Ambient air quality monitoring data correction system

Abstract

The invention relates to an ambient air quality monitoring data correction system, which belongs to the technical field of computer data processing and modeling, and comprises an environment state judgment module used for acquiring original monitoring data of a micro station, determining a sea fog interference state mark based on the original monitoring data, and sending the sea fog interference state mark to the micro station; the original monitoring data comprises relative humidity and original mass concentration of particulate matters; the first correction module is used for processing the original mass concentration of the particulate matters in the original monitoring data when the sea fog interference state flag is 1 so as to generate first particulate matter correction data; and the second correction module is used for processing the original monitoring data to generate second correction data when the sea fog interference state mark is zero, and technical guarantee is provided for advanced applications which depend on long-term data quality, such as pollution source tracing and environmental governance effect evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of computer data processing and modeling, and in particular to an ambient air quality monitoring data correction system. Background Art

[0002] A hybrid monitoring network of reference stations and microstations is commonly used in this field to monitor the environment of industrial parks. Microstation data must be corrected by a calibration model before use. Currently, the most common calibration techniques are based on conventional multivariate linear regression methods. This method has two core flaws when dealing with complex environments such as coastal chemical parks: First, high ozone concentrations in summer can interfere with some electrochemical sensors, contaminating the model input source and causing inaccurate long-term trends in the calibrated data. Second, high-humidity coastal sea fog can cause PM2.5 optical sensor readings to be artificially high. Conventional models cannot remove this interference, resulting in false alarms in instantaneous data. Existing technologies have failed to provide an effective solution that can simultaneously address these two different physical mechanisms of interference.

[0003] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0004] The object of the present invention is to provide an ambient air quality monitoring data correction system to solve the problems raised in the above background technology.

[0005] The technical solution of the present invention comprises: an environmental state discrimination module for acquiring original monitoring data of a micro station and determining a sea fog interference state flag based on the original monitoring data, wherein the original monitoring data includes relative humidity and original mass concentration of particulate matter;

[0006] a first correction module, configured to process the original mass concentration of particulate matter in the original monitoring data to generate first particulate matter correction data when the sea fog interference state flag is 1;

[0007] a second correction module, configured to process the original monitoring data to generate second correction data when the sea fog interference state flag is 0;

[0008] The data fusion module is used to select data from the first particulate matter correction data and the second correction data based on the sea fog interference state flag to generate final correction data.

[0009] Preferably, the environmental state discrimination module determines the sea fog interference state flag, including:

[0010] Based on the original mass concentration of particulate matter, the short-time growth rate is calculated; the relative humidity is compared with the preset high humidity discrimination threshold; the short-time growth rate is compared with the preset particulate matter sudden increase discrimination threshold; when the relative humidity exceeds the high humidity discrimination threshold and the short-time growth rate exceeds the particulate matter sudden increase discrimination threshold, the sea fog interference status flag is determined to be 1; in other cases, the sea fog interference status flag is determined to be 0.

[0011] Preferably, it also includes:

[0012] an ozone disturbance discrimination unit, configured to obtain an ozone mass concentration and compare the ozone mass concentration with a preset ozone disturbance discrimination threshold value to determine an ozone disturbance state flag;

[0013] The processing of the second correction module adopts a correction model adjusted by the ozone disturbance state flag.

[0014] Preferably, the first correction module uses a sea fog interference stripping model for processing. The model constructs a correction factor that grows nonlinearly with relative humidity and divides the original mass concentration of the particulate matter by the correction factor to deduct the false increase in mass concentration caused by humidity.

[0015] Preferably, the second correction module adopts a dynamic regression model with an ozone impact attenuation factor for processing, and the second correction module first determines the dynamic attenuation factor based on the ozone mass concentration and the ozone disturbance state flag.

[0016] Preferably, the second correction module is further used to combine the dynamic attenuation factor and the preset reference regression coefficient to construct a dynamic regression model; and apply the dynamic regression model to the original monitoring data to generate second correction data.

[0017] Preferably, the reference regression coefficient is determined by screening preset learning period data during a period when the ozone mass concentration is lower than the ozone disturbance discrimination threshold, and fitting the learning period data using a standard least squares method.

[0018] Preferably, the data fusion module generates final correction data, including:

[0019] When the processing target is particulate matter, if the sea fog interference status flag is 1, the first particulate matter correction data is used as the final correction data;

[0020] When the processing target is particulate matter, if the sea fog interference status flag is 0, the particulate matter correction result in the second correction data is used as the final correction data;

[0021] When the processing target is a pollutant other than particulate matter, the corresponding pollutant correction result in the second correction data is used as the final correction data.

[0022] The present invention provides an improved ambient air quality monitoring data correction system, which has the following improvements and advantages compared with the prior art:

[0023] 1. By implementing an environmental state discrimination module, we achieve precise identification and classification of complex interference sources, laying a solid foundation for subsequent targeted corrections. This module accurately distinguishes transient data anomalies caused by high-humidity sea fog based on relative humidity and the short-term growth rate of particulate matter, preventing meteorological interference from being misidentified as pollution events. This directly improves the monitoring system's early warning accuracy and reduces unnecessary administrative and social costs caused by false alarms.

[0024] 2. The first correction module and its adopted sea fog interference stripping model effectively correct instantaneous particulate matter concentration data, ensuring the real-time accuracy of the data. In coastal areas, sea fog is the main instantaneous factor causing inaccurate PM2.5 monitoring data. The first correction module in this invention can quantitatively and nonlinearly strip away the false increase in mass concentration caused by humidity, outputting a particulate matter concentration value that is closer to physical reality. This effect is of vital application value in scenarios that rely on real-time data for emergency response and short-term forecasting.

[0025] 3. The second correction module and its built-in dynamic regression model with an ozone impact attenuation factor effectively overcome the long-term data drift caused by ozone cross-interference, ensuring the long-term stability and trend consistency of the corrected data. Especially during the peak ozone season in summer, the module can dynamically suppress the weight of the interfered sensor data in the model, ensuring that the corrected data series maintains a high degree of long-term consistency with the baseline reference station. This provides technical support for advanced applications that rely on long-term data quality, such as pollution source tracing and environmental governance effectiveness assessment.

[0026] 4. The introduction of the data fusion module establishes a logically self-consistent and strategically flexible data output endpoint, greatly enhancing the overall robustness and scenario applicability of the system. This module intelligently matches the optimal correction results for different types of pollutants based on the sea fog interference status indicators. This refined decision-making mechanism ensures that the system can provide a physically clear and logically reasonable correction value under any foreseeable environmental conditions, making the data output of the entire monitoring network unprecedentedly stable and reliable when facing different types of interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will be further explained below in conjunction with the accompanying drawings and Examples:

[0028] Figure 1 It is a flow chart of the system of the present invention. DETAILED DESCRIPTION

[0029] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0030] Example 1:

[0031] See also Figure 1 , the present invention provides an ambient air quality monitoring data correction system, comprising: an environmental state discrimination module, for acquiring original monitoring data of a micro station and determining a sea fog interference state flag based on the original monitoring data, the original monitoring data including relative humidity and original mass concentration of particulate matter;

[0032] a first correction module, configured to process the original mass concentration of particulate matter in the original monitoring data to generate first particulate matter correction data when the sea fog interference state flag is 1;

[0033] a second correction module, configured to process the original monitoring data to generate second correction data when the sea fog interference state flag is 0;

[0034] The data fusion module is used to select data from the first particulate matter correction data and the second correction data based on the sea fog interference state flag to generate final correction data.

[0035] This embodiment provides an ambient air quality monitoring data correction system. Through a logically cohesive modular data processing process, the system achieves high-precision and high-stability correction of monitoring data in complex environments. The system's architecture is based on an environmental state discrimination module that qualitatively analyzes the input raw data to determine whether sea fog interference is present. Based on this determination, the system then activates one of two dedicated correction channels. When sea fog interference is detected (i.e., the sea fog interference status flag is 1), the system activates a first correction module designed specifically to remove the effects of water vapor and perform instantaneous accuracy correction on particulate matter data. Conversely, in the absence of sea fog interference (i.e., the sea fog interference status flag is 0), the more universal second correction module is activated, responsible for long-term stability correction of various pollutants, including particulate matter. The data fusion module, serving as the decision-making output unit of the entire process, intelligently selects the best output from the two channels based on the sea fog interference status flag, integrating the final correction data into a set of data that combines instantaneous accuracy and long-term consistency. This design decomposes the complex mixed interference problem to achieve targeted processing of interference from different physical mechanisms, thereby systematically improving data quality across the entire monitoring network.

[0036] Example 2

[0037] The environmental status identification module determines the sea fog interference status flag, including:

[0038] Based on the original mass concentration of particulate matter, the short-time growth rate is calculated; the relative humidity is compared with the preset high humidity discrimination threshold; the short-time growth rate is compared with the preset particulate matter sudden increase discrimination threshold; when the relative humidity exceeds the high humidity discrimination threshold and the short-time growth rate exceeds the particulate matter sudden increase discrimination threshold, the sea fog interference status flag is determined to be 1; in other cases, the sea fog interference status flag is determined to be 0.

[0039] In this embodiment, to ensure the accuracy of subsequent correction decisions, the core function of the environmental state discrimination module is to accurately capture the characteristics of sea fog events through the simultaneous monitoring of two key physical indicators. The inherent logic is to effectively distinguish between falsely high particulate matter readings caused by sea fog and true pollution events. The basis for this module's discrimination is reflected in the following formula:

[0040]

[0041] This formula simulates the empirical judgment logic for sea fog, which is often accompanied by a sharp increase in relative humidity and a transient jump in the optical particulate matter sensor reading. Using these two as a combined criterion, sea fog interference is only confirmed when both exceed their respective thresholds simultaneously, improving the accuracy of the judgment.

[0042] It should be noted that while this embodiment uses sea fog as a typical interference scenario, its core physical criterion—the coupling of high humidity and a brief spike in particle readings—is also applicable to other types of high-humidity weather events, such as advection fog and radiation fog, that cause similar interference to optical particle sensors. Therefore, the applicability of the discrimination logic of the present invention is not limited to coastal areas but encompasses all environments where such physical interference may occur, thereby enhancing the system's universality.

[0043] Among them, S fog is the sea fog interference status flag, which is the output decision signal of the module; RH(t) is the relative humidity measured by the micro station at the current time t, in percentage%, which is the real-time input of the system; RH th The high humidity discrimination threshold is set based on the physical critical point of aerosol hygroscopic growth, usually between 85% and 90%. The value can be optimized by statistical analysis of humidity data of historical sea fog events; G th is the threshold for distinguishing sudden increase of particulate matter, which is set to distinguish the reading jump caused by sea fog from the slow accumulation of pollution sources. It is usually set to 100% or higher. This value can also be determined statistically based on the distribution of concentration change rates of the two types of events in historical data; t is the current time; G PM2.5is the short-term growth rate of PM2.5; if is used to introduce one or more conditions that must be met to trigger a specific result; AND is a logical operator, indicating that the two conditions it connects must be met; otherwise is in other cases or otherwise;

[0044] The optimization value method is as follows:

[0045] Construct a learning dataset containing at least one year of historical monitoring data from micro-stations and nearby national reference stations, and have domain experts mark all known sea fog interference events and real heavy pollution events in this dataset;

[0046] For high humidity discrimination threshold RH th and the particle matter sudden increase discrimination threshold G th , respectively set a reasonable optimization range, such as RH th ∈[80%,95%], with a step size of 1%; G th ∈[50%,200%], with a step size of 10%.

[0047] Traverse all possible (RH th ,G th ) combination, apply formula S fog =1ifRH(t)>RH th ,AND G PM2.5 >G th , else 0 discriminates the learning data set;

[0048] Traverse all possible threshold combinations, discriminate the learning data set, and calculate the true positive rate under each threshold, that is, the proportion of correctly identifying sea fog, and the false positive rate, that is, the proportion of real pollution mistakenly identified as sea fog, and draw the receiver operating characteristic curve based on this; select the point closest to the upper left corner of the curve, that is, the point with the highest true positive rate and the lowest false positive rate, and its corresponding (RH th ,G th ) is the optimal threshold combination for the site; this method transforms principled statistical analysis into a clear and reproducible optimization step;

[0049] And G PM2.5 is the short-term growth rate of PM2.5, through the formula:

[0050]

[0051] Calculation shows that here C PM2.5,raw (t) and C PM2.5,raw (t-Δt) are the original mass concentrations of PM2.5 at the current time t and the previous time t-Δt (μg / m 3), and Δt is the time window that matches the data reporting frequency, for example, 5 minutes; the module performs this logical operation after receiving real-time data, and the result S fog As a binary signal, it is directly transmitted to the subsequent correction and fusion module, acting as a switch to trigger different algorithm paths, thereby automatically identifying and marking PM2.5 false alarms caused by sea fog, avoiding unnecessary emergency responses and ensuring the reliability of the early warning system; C PM2.5,raw is the original mass concentration of PM2.5, in μg / m 3 .

[0052] Example 3

[0053] Also includes:

[0054] an ozone disturbance discrimination unit, configured to obtain an ozone mass concentration and compare the ozone mass concentration with a preset ozone disturbance discrimination threshold value to determine an ozone disturbance state flag;

[0055] The processing of the second correction module adopts a correction model adjusted by the ozone disturbance state flag.

[0056] In this embodiment, to address another type of data inaccuracy caused by chemical cross-interference, the system further integrates an ozone disturbance discrimination unit. This technical purpose is to address the cross-interference caused by high ozone concentrations in summer on some electrochemical sensors, such as NO2 and SO2 sensors. This interference can contaminate the input source of the calibration model, thereby causing long-term systematic deviations in the calibration results. The discrimination logic of this unit is defined by the following formula:

[0057]

[0058] This formula establishes a clear boundary to define when ozone concentrations are sufficient to produce non-negligible interference, which is the first step in achieving dynamic model adjustments and targeted compensation.

[0059] Among them, S O3 C is the ozone disturbance status flag, which is the command signal sent by the unit to the second correction module; O3 (t) is the ozone mass concentration measured at the microstation at the current time t (μg / m 3 );C O3,th is the ozone disturbance discrimination threshold. This threshold can be determined by referring to the national ambient air quality standard, such as the first-level concentration limit of 160 μg / m in GB3095-2012. 3 , or through experimental calibration, find the critical point of ozone concentration that can induce statistically significant deviations between other sensor readings and reference station data;

[0060] The experimental calibration method is as follows: during the learning period, the synchronous monitoring data of the micro-station and the reference station are screened; taking the affected pollutant, such as NO2, as an example, the deviation Δ between the micro-station reading and the reference station reading is calculated. NO2 =C NO2,micro -C NO2,ref ; Among them, Δ NO2 is the deviation between the micro station reading and the reference station reading; C NO2,micro is the NO2 reading of the micro station; C NO2,ref For the NO2 readings at the reference station, convert all data points to ozone concentration C O3 (t) Sort from low to high and plot Δ NO2 About C O3 (t) scatter plot; by observing or fitting the curve, find the deviation Δ NO2 The inflection point of systematic and nonlinear growth begins to appear, and the ozone concentration corresponding to this inflection point can be determined as the ozone disturbance discrimination threshold C O3,th ;

[0061] This S O3 The core function of the marker is to serve as a regulating signal for the internal model of the second correction module; once S O3 When the value is 1, the second correction module will activate its internal attenuation mechanism to weaken the weight of the interfered input item, so that the entire correction system can effectively suppress the cross-interference effect of ozone and prevent the long-term model drift caused by input data pollution from the source. This is of key value in ensuring the long-term trend consistency of data and meeting the needs of advanced data applications such as pollution tracing.

[0062] Example 4

[0063] The first correction module uses the sea fog interference stripping model for processing. The model constructs a correction factor that grows nonlinearly with relative humidity and divides the original mass concentration of particulate matter by the correction factor to deduct the false increase in mass concentration caused by humidity.

[0064] In this embodiment, when the environmental state discrimination module confirms that sea fog interference exists, that is, S fog = 1, the first correction module is activated. The core of this module is the sea fog interference removal model, which creates a mathematical tool that can quantitatively describe and remove the false readings caused by liquid water components in high humidity environments on the optical measurement method of particulate matter. The mathematical expression is as follows:

[0065]

[0066] The formula is based on the simplification and engineering application of the physical model of aerosol hygroscopic growth. Its innovation lies in the construction of a dimensionless correction factor, the denominator, which quantifies the inflated effect of humidity on PM2.5 readings when it exceeds a critical point. This effect grows nonlinearly exponentially with humidity, highly consistent with physical reality.

[0067] Among them, C PM2.5,corrA (t) is the PM2.5 concentration after correction by this channel, μg / m 3 ; C PM2.5,raw (t) is the original mass concentration of PM2.5 at the microstation in μg / m 3 ; RH(t) is the current relative humidity%; RH th is the high humidity discrimination threshold; k h is a dimensionless sea fog influence coefficient, which characterizes the average hygroscopic growth potential of aerosol components in the monitoring area; γ is also dimensionless and is called the humidity response curvature factor, which is used to adjust the nonlinear rate of correction intensity growth with humidity; the key parameter k h and γ are the unknown coefficients of the model, which need to be determined by iteratively optimizing the nonlinear optimization algorithm on the learning period data containing historical sea fog events. The optimization goal is to minimize the root mean square error between the correction value and the true value of the reference station;

[0068] The specific optimization process can use mature nonlinear least squares algorithms such as Levenberg-Marquardt to select a learning subset from historical data. This subset should contain several complete sea fog events, namely S fog The process of changing from 0 to 1 and then back to 0, and having the corresponding true value of PM2.5 concentration at the national reference station. Then, on this subset, the objective function is minimized:

[0069]

[0070] The above formula is the goal, and iterative calculation is performed to solve the optimal parameter k h and γ; where C PM2.5,corrA (t) is the correction value calculated according to formula (2), and C PM2.5,ref (t) is the true value of the reference station; RMSE is the root mean square error, which is used as the objective function for optimization; N is the total number of samples in the learning subset; t is the summed count variable, representing the time point;

[0071] The application effect of this model is reflected in the fact that fog=1 signal is triggered, by dividing the original reading by the dynamically calculated correction factor, the inflated mass concentration caused by water vapor condensation can be accurately stripped off, so that under sea fog conditions, the PM2.5 data output by the system can still be close to the true value, thereby greatly improving the instantaneous accuracy of the data and effectively suppressing the false alarm rate of the monitoring network.

[0072] Example 5

[0073] The second correction module uses a dynamic regression model with an ozone impact attenuation factor for processing, and the second correction module first determines the dynamic attenuation factor based on the ozone mass concentration and the ozone disturbance state flag;

[0074] The second correction module is further used to combine the dynamic attenuation factor and the preset baseline regression coefficient to construct a dynamic regression model; and apply the dynamic regression model to the original monitoring data to generate second correction data;

[0075] The baseline regression coefficient is determined by screening the preset learning period data during the period when the ozone mass concentration is lower than the ozone disturbance discrimination threshold, and fitting the learning period data using the standard least squares method.

[0076] In this embodiment, under all working conditions without sea fog interference, that is, S fog =0, the second correction module plays a role as the main correction channel of the system. The core technology lies in the dynamic regression model with ozone influence attenuation factor. This model is an innovation of the traditional multiple linear regression method. The construction and application of this model begins with the determination of the dynamic attenuation factor. In order to quantitatively evaluate the interference intensity of ozone and achieve a smooth transition of interference evaluation, the Sigmoid function is used to model the dynamic attenuation factor A. O3 (t) is determined by the following formula:

[0077]

[0078] In this formula, C O3 (t) and C O3,th The definition is the same as above; e is the base of natural logarithm; k O3 is the interference response sensitivity coefficient, the dimension is the inverse of the concentration (m 3 / μg) to ensure that the exponential term is dimensionless; based on this attenuation factor, a dynamic regression model is constructed. Its internal logic is to embed the attenuation factor as a regulator into the standard MLR model framework to dynamically adjust the weights of input variables affected by ozone interference; A O3 is the dynamic attenuation factor; the dynamic regression model expression is:

[0079]

[0080] In this formula, YcorrB (t) is the target pollutant concentration output after correction by this channel; X i (t) and X j (t) represents the various input variables measured by the microstation; β0 is the constant term in the baseline regression coefficient; β i is the baseline regression coefficient corresponding to the unaffected variable; X i is the unaffected part of the various input variables measured by the microstation; i is the index used to traverse each variable in the set; j is the index of the set of input variables affected by ozone cross interference; β j is the baseline regression coefficient of the affected variable; X j The affected part of each input variable measured by the micro station; Ω aff is a set of input variables that are predetermined by the sensor specifications and are subject to ozone crosstalk, and Ω unaff is the set of other unaffected variables; the core (1-λ·A O3 (t)) is the dynamic attenuation term, which directly acts on the affected variable; λ is a dimensionless global attenuation scale factor with a value in the interval [0, 1]; the baseline regression coefficients β0, β i ,β j The determination method is as follows: in the historical data of the learning period, a clean data subset under the benchmark conditions is selected. The selection criteria of this subset are to simultaneously meet the sea fog interference state flag S fog = 0 and the ozone concentration is lower than the disturbance threshold C O3 (t)≤C O3,th ; For this subset, the reference station data is used as the true value and the standard least squares method is applied to fit; such screening ensures that the benchmark coefficients are not affected by the two main interferences; while the other two unknown parameters λ and k O3 , after obtaining the benchmark coefficients, the full learning dataset containing high ozone events is used to collaboratively determine the coefficients by optimizing methods such as grid search with the goal of minimizing the correction error;

[0081] An example of the grid search method is: on a full learning dataset containing high ozone events, limit the number of fog = 0, set the search space and step size for the undetermined parameters. For example, the search range of the global attenuation scale factor λ is [0, 1], with a step size of 0.05; the interference response sensitivity coefficient k O3 The search range can be set according to the concentration unit, for example [0.005,0.1] (unit: m 3 / μg), with a step size of 0.005. O3) parameter pair combination, apply formulas (4) and (5) to calculate the correction results of the full learning data set, and determine the optimal parameter combination based on the minimum root mean square error between the correction result and the true value of the reference station;

[0082] After receiving the original data, the module O3 Sign Calculation A O3 (t) and substituted it into the dynamic regression model for correction, so that the model can actively suppress the influence of interference sources when the ozone concentration increases, ensuring the long-term trend consistency of the correction data.

[0083] Example 6

[0084] The data fusion module generates the final corrected data, including:

[0085] When the processing target is particulate matter, if the sea fog interference status flag is 1, the first particulate matter correction data is used as the final correction data;

[0086] When the processing target is particulate matter, if the sea fog interference status flag is 0, the particulate matter correction result in the second correction data is used as the final correction data;

[0087] When the processing target is a pollutant other than particulate matter, the corresponding pollutant correction result in the second correction data is used as the final correction data.

[0088] In this embodiment, the data fusion module serves as the final decision-making and output unit of the correction process. Its technical purpose is to select the most appropriate correction result for each monitored pollutant based on the current most important environmental contradiction, thereby ensuring the logical consistency of the output data. The operating logic of this module is a refined decision matrix based on the pollutant type and environmental status. When the treatment target is particulate matter PM2.5, the core criterion for decision-making is the sea fog interference status flag S fog :If S fog =1, then select the output of the first correction module, that is, Y final,PM2.5 (t) = C PM2.5,corrA (t); if S fog =0, then select the PM2.5 correction result calculated by the second correction module, that is, Y final,PM2.5 (t) = Y corrB,PM2.5 (t); For all pollutants other than PM2.5, such as NO2, SO2, O3, etc., since their measurement principles are generally not directly affected by the high humidity environment of sea fog, regardless of S fog The correction task is always completed by the second correction module, and the output is constantly taken from channel B, that is, Y final,Y≠PM2.5 (t) = Y corrB,Y(t); Through this intelligent integration of pollutant type differentiation, it is ensured that under any environmental conditions, each pollutant data output from the system undergoes the most appropriate physical process correction, thereby achieving logically self-consistent and physically clear corrections for all monitored pollutants, forming the technical closed loop of the present invention;

[0089] As a preferred embodiment, in order to avoid the sea fog state mark S fog At the moment of switching, for example, RH is at RH th When the relative humidity RH fluctuates around RH, the final output of the particle correction data will produce step or burrs. A smooth transition mechanism can be introduced in the data fusion module. For example, a relative humidity RH is defined as the input and the relative humidity RH is defined as the input. th The weight function W(RH) with smooth transition near RH is much lower than RH. th When W(RH)≈0, when RH is much higher than RH th When W(RH)≈1, the particle correction data can be obtained by weighted fusion of the outputs of the two modules:

[0090] Y final,PM2.5 (t)=(1-W(RH))·Y corrB,PM2.5 (t)+W(RH)·C PM2.5,corrA (t)

[0091] Among them, Y final,PM2.5 The final output of the particle correction data; W(RH) is a relative humidity RH input, th The weight function of the smooth transition near RH is relative humidity; Y corrB,PM2.5 C is the PM2.5 correction result calculated by the second correction module; PM2.5,corrA This is the PM2.5 correction output of the first correction module. This approach can ensure a smooth data transition at the state switching boundary, thereby improving the continuity and quality of the final output data sequence.

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An ambient air quality monitoring data correction system, characterized in that: include: An environmental status identification module is used to obtain the original monitoring data of the micro-station and determine the sea fog interference status flag based on the original monitoring data. The original monitoring data includes relative humidity and original mass concentration of particulate matter; a first correction module, configured to process the original mass concentration of particulate matter in the original monitoring data to generate first particulate matter correction data when the sea fog interference state flag is 1; a second correction module, configured to process the original monitoring data to generate second correction data when the sea fog interference state flag is 0; The data fusion module is used to select data from the first particulate matter correction data and the second correction data based on the sea fog interference state flag to generate final correction data.

2. The ambient air quality monitoring data correction system according to claim 1, characterized in that: The environmental state discrimination module determines the sea fog interference state flag, including: Based on the original mass concentration of particulate matter, the short-time growth rate is calculated; the relative humidity is compared with the preset high humidity discrimination threshold; the short-time growth rate is compared with the preset particulate matter sudden increase discrimination threshold; when the relative humidity exceeds the high humidity discrimination threshold and the short-time growth rate exceeds the particulate matter sudden increase discrimination threshold, the sea fog interference status flag is determined to be 1; in other cases, the sea fog interference status flag is determined to be 0.

3. The ambient air quality monitoring data correction system according to claim 1, characterized in that: Also includes: an ozone disturbance discrimination unit, configured to obtain an ozone mass concentration and compare the ozone mass concentration with a preset ozone disturbance discrimination threshold value to determine an ozone disturbance state flag; The processing of the second correction module adopts a correction model adjusted by the ozone disturbance state flag.

4. The ambient air quality monitoring data correction system according to claim 1, characterized in that: The first correction module uses a sea fog interference stripping model for processing. The model constructs a correction factor that grows nonlinearly with relative humidity and divides the original mass concentration of particulate matter by the correction factor to deduct the false increase in mass concentration caused by humidity.

5. The ambient air quality monitoring data correction system according to claim 3, characterized in that: The second correction module uses a dynamic regression model with an ozone impact attenuation factor for processing, and the second correction module first determines the dynamic attenuation factor based on the ozone mass concentration and the ozone disturbance state flag.

6. The ambient air quality monitoring data correction system according to claim 5, characterized in that: The second correction module is further used to construct a dynamic regression model by combining the dynamic attenuation factor and a preset reference regression coefficient; and apply the dynamic regression model to the original monitoring data to generate second correction data.

7. The ambient air quality monitoring data correction system according to claim 6, characterized in that: The reference regression coefficient is determined by screening the preset learning period data during the period when the ozone mass concentration is lower than the ozone disturbance discrimination threshold, and fitting the learning period data using the standard least squares method.

8. The ambient air quality monitoring data correction system according to claim 1, characterized in that: The data fusion module generates final correction data, including: When the processing target is particulate matter, if the sea fog interference status flag is 1, the first particulate matter correction data is used as the final correction data; When the processing target is particulate matter, if the sea fog interference status flag is 0, the particulate matter correction result in the second correction data is used as the final correction data; When the processing target is a pollutant other than particulate matter, the corresponding pollutant correction result in the second correction data is used as the final correction data.

Images