​​corresponding to several atmospheric state inversion events, the scene evaluation value of the atmospheric state is determined. Continuing with the example of the atmospheric state with five atmospheric state inversion events, for example, the event evaluation value of event 1 is 0.3, the event evaluation value of event 2 is 0.8, the event evaluation value of event 3 is 0.7, the event evaluation value of event 4 is 0.6, and the event evaluation value of event 5 is 0.9. When determining the scene evaluation value of the atmospheric state, it is necessary to comprehensively consider these five event evaluation values. A weighted average approach can be used, for example, by assigning different weights to each event based on its importance in the overall atmospheric state measurement.

[0125] For example, event 1 is an abnormal fluctuation event and occurs at the beginning of the measurement, so its weight is set to 0.2; event 2 is a non-abnormal fluctuation event with a small fluctuation, so its weight is set to 0.15; event 3 is weighted to 0.2; event 4 is weighted to 0.2; and event 5 is weighted to 0.25. The following calculation (0.3 × 0.2 + 0.8 × 0.15 + 0.7 × 0.2 + 0.6 × 0.2 + 0.9 × 0.25) yields a scenario evaluation value of 0.68 for the atmospheric state (0.68 is an example value here). This scenario evaluation value is used to indicate the confidence factor of the atmospheric state.

[0126] Failure to accurately determine the confidence factor for atmospheric conditions can lead to problems in subsequent measurements and data processing. For example, when measuring atmospheric aerosol optical depth (AOD), an incorrect estimate of the atmospheric state confidence factor will result in inaccurate measurement results. For example, when the atmospheric state confidence factor is correctly determined, the error between the measured AOD and the actual value is within 0.02. However, if the atmospheric state scene assessment value is incorrectly calculated, resulting in an overestimate or underestimate of the confidence factor—for example, misinterpreting a scene assessment value of 0.68 (indicating some uncertainty) as 0.9 (indicating high confidence)—then some factors that should have been considered may be overlooked in subsequent data processing, causing the error between the measured AOD and the actual value to exceed 0.1. Similarly, when measuring the absorption of solar radiation by water vapor in the atmosphere, an inaccurate atmospheric state confidence factor can also increase measurement errors, affecting the accurate measurement of atmospheric environment parameters.

[0127] The design objective of this embodiment is to accurately characterize the atmospheric state confidence coefficient by distinguishing the types of atmospheric state inversion events, using different algorithms to calculate event evaluation values, and then determining a scene evaluation value for the atmospheric state. This helps improve the accuracy of atmospheric state assessments, thereby reducing measurement errors caused by inaccurate atmospheric state confidence coefficients. For example, certain measurement errors can be reduced from a possible value of over 0.1 to within 0.02, thereby improving the data accuracy and reliability of multispectral sun photometers in applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0128] Based on S101-S105, the method also includes: labeling each of the multispectral solar photometers to be corrected based on the temperature correction strategy corresponding to each of the multispectral solar photometers to be corrected, and determining a distributed correction device for each of the multispectral solar photometers to be corrected based on the labeling information; or, determining a sequence tag for each of the multispectral solar photometers to be corrected based on the temperature correction strategy corresponding to each of the multispectral solar photometers to be corrected, and assigning each of the multispectral solar photometers to be corrected to a corresponding distributed correction device for processing based on the sequence tag of each of the multispectral solar photometers to be corrected.

[0129] In the technical solution of the multispectral sun photometer, based on the aforementioned steps S101 to S105, this embodiment involves the expansion and optimization of the labeling process.

[0130] First, label each multispectral sunphotometer to be corrected based on its corresponding temperature correction strategy. For example, consider a group of ten multispectral sunphotometers to be corrected, labeled Photometer 1 through Photometer 10. For Photometer 1, its temperature correction strategy encompasses multiple aspects, including an atmospheric state correction perspective, a relationship correction strategy, and a temperature-sensitive correlation strategy. For example, the atmospheric state correction perspective provides a confidence coefficient of 0.8, and the relationship correction strategy employs a linear correction strategy for the 1020nm channel. (This linear correction strategy may be based on the specific relationship between irradiance and temperature at different temperatures for this channel, such as the linear relationship between irradiance and temperature within the 15°C-30°C temperature range.) The temperature-sensitive correlation strategy adjusts the detector response of the 1640nm channel at low temperatures. (For example, if the detector response is low within the -5°C-5°C temperature range, a specific correction factor, such as 1.1, is applied based on the temperature-sensitive correlation strategy.) Based on this temperature correction strategy, Photometer 1 can be labeled. The annotations may include key features of its temperature correction strategy, such as a confidence coefficient of 0.8, linear correction for the 1020nm channel, and a low-temperature correction factor of 1.1 for the 1640nm channel.

[0131] Similarly, photometers 2 through 10 are annotated according to their respective temperature correction strategies. For example, the atmospheric state correction confidence coefficient for photometer 2 is 0.75, and the relationship correction strategy for one channel is a quadratic function correction strategy (this is because the relationship between irradiance and temperature at different temperatures in this channel is more consistent with the characteristics of a quadratic function, such as the existence of a quadratic function relationship between irradiance and temperature in the temperature range of 20°C-40°C). The temperature-sensitive correlation strategy is targeted at the high temperature conditions of another channel (for example, in the temperature range of 30°C-45°C, the detector response value has abnormal fluctuations, and the corresponding adjustment rules are adopted). Photometer 2 is then annotated to include these characteristics.

[0132] Next, based on the annotation information, the distributed correction device for each multispectral sun photometer to be corrected is determined. For example, there are three types of distributed correction devices: Device A, Device B, and Device C. Device A is suitable for photometers with high confidence coefficients and relatively simple correction strategies. For example, Device A is suitable for photometers with a confidence coefficient of 0.8 or above for atmospheric state correction and a calibration strategy that primarily uses linear correction. Based on the aforementioned annotation information, Photometer 1 has a confidence coefficient of 0.8 and a linear correction for its 1020nm channel, so Photometer 1 can be assigned to Device A for processing. Device B, on the other hand, may be suitable for photometers with medium confidence coefficients and a calibration strategy that includes quadratic function correction. For example, Photometer 2 has a confidence coefficient of 0.75 and a quadratic function correction strategy, so Photometer 2 can be assigned to Device B for processing. Device C may be suitable for photometers with special temperature-sensitive correlation strategies (such as special correction rules for a particular channel at extremely low or high temperatures). For example, if a temperature-sensitive correlation strategy of a photometer has special correction rules for a certain channel below -10°C or above 45°C, it can be assigned to device C for processing.

[0133] Another approach is to determine the sequence label for each multispectral solar photometer to be corrected based on its corresponding temperature correction strategy. For example, for photometer 1, a sequence label can be determined based on the characteristics of its temperature correction strategy. This sequence label may be a coded form based on factors such as its confidence coefficient and channel correction strategy type. For example, based on photometer 1's characteristics such as a confidence coefficient of 0.8, linear correction of the 1020nm channel, and low-temperature correction of the 1640nm channel, its sequence label is determined to be "08L1T1" (this is just an example encoding form, where "08" represents a confidence coefficient of 0.8, "L" represents linear correction, "1" represents the 1020nm channel, and "T1" represents the low-temperature correction-related characteristics of the 1640nm channel). Similarly, for photometer 2, based on its confidence coefficient of 0.75, quadratic function correction in one channel, and high-temperature correction in another channel, its sequence label is determined to be "075Q1H1" ("075" represents a confidence coefficient of 0.75, "Q" represents quadratic function correction, "1" represents a certain channel, and "H1" represents high-temperature correction-related features in another channel).

[0134] Then, based on the sequence tags of each multispectral photometer to be corrected, each is assigned to the corresponding distributed correction device for processing. For example, if device A specifies that it processes photometers with sequence tags starting with "08" and containing "L" (a characteristic associated with linear correction), then photometer 1 can be assigned to device A. If device B specifies that it processes photometers with sequence tags containing "075" and containing "Q" (a characteristic associated with quadratic correction), then photometer 2 can be assigned to device B.

[0135] This allocation method is crucial for the correction of multispectral sun photometers. Without such a precise allocation, correction efficiency or inaccurate results can be inefficient. For example, if photometer 1, which should have been assigned to device A (suitable for simple linear correction and high confidence coefficients), is mistakenly assigned to device B (suitable for quadratic function correction, etc.), device B may not be able to effectively implement the temperature correction strategy of photometer 1. When measuring atmospheric aerosol optical depth, improper processing can increase the error between the measured aerosol optical depth and the actual value. For example, when correctly assigned to devices, the measurement error is within 0.02, but when incorrectly assigned, the error may exceed 0.1. Similarly, when measuring the absorption of solar radiation by water vapor in the atmosphere, incorrect allocation can lead to increased measurement errors, compromising the accurate measurement of atmospheric environmental parameters.

[0136] The core of this technology lies in assigning multispectral sunphotometers to be corrected to corresponding distributed correction devices by labeling or determining sequence tags based on a temperature correction strategy. This helps improve the accuracy and efficiency of the correction process, thereby reducing measurement errors caused by improper allocation. For example, it can reduce certain measurement errors from a potential of over 0.1 to within 0.02, thereby improving the data accuracy and reliability of multispectral sunphotometers in applications such as atmospheric optical properties research, meteorological monitoring, and environmental science.

[0137] Furthermore, Figure 2 Schematic diagram of a photometer detection temperature correction system 200 provided in an embodiment of the present application. Figure 2 The photometer detection temperature correction system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0138] Alternatively, as Figure 2 As shown, the photometer detection temperature correction system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.

[0139] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0140] Alternatively, as Figure 2 As shown, the photometer detection temperature correction system 200 may further include a transceiver 220. The processor 210 may control the transceiver 220 to interact with other devices. Specifically, the transceiver 220 may send information or data to other devices or receive information or data sent by other devices.

[0141] Optionally, the photometer detection temperature correction system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application. For the sake of brevity, they are not repeated here.

[0142] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities.

[0143] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0144] Based on the above, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.

Claims

1. A method for correcting temperature detected by a photometer based on machine learning, characterized in that: The method is applied to a photometer detection temperature correction system, and the method comprises: Acquire an environmental detection dataset of a multispectral sun photometer to be corrected, the environmental detection dataset including atmospheric state inversion data, channel irradiance relationship data, and temperature environment change data, the atmospheric state inversion data including an inversion cluster indicator variable associated with the atmospheric state corresponding to the multispectral sun photometer, the channel irradiance relationship data including linear data associated with the channel irradiance of the multispectral sun photometer, and the temperature environment change data including model data associated with temperature disturbance in the multispectral sun photometer; Determining, based on the atmospheric state inversion data of the multispectral sun photometer to be corrected, an atmospheric state correction viewpoint of the multispectral sun photometer to be corrected, wherein the atmospheric state correction viewpoint of the multispectral sun photometer to be corrected is used to indicate a confidence coefficient of the atmospheric state corresponding to the multispectral sun photometer to be corrected; Based on the channel irradiance relationship data of the multispectral sun photometer to be corrected, determining a relationship correction strategy for the multispectral sun photometer to be corrected from a plurality of initial relationship correction strategies, wherein the relationship correction strategy for the multispectral sun photometer to be corrected is used to indicate a channel irradiance correction rule corresponding to a detection disturbance of the multispectral sun photometer to be corrected in a temperature environment; Based on the temperature environment change data of the multispectral sunphotometer to be corrected, determining a temperature-sensitive association strategy of the multispectral sunphotometer to be corrected from a plurality of initial temperature-sensitive association strategies, wherein the temperature-sensitive association strategy of the multispectral sunphotometer to be corrected is used to indicate a temperature-sensitive disturbance type corresponding to the multispectral sunphotometer to be corrected; A temperature correction strategy for the multispectral sun photometer to be corrected is generated. The temperature correction strategy for the multispectral sun photometer to be corrected includes an atmospheric state correction viewpoint, a relationship correction strategy, and a temperature-sensitive correlation strategy for the multispectral sun photometer to be corrected.

2. The method according to claim 1, wherein The step of determining a relationship correction strategy for the multispectral sun photometer to be corrected from a plurality of initial relationship correction strategies based on the channel irradiance relationship data of the multispectral sun photometer to be corrected includes: Calculating the commonality matching weights between the channel irradiance relationship data of the multispectral sun photometer to be corrected and the plurality of initial relationship correction strategies; An initial relationship correction strategy having the largest commonality matching weight with the channel irradiance relationship data of the multispectral sun photometer to be corrected is determined as the relationship correction strategy of the multispectral sun photometer to be corrected.

3. The method according to claim 1, wherein The method further comprises: Acquiring channel irradiance relationship data corresponding to a plurality of reference multispectral sun photometers, wherein the reference multispectral sun photometers refer to multispectral sun photometers that have completed correction operations; Clustering the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers to obtain at least one channel irradiance cluster data cluster, each channel irradiance cluster data cluster including the channel irradiance relationship data of at least one reference multispectral sun photometer; For each channel irradiance cluster data cluster, a relationship correction strategy corresponding to the channel irradiance cluster data cluster is determined based on the channel irradiance relationship data of the reference multispectral sun photometer included in the channel irradiance cluster data cluster.

4. The method according to claim 3, wherein The clustering of the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers to obtain at least one channel irradiance cluster data cluster includes: Determining a channel irradiance correlation between every two reference multispectral sun photometers among the plurality of reference multispectral sun photometers, wherein the channel irradiance correlation is used to indicate a commonality matching weight between channel irradiance relationship data respectively corresponding to the two reference multispectral sun photometers; Determining a first clustering decision feature based on the channel irradiance correlation, the detected thermal values ​​of the plurality of reference multispectral sun photometers, and a first clustering indicator variable, wherein the first clustering decision feature is used to adjust the channel irradiance correlation between the reference multispectral sun photometers belonging to the same channel irradiance cluster data cluster, and the first clustering indicator variable refers to a cycle round of the clustering process; Based on the first clustering decision feature and the second clustering decision feature, the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers are clustered to obtain the at least one channel irradiance cluster data cluster, wherein the second clustering decision feature is used to adjust the detected thermal value of the reference multispectral sun photometer included in each channel irradiance cluster data cluster; The step of clustering the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers based on the first clustering decision feature and the second clustering decision feature to obtain the at least one channel irradiance cluster data cluster includes: During the u-th cycle, based on the first clustering decision feature and the second clustering decision feature, the channel irradiance relationship data corresponding to the plurality of reference multispectral sun photometers are clustered to obtain at least one initial channel irradiance clustering data cluster; wherein u represents the first clustering indicator variable, and the default value of u is 1; Calculating a channel irradiation quality characteristic value of a first channel irradiation cluster data cluster in the at least one initial channel irradiation cluster data cluster, wherein the first channel irradiation cluster data cluster refers to a set of maximum detected thermal values ​​of a reference multispectral sun photometer included in the at least one initial channel irradiation cluster data cluster, and the channel irradiation quality characteristic value is used to characterize a noise correction weight of the first channel irradiation cluster data cluster; On the basis that the channel irradiation quality characteristic value is greater than a first set characteristic value, determining the at least one initial channel irradiation cluster data cluster as the at least one channel irradiation cluster data cluster; On the basis that the channel irradiation quality characteristic value is less than the first set characteristic value and u is less than the second set characteristic value, setting u=u+1, and again performing the step of clustering the channel irradiation relationship data corresponding to the plurality of reference multispectral sun photometers based on the first clustering decision feature and the second clustering decision feature to obtain at least one initial channel irradiation cluster data cluster; On the basis that the channel irradiation quality characteristic value is less than the first set characteristic value and u is equal to the second set characteristic value, at least one initial channel irradiation cluster data cluster obtained in a cycle process with the largest channel irradiation quality characteristic value is determined as the at least one channel irradiation cluster data cluster.

5. The method according to claim 3, wherein The determining of the relationship correction strategy corresponding to the channel irradiation cluster data cluster based on the channel irradiation relationship data of the reference multispectral sun photometer included in the channel irradiation cluster data cluster comprises: mining channel irradiation relationship quantitative knowledge corresponding to the channel irradiation relationship data of each reference multispectral sun photometer included in the channel irradiation cluster data cluster, wherein the channel irradiation relationship quantitative knowledge is used to characterize relationship state information of the channel irradiation relationship data; Determining cluster description knowledge corresponding to the channel irradiation cluster data cluster based on the channel irradiation relationship quantification knowledge corresponding to the channel irradiation relationship data of each reference multispectral sun photometer, wherein the cluster description knowledge corresponding to the channel irradiation cluster data cluster is used to characterize the temperature-noise relationship state information between the channel irradiation relationship data of each reference multispectral sun photometer; Based on the cluster description knowledge corresponding to the channel irradiation cluster data cluster, a relationship correction strategy corresponding to the channel irradiation cluster data cluster is determined.

6. The method according to claim 1, wherein The method of determining the temperature-sensitive association strategy of the multispectral sunphotometer to be corrected from a plurality of initial temperature-sensitive association strategies based on the temperature environment change data of the multispectral sunphotometer to be corrected includes: Calculating the commonality matching weights between the temperature environment change data of the multispectral sun photometer to be corrected and the plurality of initial temperature-sensitive association strategies; An initial temperature-sensitive association strategy having the largest commonality matching weight with the temperature environment change data of the multispectral sun photometer to be corrected is determined as the temperature-sensitive association strategy of the multispectral sun photometer to be corrected.

7. The method according to claim 1, wherein The method further comprises: Acquiring temperature environment change data corresponding to a plurality of reference multispectral sun photometers, wherein the reference multispectral sun photometers refer to multispectral sun photometers that have completed correction operations; Clustering the temperature environment change data corresponding to the plurality of reference multispectral sun photometers to obtain at least one temperature environment cluster set, each temperature environment cluster set including the temperature environment change data of at least one reference multispectral sun photometer; For each temperature environment cluster set, a temperature-sensitive association strategy corresponding to the temperature environment cluster set is determined based on the temperature environment change data of the reference multispectral sun photometer included in the temperature environment cluster set.

8. The method according to claim 1, wherein The method further comprises: Acquiring atmospheric state inversion data of at least one atmospheric state, the atmospheric state inversion data being used to characterize a correction record of a reference multispectral sun photometer corresponding to the atmospheric state, the atmospheric state inversion data including a plurality of atmospheric state inversion events; For each atmospheric state, identifying, based on a plurality of atmospheric state inversion events included in the atmospheric state inversion data of the atmospheric state, event evaluation values ​​corresponding to the plurality of atmospheric state inversion events, wherein the event evaluation values ​​corresponding to the atmospheric state inversion events are used to indicate a confidence coefficient of the atmospheric state determined based on the atmospheric state inversion events; Determining a scene evaluation value of the atmospheric state based on event evaluation values ​​corresponding to the plurality of atmospheric state inversion events, wherein the scene evaluation value is used to indicate a confidence coefficient of the atmospheric state; The method of identifying the event evaluation values ​​corresponding to the plurality of atmospheric state inversion events included in the atmospheric state inversion data based on the atmospheric state comprises: On the basis that the atmospheric state inversion event is an abnormal fluctuation event, an event evaluation value corresponding to the atmospheric state inversion event is calculated using a first regression algorithm, wherein a reverse gradient of a linear weight value of the first regression algorithm has a mapping relationship with a temporal feature of the atmospheric state inversion event; On the basis that the atmospheric state inversion event is a non-abnormal fluctuation event, an event evaluation value corresponding to the atmospheric state inversion event is calculated by a second regression algorithm, where the second regression algorithm is a least squares algorithm.

9. The method according to any one of claims 1 to 8, wherein The method further comprises: Based on the temperature correction strategies corresponding to the multispectral sun photometers to be corrected, each of the multispectral sun photometers to be corrected is labeled, and based on the labeled information, a distributed correction device for each of the multispectral sun photometers to be corrected is determined; Alternatively, based on the temperature correction strategy corresponding to each of the multispectral sun photometers to be corrected, the sequence label of each of the multispectral sun photometers to be corrected is determined, and based on the sequence label of each of the multispectral sun photometers to be corrected, each of the multispectral sun photometers to be corrected is assigned to a corresponding distributed correction device for processing.

10. A photometer detection temperature correction system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.

Images