A pm2.5 data calibration method and system for pollution peaks

Abstract

The present application relates to the technical field of data processing, in particular to a PM2.5 data calibration method and system for pollution peak value. In the present application, the logic of constructing a dynamic interval based on model operation fine particulate matter fluctuation characteristics can capture the nonlinear evolution law and thus get rid of the static threshold limit, the way of constructing a plane topological network for suspected peak value and measuring the spatial diffusion gradient value effectively fuses the core geometric constraint condition, the mechanism of deriving the directional flux value relying on the sine constant of the internal angle of a triangle and generating the dispersity scalar to participate in the benchmark comparison, the process of converting the simple concentration comparison into the quantitative evaluation process of the transport intensity with meteorological physical support, the accurate distinction between the real extreme value driven by the meteorological external cause and the false burr caused by the equipment failure, the targeted replacement and retention operation according to the identification result, the breakthrough of the technical barrier that the conventional means is easy to confuse the true and false of the peak value, and the physical real property and the traceability analysis accuracy of the data calibration sequence are effectively ensured.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for calibrating PM2.5 data for pollution peaks. Background Technology

[0002] In the field of data processing technology, in environmental monitoring and air quality assessment scenarios, data processing typically revolves around the deployment of sensors at monitoring stations to sample raw concentration records, database creation and storage, calculation of daily and hourly average values, and comparison of historical data. It also combines meteorological statistical analysis and forecasting methods to organize and analyze the changing trends of pollutant concentrations in order to form a continuous and stable air quality data sequence.

[0003] Among them, the PM2.5 data calibration method for pollution peaks in pollution meteorology refers to the process of manually comparing values ​​from adjacent time periods, setting a fixed threshold range, calculating the difference by combining the historical average concentration for the same period, using a sliding time window to calculate the average value of several sampling points before and after, replacing or correcting values ​​that exceed the limit, truncating records that exceed the range according to the measurement range in the sampling equipment manual, and comparing data from multiple monitoring stations in the same monitoring area hourly to determine whether abnormal peaks should be retained or adjusted, thereby reorganizing and calibrating the original PM2.5 concentration sequence.

[0004] Existing technologies primarily rely on a comparison and discrimination model that uses fixed thresholds and sliding time windows to calculate average values ​​in the monitoring data processing stage. The logic based on static parameters is difficult to adapt to the nonlinear fluctuation characteristics of fine particulate matter concentration in complex environments. Furthermore, in the identification of abnormal peaks, relying solely on hourly one-way comparisons of multi-site data deviates from the actual diffusion and attenuation laws of spatial pollutants and real meteorological evolution conditions in pollution meteorology. When faced with terrain obstacles or cross-regional advection transport scenarios, it is easy to confuse false spikes caused by local faults with real pollution peaks driven by external weather factors. This results in the output calibration data sequence frequently exhibiting deviations such as erroneous deletion of real extreme values ​​or omission of fault interference, severely weakening the physical authenticity of the data and the reliability of subsequent pollution source tracing operations. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and to propose a PM2.5 data calibration method and system for pollution peaks.

[0006] To achieve the above objectives, the present invention employs the following technical solution: a PM2.5 data calibration method for pollution peaks, comprising the following steps:

[0007] S1: Obtain the continuous PM2.5 data sequence of the target station, extract the corresponding fluctuation characteristics from it, calculate the conditional variance, compare the conditional variance with the preset constant, and construct the dynamic confidence discrimination interval of pollution peak;

[0008] S2: When the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence discrimination interval of the pollution peak, construct the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station, integrate the synchronous concentration information of the second two-dimensional spatial plane coordinates, and generate basic topology parameters.

[0009] S3: Based on the aforementioned basic topological parameters, determine the spatial diffusion gradient value of the pollution peak by using the Euclidean distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates;

[0010] S4: Construct a triangular network based on the basic topological parameters and select the minimum interior angle value. Calculate the directional flow flux value based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak.

[0011] S5: Summing the directional flow flux values ​​of multiple sets to generate a global convergence dispersion scalar, comparing it with the preset frontal advection critical benchmark to determine the authenticity of the pollution peak, and performing replacement or retention on the real-time PM2.5 data concentration to obtain the pollution peak data calibration processing sequence.

[0012] As a further aspect of the present invention, the pollution peak dynamic confidence discrimination interval includes an upper limit and a lower limit of the discrimination interval, the basic topological parameters include a first two-dimensional spatial plane coordinate, a second two-dimensional spatial plane coordinate and synchronous concentration information, the pollution peak spatial diffusion gradient value is specifically calculated by Euclidean space straight-line distance, the directional flow flux value is specifically calculated by a sine constant and the pollution peak spatial diffusion gradient value, and the pollution peak data calibration processing sequence includes replacement concentration data and retention concentration data.

[0013] As a further aspect of the present invention, the step of obtaining the dynamic confidence interval of the pollution peak is specifically as follows:

[0014] S111: Set a sliding time window to obtain the continuous PM2.5 data sequence of the target station within the current window period, extract the attached time node parameters from it, call the preset generalized autoregressive conditional heteroscedasticity model to calculate the time-series change state variance term of the continuous PM2.5 data sequence, and use it as the concentration fluctuation characteristic of PM2.5 data.

[0015] S112: Extract the heteroscedastic component values ​​of the concentration fluctuation feature through the generalized autoregressive conditional heteroscedasticity model, perform numerical conditional fitting on the heteroscedastic component values, and deduce the conditional variance values.

[0016] S113: Calculate the constant sum and constant difference components of the conditional variance value and the preset initial constant scalar, combine the constant sum and constant difference components to determine the upper and lower limits of the confidence discrimination of the data distribution, and obtain the dynamic confidence discrimination interval of the pollution peak.

[0017] As a further aspect of the present invention, the step of obtaining the basic topology parameters specifically includes:

[0018] S211: Obtain the real-time PM2.5 data concentration of the target station and the dynamic confidence interval of the pollution peak, compare the real-time PM2.5 data concentration with the dynamic confidence interval of the pollution peak to determine the suspected pollution peak triggering state, and obtain the first geographic latitude and longitude coordinates of the target station under the real-time PM2.5 data concentration peak triggering condition, and project and transform them into the first two-dimensional spatial plane coordinates.

[0019] S212: Use the Deloni triangulation algorithm to establish the network topology architecture of the first two-dimensional spatial plane coordinates, search for first-order adjacent stations with the target station along the network topology architecture, obtain the second two-dimensional spatial plane coordinates of the first-order adjacent stations and the PM2.5 data concentration of the adjacent stations, and obtain the characteristic parameters of the adjacent stations.

[0020] S213: Extract the first two-dimensional spatial plane coordinates corresponding to the target station, and concatenate the first two-dimensional spatial plane coordinates with the second two-dimensional spatial plane coordinates in the adjacent station feature parameters and the adjacent synchronous PM2.5 data concentration to obtain the basic topology parameters.

[0021] As a further aspect of the present invention, the step of obtaining the spatial diffusion gradient value of the pollution peak is specifically as follows:

[0022] S311: Read the real-time PM2.5 data concentration of the target station and the synchronous PM2.5 data concentration of the first-order adjacent station from the basic topology parameters, calculate the absolute value of the scalar difference characteristic between the two types of data concentrations, and obtain the absolute concentration deviation value.

[0023] S312: Obtain the basic topology parameters along with the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates, and calculate the Euclidean distance parameter between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates;

[0024] S313: Calculate the ratio between the absolute concentration deviation value and the Euclidean distance parameter as the concentration change data under Euclidean distance, associate the concentration change data with the two-dimensional spatial plane formed by the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates, and calculate the pollution peak spatial diffusion gradient value by combining the concentration evolution attribute of the corresponding topological direction in the two-dimensional spatial plane.

[0025] As a further aspect of the present invention, the step of obtaining the directional flow flux value specifically includes:

[0026] S411: Construct a local Deloni triangle structure with the target station as the vertex based on the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates in the basic topology parameters. Scan multiple sets of internal angle values ​​of the triangle structure and sort them from smallest to largest. Extract the smallest interior angle feature value at the beginning of the sorted sequence.

[0027] S412: Based on the minimum interior angle characteristic value and the pollution peak spatial diffusion gradient value, calculate the sinusoidal trigonometric proportional constant of the corresponding diffusion direction, obtain the preset environmental diffusion coefficient of the same region as the pollution peak spatial diffusion gradient value, and establish the scalarized basic diffusion flux.

[0028] S413: Extract the sine trigonometric proportionality constant associated with the minimum interior angle feature value, calculate the product between the two parameters of the scalarized basic diffusion flux and the sine trigonometric proportionality constant, characterize the state of pollution peak contraction due to the obstruction of narrow topological space, and obtain the directional flow flux value.

[0029] As a further aspect of the present invention, the step of obtaining the pollution peak data calibration processing sequence specifically includes:

[0030] S511: Collect multiple sets of the directional flow flux values ​​around the target site, calculate the sum of all the directional flow flux values ​​to quantify the overall diffusion and convergence intensity of the spatial site cluster, and obtain the global convergence dispersion scalar.

[0031] S512: Compare the global convergence dispersion scalar with the preset frontal advection critical benchmark. If the current global convergence dispersion scalar is less than the frontal advection critical benchmark, it is determined that the real-time PM2.5 data concentration belongs to a false pollution peak caused by a local fault. If it reaches the frontal advection critical benchmark, it is determined to be a real peak. Establish a peak authenticity identification mark.

[0032] S513: Based on the peak authenticity identification identifier, for false pollution peak states, retrieve the upper limit value of the corresponding interval of the dynamic confidence judgment interval of pollution peak, apply the upper limit value of the interval to replace the real-time PM2.5 data concentration to perform abnormal concentration data coverage calibration, and for true pollution peak states, maintain the original real-time PM2.5 data concentration characteristics in a constant retention state to obtain the pollution peak data calibration processing sequence.

[0033] A PM2.5 data calibration system for pollution peaks, the system comprising:

[0034] The confidence discrimination module acquires continuous PM2.5 data sequences from the target station, extracts the corresponding fluctuation characteristics to calculate the conditional variance, compares the conditional variance with a preset constant, and constructs a dynamic confidence discrimination interval for pollution peaks.

[0035] The topology construction module constructs the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station when the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence discrimination interval of the pollution peak. It integrates the synchronous concentration information of the second two-dimensional spatial plane coordinates to generate basic topology parameters.

[0036] The gradient calculation module, based on the basic topological parameters, determines the spatial diffusion gradient value of the pollution peak by using the Euclidean straight-line distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates.

[0037] The flux calculation module constructs a triangular network based on the basic topology parameters and filters the minimum interior angle values. Based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak, it calculates the directional flow flux value.

[0038] The calibration judgment module sums multiple sets of directional flow flux values ​​to generate a global convergence dispersion scalar, compares it with a preset frontal advection critical benchmark to determine the authenticity of the pollution peak, and performs replacement or retention on the real-time PM2.5 data concentration to obtain a pollution peak data calibration processing sequence.

[0039] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0040] In this invention, the logic of constructing a dynamic interval based on the fluctuation characteristics of fine particulate matter using model calculations can capture nonlinear evolution patterns and thus overcome the limitations of static thresholds. The method of constructing a planar topological network for suspected peaks and calculating spatial diffusion gradient values ​​effectively integrates core geometric constraints. The mechanism of deriving directional flux values ​​based on the sine constant of the interior angles of a triangle and generating a discrete scalar to participate in benchmark comparison transforms simple concentration comparison into a quantitative assessment process of transport intensity supported by pollution meteorology and meteorological physics. It accurately distinguishes between true extreme values ​​driven by meteorological externalities and false spikes caused by equipment failures. Based on the identification results, targeted replacement and retention operations are performed, overcoming the technical barrier that conventional methods easily confuse the authenticity of peaks and effectively ensuring the physical authenticity of the data calibration sequence and the accuracy of source tracing analysis. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] Please see Figure 1 This invention provides a technical solution, a method for calibrating PM2.5 data for pollution peaks, comprising the following steps:

[0044] S1: Obtain the continuous PM2.5 data sequence of the target station, extract the corresponding fluctuation characteristics from it, calculate the conditional variance, compare the conditional variance with the preset constant, and construct the dynamic confidence discrimination interval of pollution peak;

[0045] S2: When the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence interval of the pollution peak, construct the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station, integrate the synchronous concentration information of the second two-dimensional spatial plane coordinates, and generate basic topological parameters.

[0046] S3: Based on the basic topological parameters, the Euclidean distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates is used to determine the spatial diffusion gradient value of the pollution peak.

[0047] S4: Construct a triangular network based on the basic topology parameters and select the minimum interior angle values. Calculate the directional flow flux value based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak.

[0048] S5: Summing multiple sets of directional flow flux values ​​to generate a global convergence dispersion scalar, comparing it with the preset frontal advection critical benchmark to determine the authenticity of the pollution peak, performing replacement or retention on the real-time PM2.5 data concentration, and obtaining the pollution peak data calibration processing sequence.

[0049] The dynamic confidence interval for pollution peaks includes an upper limit and a lower limit. The basic topological parameters include the first two-dimensional spatial plane coordinates, the second two-dimensional spatial plane coordinates, and synchronous concentration information. The spatial diffusion gradient value of the pollution peak is specifically calculated using the straight-line distance in Euclidean space. The directional flow flux value is specifically calculated using the sine constant and the spatial diffusion gradient value of the pollution peak. The pollution peak data calibration processing sequence includes replacement concentration data and retained concentration data.

[0050] The specific steps for obtaining the dynamic confidence interval for pollution peaks are as follows:

[0051] S111: Set a sliding time window to obtain the continuous PM2.5 data sequence of the target station within the current window period, extract the attached time node parameters from it, call the preset generalized autoregressive conditional heteroscedasticity model to calculate the time-series change state variance term of the continuous PM2.5 data sequence, and use it as the concentration fluctuation characteristic of PM2.5 data.

[0052] The time-series state variance term of the continuous PM2.5 data series is calculated by calling the preset generalized autoregressive conditional heteroscedasticity model. Specifically:

[0053] The generalized autoregressive conditional heteroscedasticity model is set by pre-configured autoregressive order parameters and moving average order parameters;

[0054] Input continuous PM2.5 data series into a generalized autoregressive conditional heteroscedasticity model with set parameters;

[0055] By combining the autoregressive order parameter and the moving average order parameter, the time-series change state variance term is extracted from the continuous PM2.5 data series.

[0056] Continuous fine particulate matter (FPM) data sequences recorded at a fixed sampling frequency within a sliding time window are extracted. Simultaneously, the timestamps corresponding to each concentration value in the continuous FPM data sequences are extracted to form time node parameters. The time window length and sampling frequency are set by statistically analyzing the average historical fluctuation period of the target station's meteorological data. This statistically derived average historical fluctuation period is used as the window length, and the sampling frequency is set to a fixed time interval. A pre-configured generalized autoregressive conditional heteroscedasticity (GHP) model is invoked. This model structure consists of a data input layer, a fluctuation fitting layer, and a variance output layer. The model execution process involves first receiving the sequence through the data input layer, then the fluctuation fitting layer calculating the mean residual of adjacent time series data, and finally, the variance output layer integrating the residuals with the historical variance for smoothing iteration. The model is dimensionally limited by the autoregressive order parameter and the moving average order parameter. The autoregressive order parameter is determined by extracting historical fine particulate matter monitoring data from the target site, calculating the partial autocorrelation distribution term (PAT) of the correlation strength between data points at different time points in the dataset, and statistically analyzing the standard deviation of the white noise sequence in the historical fine particulate matter monitoring data set, using this standard deviation as a preset constant limit. The lag order of the PAT distribution term, which decays to within the preset constant limit, is then assigned to the autoregressive order parameter. Several possible orders exist: if the PAT value decays rapidly to below the limit within the range of 1 to 2, it indicates extremely strong short-term correlation, and the order is extracted as 1 or 2; if it decays slowly within the range of 3 to 5, it indicates a long memory effect, and the order is extracted as 3 to 5. The corresponding order is extracted based on the actual decay results and assigned to the autoregressive order parameter. The moving average order parameter is calculated similarly, calculating the lag order of the decaying autocorrelation distribution term and assigning it. The continuous fine particulate matter data sequence is then input into the generalized autoregressive conditional heteroscedasticity model with the pre-defined autoregressive and moving average order parameters. During the calculation, the residual data derived by subtracting the concentration value at the current time from the concentration value at the previous time, based on the autoregressive order parameter and the moving average order parameter, is extracted and substituted into the variance calculation step to complete the weighted summation of each value to obtain the time-series iteration state. To more accurately extract the time-series iteration state, a generalized autoregressive conditional heteroscedasticity nonlinear update algorithm is introduced, with the corresponding calculation formula as follows: The specific settings and physical meanings of each parameter in the formula are explained in detail: The variance term representing the temporal change state at the current time point is obtained through the model output. The weights for the squared residuals are determined by the arithmetic mean of the short-term extreme fluctuation ranges in the statistical historical data series. When the average is between 10.0 and 20.0, it is considered a slight fluctuation, and the weight is set between 0.0 and 0.15. When the average is between 20.0 and 50.0, it is considered an extreme fluctuation, and the weight is set between 0.15 and 0.4. In this embodiment, the weight is assigned as 0.2 based on the arithmetic mean of the fluctuation range of the target site being 30.0. This indicates that the residual data generated from the extraction is obtained by directly subtracting the concentration from the previous time from the current concentration. The weighting value represents the variance at the previous moment. It is set based on the average duration of the stable concentration in the historical monitoring data. The average duration is weakly stable in the range of 1 to 5 hours, and the weighting value is 0.4 to 0.6. The duration is strongly stable in the range of 5 to 12 hours, and the weighting value is 0.6 to 0.9. In this embodiment, the average duration is 8 hours and the weighting value is 0.8. This indicates the initial variance term generated in the previous time step, which is directly extracted from the historical operation state; The extreme mutation compensation coefficient is set based on the average probability of extreme dust storms that have occurred in the target geographical area in the past. The probability is 0 when it is between 0% and 1%, and 0.1 to 0.2 when it is between 1% and 5%. In this embodiment, the probability is 0% when it is 0%. To prevent arithmetic overflow caused by a denominator of 0 when the residual data is completely zero during the calculation, a small positive real number compensation constant is set to 0.01 in this embodiment. The generalized autoregressive conditional heteroscedasticity model labels the final variance term output after formula calculation as the concentration fluctuation characteristic of the continuous fine particulate matter data sequence. Subtracting the first monitoring value 45 and the second monitoring value 50 in the sequence yields a residual value of 5.0. The initial variance term is set to 0. Substituting the extracted data into the corresponding calculation formula yields the result: The final calculated output of the concentration fluctuation characteristic value is 5.0.

[0057] S112: Extract the heteroscedastic component values ​​of concentration fluctuation characteristics through a generalized autoregressive conditional heteroscedasticity model, perform numerical conditional fitting on the heteroscedastic component values, and deduce the conditional variance values.

[0058] The extracted concentration fluctuation characteristics are analyzed to extract heteroscedasticity components, which measure the discrete dimension of the data. Numerical conditional fitting is then performed on these heteroscedasticity components to derive the conditional variance. The heteroscedasticity components are extracted by calculating the square root of the extracted concentration fluctuation characteristics and removing absolute dimension bias. The specific process of performing numerical conditional fitting involves setting conditional fitting constant parameters and conditional fitting slope parameters. The value of the conditional fitting constant parameter is determined by collecting annual fine particulate matter fluctuation range data sets from independent monitoring stations surrounding the target station, calculating the standard deviation of the range data set, and multiplying it by a spatial heterogeneity difference statistical index as a specific multiple. Several possibilities exist: if the spatial heterogeneity difference statistical index is in the range of 0.1 to 0.3, it indicates extremely similar climates between stations, and a multiple of 0.3 is used; if it is in the range of 0.3 to 0.6, it indicates some climatic barriers, and a multiple of 0.5 is used; if it is in the range of 0.6 to 0.9, it indicates severe geographical barriers, and a multiple of 0.8 is used. Based on the statistical heterogeneity, a corresponding multiple is extracted. The standard deviation of the calculated range dataset is multiplied by the determined multiple to obtain the conditional fitting constant parameter. The value of the conditional fitting slope parameter is determined based on the average proportional change rate of the range parameters at adjacent time points in the extracted annual fine particulate matter fluctuation range dataset. Several possibilities exist: if the average proportional change rate is greater than 1.0, it indicates an increasing divergence trend in fluctuations, and the slope is set between 1.0 and 1.5; if the average proportional change rate is between 0.5 and 1.0, it indicates a gradual convergence of fluctuations, and the slope is set between 0.5 and 1.0. The appropriate situation is determined based on the actual extracted average proportional change rate, and the corresponding conditional fitting slope parameter is set. The input heteroscedasticity component values ​​are extracted, multiplied by the conditional fitting slope parameter, and then the product is added to the conditional fitting constant parameter. This completes the univariate first-order linear fitting calculation process, and the final combined output value is determined as the conditional variance value.

[0059] S113: Calculate the constant sum and constant difference components of the conditional variance value and the preset initial constant scalar, combine the constant sum and constant difference components to determine the upper and lower limits of the confidence discrimination of the data distribution, and obtain the dynamic confidence discrimination interval of the pollution peak.

[0060] The conditional variance values ​​output from the univariate first-order linear fitting calculation are extracted, and a preset initial constant scalar is retrieved. The initial constant scalar is set by using the set of baseline variance values ​​from the past few years' air quality monitoring data of the geographical area where the target station is located, during periods of stable pollution. The arithmetic mean of the baseline variance values ​​is calculated and assigned to the initial constant scalar. The conditional variance values ​​and the initial constant scalar are superimposed to obtain the constant sum component, and simultaneously, the conditional variance values ​​and the initial constant scalar are subtracted to obtain the constant difference component. The average actual fine particulate matter concentration at the last time node of the time window is retrieved to form a baseline anchor point. Addition and subtraction operations are performed between the baseline anchor point and the constant difference component to obtain the lower confidence limit of the data distribution. Addition operations are performed between the baseline anchor point and the constant sum component to obtain the upper confidence limit of the data distribution. The extracted lower and upper confidence limits of the data distribution are combined to establish the corresponding dynamic confidence interval for pollution peaks.

[0061] The specific steps for obtaining the basic topology parameters are as follows:

[0062] S211: Obtain the real-time PM2.5 data concentration and the dynamic confidence interval of the pollution peak at the target site, compare the real-time PM2.5 data concentration and the dynamic confidence interval of the pollution peak to determine the suspected pollution peak triggering state, and obtain the first geographic latitude and longitude coordinates of the target site under the real-time PM2.5 data concentration peak triggering condition, and project and transform them into the first two-dimensional spatial plane coordinates.

[0063] The system acquires real-time fine particulate matter (PM2.5) concentration data directly from the sensor output at the target site and retrieves the dynamic confidence interval for pollution peaks established by previous operations. It then compares the real-time PM2.5 concentration with the upper and lower confidence limits set within the dynamic confidence interval. Several possibilities exist for the comparison results: if the real-time concentration is below the lower confidence limit, it indicates an abnormal drop or loss of monitoring data, triggering a low-value warning; if the real-time concentration is between the lower and upper confidence limits, it indicates the monitoring data is within a normal and reasonable fluctuation range, and a release write command is executed; if the real-time concentration is above the upper confidence limit, it indicates a high-concentration anomaly, confirming the target site is currently in a suspected pollution peak trigger state. Based on the actual relationship between the acquired real-time concentration and the confidence interval, under the peak trigger condition where the real-time PM2.5 concentration is above the upper confidence limit, the system retrieves the target site's equipment attribute information database and extracts the target site's first geographic latitude and longitude coordinates. It then extracts the transformation matrix parameters for mapping latitude and longitude spherical coordinates to a planar grid and the scale factor of the corresponding central meridian. The transformation matrix parameters and scaling factors are set based on arithmetic derivation of the semi-major axis and flattening parameters of the Earth's reference ellipsoid. The first geographic latitude and longitude coordinates of the target station are input into a polynomial expansion conversion process from spherical to planar coordinates. Latitude and longitude Gaussian zonal mapping is performed, and numerical superposition is performed using the extracted lateral offset parameters. Then, geometric mapping projection expansion is performed from Earth's spherical coordinates to two-dimensional planar coordinates. After completing the geometric mapping projection expansion, the horizontal and vertical coordinate values ​​of the plane are output as the first two-dimensional spatial planar coordinates.

[0064] S212: Use the Deloni triangulation algorithm to establish the network topology architecture of the first two-dimensional spatial plane coordinates, search for first-order adjacent stations with the target station along the network topology architecture, obtain the second two-dimensional spatial plane coordinates of the first-order adjacent stations and the PM2.5 data concentration of the adjacent stations, and obtain the characteristic parameters of the adjacent stations.

[0065] Extract the first two-dimensional spatial plane coordinates and the pre-stored two-dimensional spatial plane coordinates of surrounding stations. Perform a circumcircle partitioning search operation on the set of plane coordinates. The specific process of partitioning search is as follows: connect any three plane coordinate points as vertices to form a triangle structure, calculate the intersection of the perpendicular bisectors of the line segments formed by the three vertices, and then extract the radius and center coordinates of the circumcircle of the triangle structure. Verify whether there are other coordinate points in the set of plane coordinates inside the circumcircle. There are several possibilities for the verification results: if the circumcircle contains other coordinate points, it is determined that the current triangle structure does not meet the empty circumcircle criterion, and the edge connection of the triangle is cancelled; if the circumcircle does not contain any other coordinate points, it is determined that the triangle structure meets the optimal partitioning condition and is retained. After completing the empty verification of all coordinate points and retaining qualified triangles, construct the network topology architecture of the first two-dimensional spatial plane coordinates. After the network topology architecture is constructed, search for the coordinates of adjacent nodes that have a direct edge connection relationship with the first two-dimensional spatial plane coordinates of the target station along the edge coordinate array output by the network topology architecture. Determine the adjacent nodes with a direct single-segment edge connection relationship as first-order adjacent stations. The second two-dimensional spatial plane coordinates of each first-order adjacent site and the adjacent synchronous fine particulate matter data concentrations detected during the same period are obtained. All extracted second two-dimensional spatial plane coordinates and adjacent synchronous fine particulate matter data concentrations are combined and categorized into corresponding arrays to obtain multiple sets of adjacent site feature parameters.

[0066] S213: Extract the first two-dimensional spatial plane coordinates corresponding to the target station, and concatenate the first two-dimensional spatial plane coordinates with the second two-dimensional spatial plane coordinates in the feature parameters of the adjacent stations and the PM2.5 data concentration of the adjacent synchronous stations to obtain the basic topology parameters;

[0067] Extract the neighboring site feature parameters generated during the network topology circumcircle decomposition retrieval operation. For each group of neighboring site feature parameters, place the first two-dimensional spatial plane coordinates corresponding to the target site alongside the second two-dimensional spatial plane coordinates and the concentration of adjacent synchronous fine particulate matter data from the corresponding neighboring site feature parameters. Establish a mapping field concatenation instruction, and concatenate the data into a single row according to the order of target coordinates, adjacent coordinates, and synchronous concentration to generate a basic topology parameter record. Repeat the concatenation process to generate corresponding basic topology parameter records for each group of feature parameters. Connect and integrate the generated basic topology parameter records, and perform a data block aggregation and packaging operation to establish a complete set of basic topology parameters containing all topological relationships and concentration information.

[0068] The specific steps for obtaining the spatial diffusion gradient value of the pollution peak are as follows:

[0069] S311: Read the real-time PM2.5 data concentration of the target station and the synchronous PM2.5 data concentration of the first-order adjacent station from the basic topology parameters, calculate the absolute value of the scalar difference characteristic between the two types of data concentrations, and obtain the absolute concentration deviation value.

[0070] Read the adjacent synchronous fine particulate matter concentration values ​​of each first-order adjacent station recorded in the basic topology parameter set. For each first-order adjacent station, perform a subtraction operation between the real-time fine particulate matter concentration value of the target station and the adjacent synchronous fine particulate matter concentration value of the corresponding first-order adjacent station. Discard the absolute value of the difference to obtain and save the absolute concentration deviation value of each first-order adjacent station.

[0071] S312: Obtain the basic topology parameters including the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates, and calculate the Euclidean distance parameter between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates;

[0072] Simultaneously extract the lateral and longitudinal coordinates of the second two-dimensional spatial plane coordinates of each first-order adjacent station from the basic topology parameter set. Calculate the lateral and longitudinal coordinate differences between the target station and the first-order adjacent stations respectively. Square the two sets of differences by multiplying them by themselves, perform an addition and superposition command on the two squared items, and then perform a square root algebraic operation on the summation result. Output the Euclidean distance parameters between each first-order adjacent station and the target station in sequence.

[0073] S313: Calculate the ratio between the absolute concentration deviation value and the Euclidean distance parameter as the concentration change data under Euclidean distance. Associate the concentration change data with the two-dimensional spatial plane formed by the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates. Combine the concentration evolution attribute of the corresponding topological direction in the two-dimensional spatial plane to calculate the spatial diffusion gradient value of the pollution peak.

[0074] Extract the saved set of absolute concentration deviation values ​​and the set of Euclidean distance parameters obtained through algebraic operations. Here, to accurately map spatial correlation characteristics, a formula for calculating the spatial diffusion gradient of concentration changes with topological damping adaptation is introduced: The physical meaning and value settings of each letter parameter in the formula are explained below: This represents the spatial diffusion gradient value of the pollution peak, which serves as the final output of the algorithm, reflecting the spatial concentration decay rate. This represents the absolute concentration deviation between the corresponding adjacent site and the target site. It is obtained by extracting the concentration difference between the two sites through prior arithmetic operations and removing the absolute value sign. The Euclidean distance parameter between the coordinates of two stations in the current calculation direction is derived from the algebraic operation of subtraction, multiplication, addition, and square root extraction of the two-dimensional spatial plane coordinates. This is the spatial topological damping coefficient, which is set based on the average elevation difference of the physical obstacles in the terrain between the two stations. Several possibilities exist: if the average elevation difference is between 100.0 meters and 500.0 meters, indicating a mountainous barrier, the coefficient is set between 0.5 and 0.7 to weaken the diffusion effect; if the average elevation difference is between 0.0 meters and 100.0 meters, indicating flat and open terrain, the coefficient is set between 0.7 and 1.0. In this embodiment, the target station is located in flat and open terrain with an average elevation difference of 10.0 meters, so the value is set to 1.0. This is a small distance compensation constant, primarily used to prevent boundary conditions where the denominator is 0 and the division operation crashes due to the complete overlap of the two-dimensional spatial coordinates of two stations caused by data anomalies. If the grid is dense, the value range is set to 0.1 to 0.5; if the grid is sparse, the value range is set to 0.0 to 0.1. In this embodiment, it is set to 0.0 for a normal topological grid. For the corresponding first-order adjacent stations, the corresponding absolute concentration deviation value of 10.0 and the Euclidean distance parameter of 5385.16 are extracted, and combined with the spatial environment characteristics, they are substituted into the corresponding calculation formula to give the results: The ratio 0.00185 obtained from the formula is used as the concentration change data for the corresponding first-order adjacent site. The concentration change data is matched to the two-dimensional space plane where the boundary line connecting the first two-dimensional spatial plane coordinates of the target site and the second two-dimensional spatial plane coordinates of the first-order adjacent site is located, generating the spatial diffusion gradient value of the pollution peak in the corresponding topological direction.

[0075] The specific steps for obtaining the value of targeted circulation flux are as follows:

[0076] S411: Construct a local Deloni triangle structure with the target station as the vertex based on the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates in the basic topology parameters. Scan multiple sets of internal angle values ​​of the triangle structure and sort them from smallest to largest. Extract the smallest interior angle feature value at the beginning of the sorted sequence.

[0077] The first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of each first-order adjacent station are extracted from the basic topology parameters. In the layer of the two-dimensional coordinate system plane, the target station coordinates are used as a single common vertex, and the virtual connections between adjacent first-order adjacent stations are used as the base to generate a local Deloni triangle structure. The straight-line distances between adjacent first-order adjacent stations are calculated. Combining the Euclidean distance parameters from each station to the target station, the cosine theorem is used to calculate the cosine ratio by adding the squares of the distances between two adjacent sides, subtracting the squares of the distances between opposite sides, and dividing by the product of a constant and the distances between the two adjacent sides. This cosine ratio is then extracted using an inverse cosine angle mapping process. The calculated angles have several possibilities: if the angle is between 0.0 and 90.0 degrees, it indicates a highly concentrated diffusion line direction, making convergence flow likely; if the angle is greater than 90.0 degrees, it indicates a divergent diffusion line direction, making accumulation unlikely. A numerical sorting algorithm from smallest to largest is executed to rearrange the multiple sets of internal angle values, forming a sequence. Extract the angle value at the head position of the sorted result sequence and determine the extracted angle value as the smallest interior angle feature value.

[0078] S412: Based on the minimum interior angle characteristic value and the spatial diffusion gradient value of the pollution peak, calculate the sinusoidal trigonometric proportional constant of the corresponding diffusion direction, obtain the preset environmental diffusion coefficient of the same region as the spatial diffusion gradient value of the pollution peak, and establish the scalarized basic diffusion flux.

[0079] The smallest interior angle feature value is extracted and combined with the diffusion direction of the corresponding pollution peak spatial diffusion gradient value to retrieve the Taylor series expansion process. The Taylor series expansion process reads the radian value corresponding to the angle and uses alternating addition and subtraction of the finite power terms of the polynomial and the factorial quotient to approximate the trigonometric function value. The smallest interior angle feature value is divided by a constant to perform a halving operation, and the result of the approximation operation is extracted as the sine trigonometric ratio constant value. A preset environmental diffusion coefficient is extracted from meteorological environmental data. The specific setting method of the preset environmental diffusion coefficient is to extract the annual average wind speed dataset records of historical meteorological monitoring in the target area and perform cumulative averaging to obtain the expected mean wind speed. The average surface vegetation coverage in the statistical area is used to determine the surface friction correction coefficient constant. If the coverage is in the range of 50% to 100%, the resistance is high, and the coefficient is set to 0.5 to 0.7; if the coverage is in the range of 0% to 50%, the resistance is low, and the coefficient is set to 0.7 to 0.9. Based on the statistical coverage, a constant coefficient is determined. The expected mean wind speed is multiplied by the surface friction correction coefficient constant to obtain the result, which is then assigned to the preset environmental diffusion coefficient. The preset environmental diffusion coefficient is then multiplied by the spatial diffusion gradient value of the pollution peak, and the resulting product is recorded as the scalarized basic diffusion flux parameter.

[0080] S413: Extract the sine trigonometric proportionality constant associated with the minimum interior angle feature value, calculate the product between the scalarized basic diffusion flux and the sine trigonometric proportionality constant, characterize the state of pollution peak contraction due to the obstruction of narrow topological space, and obtain the directional flow flux value.

[0081] The standardized basic diffusion flux parameter and the sine trigonometric proportionality constant value extracted by Taylor series expansion are extracted. Using the standardized basic diffusion flux parameter and the sine trigonometric proportionality constant value as dual-source data inputs, a multiplication operation is performed between the two data points. The sine trigonometric proportionality constant term represents the flow cross-sectional impedance parameter induced by the reduction of the spatial node angle in the multiplication structure, enabling the multiplication process to output a quantified index and objectively simulate the physical limitation of the flow cross-sectional area in the narrow constriction region corresponding to the local minimum angle at the target site. The multiplication yields a merged result, and the final quantized data output from the numerical multiplication operation is defined and saved as the directional flow flux value.

[0082] The specific steps for obtaining the pollution peak data calibration processing sequence are as follows:

[0083] S511: Collect multiple sets of directional flow flux values ​​around the target site, calculate the sum of all directional flow flux values ​​to quantify the overall diffusion and convergence intensity of the spatial site cluster, and obtain the global convergence dispersion scalar.

[0084] Multiple sets of directional flux values ​​distributed along the diffusion directions of the topological radiation lines at the target site are collected. These directional flux values ​​are continuously accumulated. Carry-over and merging are performed sequentially, and the arithmetic result of the continuous accumulation of all directional flux values ​​demonstrates the spatial comprehensive release and convergence intensity characteristics of fine particulate matter pollutants diffused from the target site to surrounding first-order adjacent sites at the current time point. The calculated final accumulated value is encapsulated in a spatial flux parameter definition header, directly identifying it as a global convergence dispersion scalar parameter.

[0085] S512: Compare the global convergence dispersion scalar with the preset frontal advection critical benchmark. If the current global convergence dispersion scalar is less than the frontal advection critical benchmark, it is determined that the real-time PM2.5 data concentration is a false pollution peak caused by a local fault. If it reaches the frontal advection critical benchmark, it is determined to be a true peak. Establish a peak authenticity identification mark.

[0086] The global convergence dispersion scalar value is extracted, and the frontal advection critical benchmark within a pre-set comparison range is retrieved. The frontal advection critical benchmark value is set based on the global convergence dispersion scalar test data corresponding to confirmed real advection front weather events extracted from historical pollution source tracing data, and the arithmetic mean of the global convergence dispersion scalar test data is calculated. The arithmetic mean is rearranged in ascending order from smallest to largest, and the arithmetic mean of the test data at a specific quantile position is extracted to establish the judgment boundary and obtain the benchmark value. The global convergence dispersion scalar and the frontal advection critical benchmark are subjected to a quantitative inequality judgment operation. There are several possibilities for the judgment result: if the global convergence dispersion scalar is greater than the upper limit of the frontal advection critical benchmark range, it belongs to the strong external meteorological driving situation, triggering the true peak affirmative response operation; if the global convergence dispersion scalar is within a reasonable transition range of the benchmark range, it belongs to the transition fusion situation; if the received scalar value is less than the lower limit of the benchmark, it belongs to the weak external factor and high fault occurrence situation, and a negative Boolean judgment code for false pollution peaks caused by local faults is output. Based on the actual results of the quantitative inequality judgment operation, the corresponding status identifier code number is directly written into the corresponding position of the monitoring data time series to construct the peak authenticity identification identifier.

[0087] S513: Based on the peak authenticity identification identifier, for false pollution peak states, retrieve the upper limit value of the corresponding interval of the dynamic confidence judgment interval of pollution peak, apply the upper limit value of the interval to replace the real-time PM2.5 data concentration to perform abnormal concentration data coverage calibration, and for true pollution peak states, maintain the original real-time PM2.5 data concentration characteristics in a constant retention state to obtain the pollution peak data calibration processing sequence.

[0088] Extract the peak value verification identifier from the real-time fine particulate matter data concentration entry recorded at the current time point. The state verification process involves multiple possible decision branches: if the extracted peak value verification identifier indicates that the current data node is in the true peak confirmation state branch, a suspension command is executed to maintain the constant isolation and locking of the original real-time fine particulate matter data concentration value, ensuring that the current record fully retains the physical truth value obtained from hardware detection and rejecting any tampering or overwriting operations; if the extracted peak value verification identifier indicates that it falls into the false pollution peak fault interference state branch, a correction command is triggered, retrieving the upper limit value of the previously generated pollution peak dynamic confidence discrimination interval, and the overwrite control process executes the command to forcibly overwrite and replace the upper limit value in the original real-time fine particulate matter data concentration record. Extract the original safe concentration data record or the safe data record node entry that has been corrected and overwritten after verification, and, according to the chronological order of occurrence in the time series, perform a head-tail pointer association mapping on each discrete node, concatenating and encapsulating them into a one-dimensional continuous time-series array. The output command is executed, and the final output is a pollution peak data calibration processing sequence with continuous temporal properties and intervention to remove and suppress false spikes.

[0089] A PM2.5 data calibration system for pollution peaks, the system comprising:

[0090] The confidence discrimination module acquires continuous PM2.5 data sequences from the target station, extracts the corresponding fluctuation characteristics to calculate the conditional variance, compares the conditional variance with a preset constant, and constructs a dynamic confidence discrimination interval for pollution peaks.

[0091] The topology construction module constructs the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station when the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence discrimination interval of the pollution peak. It integrates the synchronous concentration information of the second two-dimensional spatial plane coordinates to generate basic topology parameters.

[0092] The gradient calculation module, based on the basic topological parameters, determines the spatial diffusion gradient value of the pollution peak by using the Euclidean distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates.

[0093] The flux calculation module constructs a triangular network based on the basic topology parameters and filters the minimum interior angle values. Based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak, it calculates the directional flow flux value.

[0094] The calibration and judgment module sums multiple sets of directional flow flux values ​​to generate a global convergence dispersion scalar, compares it with the preset frontal advection critical benchmark to determine the authenticity of the pollution peak, and performs replacement or retention on the real-time PM2.5 data concentration to obtain the pollution peak data calibration processing sequence.

[0095] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for calibrating PM2.5 data for pollution peaks, characterized in that, Includes the following steps: S1: Obtain the continuous PM2.5 data sequence of the target station, extract the corresponding fluctuation characteristics from it, calculate the conditional variance, compare the conditional variance with the preset constant, and construct the dynamic confidence discrimination interval of pollution peak; S2: When the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence discrimination interval of the pollution peak, construct the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station, integrate the synchronous concentration information of the second two-dimensional spatial plane coordinates, and generate basic topology parameters. S3: Based on the aforementioned basic topological parameters, determine the spatial diffusion gradient value of the pollution peak by using the Euclidean distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates; S4: Construct a triangular network based on the basic topological parameters and select the minimum interior angle value. Calculate the directional flow flux value based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak. S5: Summing the multiple sets of directional flow flux values ​​to generate a global convergence dispersion scalar, comparing it with the preset frontal advection critical benchmark to determine the authenticity of the pollution peak, and performing replacement or retention on the real-time PM2.5 data concentration to obtain the pollution peak data calibration processing sequence. The specific steps for obtaining the targeted circulation flux value are as follows: S411: Construct a local Deloni triangle structure with the target station as the vertex based on the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates in the basic topology parameters. Scan multiple sets of internal angle values ​​of the triangle structure and sort them from smallest to largest. Extract the smallest interior angle feature value at the beginning of the sorted sequence. S412: Based on the minimum interior angle characteristic value and the pollution peak spatial diffusion gradient value, calculate the sinusoidal trigonometric proportional constant of the corresponding diffusion direction, obtain the preset environmental diffusion coefficient of the same region as the pollution peak spatial diffusion gradient value, and establish the scalarized basic diffusion flux. S413: Extract the sine trigonometric proportionality constant associated with the minimum interior angle feature value, calculate the product between the two parameters of the scalarized basic diffusion flux and the sine trigonometric proportionality constant, characterize the state of pollution peak contraction due to the obstruction of narrow topological space, and obtain the directional flow flux value.

2. The PM2.5 data calibration method for pollution peaks according to claim 1, characterized in that, The pollution peak dynamic confidence discrimination interval includes an upper limit and a lower limit. The basic topological parameters include first two-dimensional spatial plane coordinates, second two-dimensional spatial plane coordinates, and synchronous concentration information. The pollution peak spatial diffusion gradient value is specifically calculated using Euclidean space linear distance. The directional flow flux value is specifically calculated using a sine constant and the pollution peak spatial diffusion gradient value. The pollution peak data calibration processing sequence includes replacement concentration data and retained concentration data.

3. The PM2.5 data calibration method for pollution peaks according to claim 1, characterized in that, The specific steps for obtaining the dynamic confidence interval for the pollution peak are as follows: S111: Set a sliding time window to obtain the continuous PM2.5 data sequence of the target station within the current window period, extract the attached time node parameters from it, call the preset generalized autoregressive conditional heteroscedasticity model to calculate the time-series change state variance term of the continuous PM2.5 data sequence, and use it as the concentration fluctuation characteristic of PM2.5 data. S112: Extract the heteroscedastic component values ​​of the concentration fluctuation feature through the generalized autoregressive conditional heteroscedasticity model, perform numerical conditional fitting on the heteroscedastic component values, and deduce the conditional variance values. S113: Calculate the constant sum and constant difference components of the conditional variance value and the preset initial constant scalar, combine the constant sum and constant difference components to determine the upper and lower limits of the confidence discrimination of the data distribution, and obtain the dynamic confidence discrimination interval of the pollution peak.

4. The PM2.5 data calibration method for pollution peaks according to claim 3, characterized in that, The specific steps for obtaining the basic topology parameters are as follows: S211: Obtain the real-time PM2.5 data concentration of the target station and the dynamic confidence interval of the pollution peak, compare the real-time PM2.5 data concentration with the dynamic confidence interval of the pollution peak to determine the suspected pollution peak triggering state, and obtain the first geographic latitude and longitude coordinates of the target station under the real-time PM2.5 data concentration peak triggering condition, and project and transform them into the first two-dimensional spatial plane coordinates. S212: Use the Deloni triangulation algorithm to establish the network topology architecture of the first two-dimensional spatial plane coordinates, search for first-order adjacent stations with the target station along the network topology architecture, obtain the second two-dimensional spatial plane coordinates of the first-order adjacent stations and the PM2.5 data concentration of the adjacent stations, and obtain the characteristic parameters of the adjacent stations. S213: Extract the first two-dimensional spatial plane coordinates corresponding to the target station, and concatenate the first two-dimensional spatial plane coordinates with the second two-dimensional spatial plane coordinates in the adjacent station feature parameters and the adjacent synchronous PM2.5 data concentration to obtain the basic topology parameters.

5. The PM2.5 data calibration method for pollution peaks according to claim 4, characterized in that, The specific steps for obtaining the spatial diffusion gradient value of the pollution peak are as follows: S311: Read the real-time PM2.5 data concentration of the target station and the synchronous PM2.5 data concentration of the first-order adjacent station from the basic topology parameters, calculate the absolute value of the scalar difference characteristic between the two types of data concentrations, and obtain the absolute concentration deviation value. S312: Obtain the basic topology parameters along with the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates, and calculate the Euclidean distance parameter between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates; S313: Calculate the ratio between the absolute concentration deviation value and the Euclidean distance parameter as the concentration change data under Euclidean distance, associate the concentration change data with the two-dimensional spatial plane formed by the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates, and calculate the pollution peak spatial diffusion gradient value by combining the concentration evolution attribute of the corresponding topological direction in the two-dimensional spatial plane.

6. The PM2.5 data calibration method for pollution peaks according to claim 1, characterized in that, The specific steps for obtaining the pollution peak data calibration processing sequence are as follows: S511: Collect multiple sets of the directional flow flux values ​​around the target site, calculate the sum of all the directional flow flux values ​​to quantify the overall diffusion and convergence intensity of the spatial site cluster, and obtain the global convergence dispersion scalar. S512: Compare the global convergence dispersion scalar with the preset frontal advection critical benchmark. If the current global convergence dispersion scalar is less than the frontal advection critical benchmark, it is determined that the real-time PM2.5 data concentration belongs to a false pollution peak caused by a local fault. If it reaches the frontal advection critical benchmark, it is determined to be a real peak. Establish a peak authenticity identification mark. S513: Based on the peak authenticity identification identifier, for false pollution peak states, retrieve the upper limit value of the corresponding interval of the dynamic confidence judgment interval of pollution peak, apply the upper limit value of the interval to replace the real-time PM2.5 data concentration to perform abnormal concentration data coverage calibration, and for true pollution peak states, maintain the original real-time PM2.5 data concentration characteristics in a constant retention state to obtain the pollution peak data calibration processing sequence.

7. The PM2.5 data calibration method for pollution peaks according to claim 3, characterized in that, The time-series state variance term of the continuous PM2.5 data series is calculated by calling the preset generalized autoregressive conditional heteroscedasticity model. Specifically: The generalized autoregressive conditional heteroscedasticity model is set by pre-configured autoregressive order parameters and moving average order parameters; Input continuous PM2.5 data series into a generalized autoregressive conditional heteroscedasticity model with set parameters; By combining the autoregressive order parameter and the moving average order parameter, the variance term of the time series change state is extracted from the continuous PM2.5 data series.

8. A PM2.5 data calibration system for pollution peaks, characterized in that, The PM2.5 data calibration method for pollution peaks according to any one of claims 1-7, the system comprising: The confidence discrimination module acquires continuous PM2.5 data sequences from the target station, extracts the corresponding fluctuation characteristics to calculate the conditional variance, compares the conditional variance with a preset constant, and constructs a dynamic confidence discrimination interval for pollution peaks. The topology construction module constructs the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates of the target station when the real-time PM2.5 data concentration of the target station exceeds the upper limit of the dynamic confidence discrimination interval of the pollution peak. It integrates the synchronous concentration information of the second two-dimensional spatial plane coordinates to generate basic topology parameters. The gradient calculation module, based on the basic topological parameters, determines the spatial diffusion gradient value of the pollution peak by using the Euclidean straight-line distance between the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates. The flux calculation module constructs a triangular network based on the basic topology parameters and filters the minimum interior angle values. Based on the sine constant of the minimum interior angle value and the spatial diffusion gradient value of the pollution peak, it calculates the directional flow flux value. The calibration judgment module sums up multiple sets of directional flow flux values ​​to generate a global convergence dispersion scalar, compares it with a preset frontal advection critical benchmark to determine the authenticity of the pollution peak, and performs replacement or retention on the real-time PM2.5 data concentration to obtain a pollution peak data calibration processing sequence. The specific steps for obtaining the targeted circulation flux value are as follows: Based on the first two-dimensional spatial plane coordinates and the second two-dimensional spatial plane coordinates in the basic topology parameters, a local Deloni triangle structure with the target station as the vertex is constructed. Multiple sets of internal angle values ​​of the triangle structure are scanned and sorted from smallest to largest. The smallest interior angle feature value at the beginning of the sorted sequence is extracted. Based on the minimum interior angle feature value and the pollution peak spatial diffusion gradient value, calculate the sinusoidal trigonometric proportionality constant of the corresponding diffusion direction, obtain the preset environmental diffusion coefficient of the same region as the pollution peak spatial diffusion gradient value, and establish the scalarized basic diffusion flux. Extract the sine trigonometric proportionality constant associated with the minimum interior angle feature value, calculate the product between the scalarized basic diffusion flux and the sine trigonometric proportionality constant, characterize the state of pollution peak contraction due to the obstruction of narrow topological space, and obtain the directional flow flux value.

Images