Configuration method and device of aerosol data prediction model and processing equipment

By constructing the matching relationship between the air pollution monitoring station and the aerosol monitoring station, and configuring an aerosol data prediction model, the problem of uneven distribution of aerosol monitoring stations is solved, better space-time coverage and data quality are achieved, and monitoring is reduced. Cost improves the efficiency of air quality management.

CN118918973APending Publication Date: 2024-11-08CHINA UNIV OF GEOSCIENCES (WUHAN)
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411159000.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing aerosol monitoring stations have limited coverage in space-time distribution, resulting in uneven aerosol monitoring effects and making it difficult to effectively monitor aerosol data.

Method used

By constructing a matching relationship between the atmospheric pollution monitoring station and the aerosol monitoring station, aerosol data prediction model is configured, and aerosol data is predicted using atmospheric pollution monitoring data to improve the coverage of spatiotemporal distribution and data quality.

Benefits of technology

It significantly improves the spatial and temporal distribution coverage and data quality of aerosol data, reduces the construction and maintenance costs of new monitoring sites, improves resource utilization and economic benefits, meets the needs of high-quality aerosol monitoring, and contributes to the atmosphere quality. Monitor and manage.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918973A_ABST
    Figure CN118918973A_ABST
Patent Text Reader

Abstract

The invention provides a configuration method and device of an aerosol data prediction model, and processing equipment, and is used for considering the construction of a matching relation between monitoring data of an atmospheric pollution monitoring station and an aerosol monitoring station under the condition that aerosol observation stations are distributed very unevenly. The aerosol data prediction model configured in this way can input air pollution monitoring data which are convenient to obtain and wide in station distribution under the actual condition to achieve the purpose of predicting aerosol data of all positions, the space-time distribution coverage and the data quality are remarkably improved, the construction and maintenance cost of newly-added monitoring stations is reduced, and the method is suitable for popularization and application. The resource utilization rate and the economic benefit are improved, the high-quality monitoring requirement on the aerosol under the actual condition can be met, and monitoring and management on the atmosphere quality are facilitated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of air quality monitoring, and specifically to a configuration method, device and processing equipment for an aerosol data prediction model. Background Art

[0002] In recent years, as public health and climate change issues have become increasingly prominent, research on air pollution and aerosols has continued to advance. Many studies have shown that aerosols are a key factor affecting the environment and human health.

[0003] In order to monitor aerosol data, many aerosol monitoring stations have been established. Aerosol monitoring stations study the distribution, composition and optical characteristics of aerosol particles. The monitoring data sources mainly include AERONET, satellite remote sensing (such as MODIS and CALIPSO) and ground-based lidar.

[0004] However, although existing aerosol monitoring stations can provide a large amount of valuable monitoring data, they still face the problem of limited coverage in terms of temporal and spatial distribution, resulting in obvious imbalance in the coverage of aerosol monitoring effects. Summary of the invention

[0005] The present application provides a configuration method, device and processing equipment of an aerosol data prediction model, which is used to consider building a matching relationship between air pollution monitoring stations and the data monitored by aerosol monitoring stations when the distribution of aerosol observation stations is very uneven. The aerosol data prediction model configured in this way can input air pollution monitoring data that is easy to obtain and has widely distributed stations in actual situations to achieve the purpose of predicting aerosol data at various locations. It has significantly improved spatiotemporal distribution coverage and data quality, reduces the construction and maintenance costs of new monitoring stations, improves resource utilization and economic benefits, can meet the high-quality monitoring needs for aerosols in actual situations, and thus helps to monitor and manage air quality.

[0006] In a first aspect, the present application provides a method for configuring an aerosol data prediction model, the method comprising:

[0007] Obtain sample air pollution data monitored by different air pollution monitoring stations within a selected time range;

[0008] Obtain sample aerosol data monitored by different aerosol monitoring stations within a selected time range;

[0009] Matching processing is performed on two types of sites, namely, different air pollution monitoring sites and different aerosol monitoring sites, to obtain site matching results, and data of corresponding sites are extracted from sample air pollution data and sample aerosol data according to the site matching results to obtain target sample data as training samples;

[0010] Based on the target sample data, an aerosol data prediction model is trained, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

[0011] In a second aspect, the present application provides a configuration device for an aerosol data prediction model, the device comprising:

[0012] A first acquisition unit is used to acquire sample air pollution data monitored by different air pollution monitoring stations within a selected time range;

[0013] A second acquisition unit is used to acquire sample aerosol data monitored by different aerosol monitoring stations within a selected time range;

[0014] A matching unit is used to match two types of sites, namely, different air pollution monitoring sites and different aerosol monitoring sites, to obtain site matching results, and to extract data of corresponding sites from sample air pollution data and sample aerosol data according to the site matching results, to obtain target sample data as training samples;

[0015] The training unit is used to train an aerosol data prediction model based on target sample data, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

[0016] In a third aspect, the present application provides a processing device, including a processor and a memory, wherein a computer program is stored in the memory, and when the processor calls the computer program in the memory, the method provided in the first aspect of the present application or any possible implementation method of the first aspect of the present application is executed.

[0017] In a fourth aspect, the present application provides a computer-readable storage medium, which stores multiple instructions, and the instructions are suitable for a processor to load to execute the method provided by the first aspect of the present application or any possible implementation of the first aspect of the present application.

[0018] From the above content, it can be concluded that the present application has the following beneficial effects:

[0019] With respect to the monitoring target of aerosols, this application considers constructing a matching relationship between air pollution monitoring stations and the data monitored by aerosol monitoring stations when the distribution of aerosol observation stations is very uneven. The aerosol data prediction model configured in this way can input air pollution monitoring data that is easy to obtain in actual situations and has widely distributed stations, so as to achieve the purpose of predicting aerosol data at various locations. It has significantly improved spatiotemporal distribution coverage and data quality, reduced the construction and maintenance costs of new monitoring stations, improved resource utilization and economic benefits, and can meet the actual needs for high-quality monitoring of aerosols, thereby facilitating the monitoring and management of air quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A schematic diagram of a flow chart of a configuration method of an aerosol data prediction model of the present application;

[0022] Figure 2 A schematic diagram of a configuration device for an aerosol data prediction model of the present application;

[0023] Figure 3 A schematic diagram of the structure of the processing equipment of this application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0025] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of steps in this application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved.

[0026] The division of modules in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between modules can be electrical or other similar forms, which are not limited in this application. In addition, the modules or submodules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed in multiple circuit modules, and some or all of the modules may be selected according to actual needs to achieve the purpose of the present application.

[0027] Before introducing the configuration method of the aerosol data prediction model provided by the present application, the background content involved in the present application is first introduced.

[0028] The configuration method, device and computer-readable storage medium of the aerosol data prediction model provided in the present application can be applied to processing equipment to consider building a matching relationship between air pollution monitoring stations and the data monitored by aerosol monitoring stations when the distribution of aerosol observation stations is very uneven. The aerosol data prediction model configured in this way can input air pollution monitoring data that is easy to obtain and has widely distributed stations under actual circumstances to achieve the purpose of predicting aerosol data at various locations. It has significantly improved spatiotemporal distribution coverage and data quality, reduces the construction and maintenance costs of new monitoring stations, improves resource utilization and economic benefits, can meet the actual high-quality monitoring needs for aerosols, and thus helps to monitor and manage air quality.

[0029] The configuration method of the aerosol data prediction model mentioned in this application can be executed by a configuration device of the aerosol data prediction model, or a server, a physical host or a user equipment (UE) and other different types of processing equipment that integrate the configuration device of the aerosol data prediction model. Among them, the configuration device of the aerosol data prediction model can be implemented in hardware or software, and the UE can be a terminal device such as a smart phone, a tablet computer, a laptop computer, a desktop computer or a personal digital assistant (PDA), and the processing device can be set in a device cluster.

[0030] It can be understood that for the processing equipment that executes the configuration method of the aerosol data prediction model provided in the present application or is equipped with the corresponding application service of the configuration method of the aerosol data prediction model provided in the present application, considering that the focus of the present application scheme is on the data expanded on the basis of the existing monitoring data without involving the specific collection operation of the monitoring data, it is usually only necessary to meet the required data processing capabilities. In this case, the processing equipment can be configured as any type of specific equipment and equipment deployment form under actual circumstances.

[0031] Of course, if specific collection operations for monitoring data are required in actual situations, the processing equipment needs to be configured with corresponding data collection capabilities. In this regard, the corresponding monitoring sites can be included in the scope of the processing equipment, or the processing equipment itself includes the monitoring system of each monitoring site.

[0032] Next, we will introduce the configuration method of the aerosol data prediction model provided by this application.

[0033] First, see Figure 1 , Figure 1 A flow chart of a configuration method of an aerosol data prediction model of the present application is shown. The configuration method of an aerosol data prediction model provided by the present application may specifically include the following steps S101 to S104:

[0034] Step S101, obtaining sample air pollution data monitored by different air pollution monitoring stations within a selected time range;

[0035] It can be understood that the construction of the model requires the configuration of training samples. As can be seen here, in the process of configuring training samples, sample air pollution data monitored by different air pollution monitoring stations may be involved. In addition, in the subsequent step S102, two aspects of training samples, namely sample aerosol data, may also be involved.

[0036] Among them, for the sample atmospheric pollution data, in order to facilitate alignment with the subsequent sample aerosol data, the same time range can be used in the acquisition process. In actual situations, this time range can be adjusted according to application requirements and is recorded as the selected time range.

[0037] As an example, the selected time range may specifically be 2014-2023.

[0038] It is worth adding that the air pollution monitoring stations themselves are widely distributed. Therefore, under the design of the present application, the air pollution monitoring data obtained by the widely distributed air pollution monitoring stations can be used to invert the aerosol data that cannot be monitored by the aerosol monitoring stations in various locations with significantly uneven distribution.

[0039] Specifically, as an exemplary embodiment here, the sample air pollution data may specifically include PM2.5, PM10, SO2, NO2, O3 and CO types of data.

[0040] Taking the sample air pollution data obtained by the National Urban Air Quality Real-time Release Platform as an example, there are:

[0041] Table 1 - Sample air pollution data examples

[0042]

[0043] Step S102, obtaining sample aerosol data monitored by different aerosol monitoring stations within a selected time range;

[0044] It can be understood that aerosol monitoring stations themselves, compared with air pollution monitoring stations, are limited in number and have obvious uneven distribution. Therefore, the aerosol data monitored by them have the problems of uneven distribution and poor quantity and quality that this application aims to overcome.

[0045] As an example, the aerosol monitoring site involved in the present application may specifically be an AERONET site.

[0046] In addition, it can be understood that there is no selection in the execution order between the previous step S101 and the step S102 here. Step S101 can be executed first, and then step S102, or step S102 can be executed first and then step S101, or step S101 and step S102 can be executed at the same time, and it can be adjusted according to actual needs.

[0047] As an exemplary embodiment herein, the sample aerosol data may specifically include several types of data, including Aerosol Optical Depth (AOD), Angstrom Exponent (AE), Single Scattering Albedo (SSA) and Fine Mode Fraction (FMF).

[0048] Taking the sample aerosol data obtained by the AERONET station as an example, we can get:

[0049] Table 2 - Sample aerosol data examples

[0050]

[0051] It can be understood that the data content of the sample aerosol data obtained here is to be used as the true value (the result that the model is to predict in theory) in the model training process.

[0052] The acquisition and processing of the monitoring data in the above two aspects are easy to understand, and are the acquisition and processing of ready-made data, which can be either manual input or autonomous extraction.

[0053] Step S103, matching processing is performed on two types of sites, namely, different air pollution monitoring sites and different aerosol monitoring sites, to obtain site matching results, and data of corresponding sites are extracted from sample air pollution data and sample aerosol data according to the site matching results to obtain target sample data as training samples;

[0054] It is understandable that after obtaining the sample atmospheric pollution data and sample aerosol data, this application does not directly carry out model training, but considers the better development of model training to promote better model training results. It is necessary to perform corresponding preprocessing operations to further enhance data quality.

[0055] As for conventional data preprocessing operations, such as removing abnormal data, unifying data formats, etc., considering that they belong to the scope of the prior art, these conventional data preprocessing operations that can be applied will not be described in detail here.

[0056] It can be seen here that the data preprocessing operation specially designed for this application involves matching processing of different types of sites, and on this basis, obtains simplified and more effective monitoring data as training samples for specific model training.

[0057] Among them, it should be noted that the matching processing here specifically refers to the matching processing between air pollution monitoring stations and aerosol monitoring stations, rather than the matching processing between stations of the same type. The specific station of the monitoring data (data source) is usually directly marked in the monitoring data. Of course, this application can also mark the corresponding specific station (data source) in the monitoring data after obtaining the monitoring data.

[0058] It is understandable that in the site matching process, not all aerosol monitoring sites have corresponding matching sites, and the matching results may be empty. Therefore, by filtering these empty value data, it can be ensured that only data with valid matching results are retained, thereby achieving the purpose of obtaining simplified and more effective monitoring data.

[0059] In specific operations, valid site matching results can be matched / retrieved on a year-by-year basis to extract qualified data and obtain target sample data that can be used as training samples for model training.

[0060] Step S104: training an aerosol data prediction model based on the target sample data, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

[0061] After obtaining the target sample data, specific model training can be carried out to train an aerosol data prediction model that can be put into practical use. For model training, the following processing model training logic can be used:

[0062] In each round of model training, a set of atmospheric pollution data is input into the model so that the model predicts the corresponding aerosol data to achieve forward propagation. Then, based on the aerosol data predicted by the model and the aerosol data (true value) corresponding to the current atmospheric pollution data in the target sample data, the loss function is calculated, and the model parameters are optimized according to the loss function to achieve reverse propagation. In this way, when different types of model training conditions such as training time, number of trainings, and prediction accuracy are met, the model training can be completed. The model at this time is recorded as an aerosol data prediction model that can be put into actual use.

[0063] Among them, the machine learning model architecture, training architecture, loss function and other model configuration contents involved in the aerosol data prediction model can be understood as directly using existing technologies, or making further optimizations and improvements on the basis of existing technologies, or using self-developed and novel technologies. All of these are possible in actual situations.

[0064] In this way, the trained aerosol data prediction model can not only predict the aerosol data of the locations of air pollution monitoring stations and aerosol monitoring stations, but also carry out the prediction and processing of aerosol data at different locations based on the overall range of the locations of the stations involved in the sample data. It has the prediction characteristics of high temporal and spatial resolution, and does not have the sparse data points existing in the prior art. In more popular terms, compared with the point prediction effect achieved by the original aerosol monitoring station, the aerosol data prediction model of the present application achieves a surface prediction effect, so the aerosol data prediction results outputted have better temporal and spatial distribution coverage and data quality.

[0065] from Figure 1 It can be seen from the illustrated embodiments that, with respect to the monitoring target of aerosols, the present application considers constructing a matching relationship between air pollution monitoring sites and the data monitored by aerosol monitoring sites when faced with a very uneven distribution of aerosol observation sites. The aerosol data prediction model configured in this way can input air pollution monitoring data that is easy to obtain in actual situations and has widely distributed sites, so as to achieve the purpose of predicting aerosol data at various locations. It has significantly improved spatiotemporal distribution coverage and data quality, reduces the construction and maintenance costs of new monitoring sites, improves resource utilization and economic benefits, can meet the actual needs for high-quality monitoring of aerosols, and thus contributes to the monitoring and management of air quality.

[0066] Continue to the above Figure 1 Each step of the illustrated embodiment and possible implementation methods thereof in practical applications are described in detail.

[0067] Specifically, the site matching process involved in step S103 in the above content can be mainly implemented by site distance or mainly based on site distance.

[0068] Focusing on the site matching process implemented by the site distance, the matching process of the two types of sites in the previous step S103 is performed to obtain the site matching result. As an exemplary embodiment, it may specifically include:

[0069] 1) Configure the site numbers of different air pollution monitoring sites and different aerosol monitoring sites;

[0070] It can be understood that the configuration of site numbers in the sample can lay a good data foundation for more convenient and better site matching processing and subsequent data processing.

[0071] The site number involved in this application may involve a digital identifier, an alphabetic identifier, a Chinese identifier, or a combination of different types of identifiers, which can be configured according to actual conditions.

[0072] 2) Based on the site number, the coordinate data corresponding to different sites are merged;

[0073] It can be understood that in order to calculate the distance between sites, the coordinates of the sites are required (specifically, they can be expressed in the form of coordinates such as longitude and latitude). In this regard, the present application can merge the coordinate data of different sites based on the site number, so that the corresponding distance calculation between sites can be carried out in a highly integrated data file.

[0074] The coordinate data of the site may be recorded in the original monitoring data, or may be processed by real-time coordinate query / determination. These are all possible, and different processing methods may be configured in actual situations.

[0075] For example, the coordinate data of different sites can be recorded in a list, and other contents such as the corresponding site numbers can also be recorded.

[0076] As an example, one could have:

[0077] Table 3 - Coordinate data examples

[0078] Monitoring site number Monitoring point name City longitude latitude 1001A Wanshou West Palace Beijing 116.366 39.8673 1002A Dingling Beijing 116.17 40.2865 1003A Dongsi Beijing 116.434 39.9522 1004A Temple of Heaven Beijing 116.434 39.8745 1005A Agricultural Exhibition Hall Beijing 116.473 39.9716

[0079] 3) In the coordinate data, the distance d between the two types of sites is calculated based on the distance quantization formula. If the distance d ≤ 50 km, it is considered a match. Otherwise, if the distance d > 50 km, it is considered a mismatch. All matching sites form a site matching result. The distance quantization formula is:

[0080]

[0081] Where R is the radius of the earth, which can be taken as a constant of 6371 km, φi is the latitude of station i in radians, i = 1 or 2, λi is the longitude of station i in radians, Δφ = φ 2 -φ 1 , Δλ=λ 2 -λ 1 .

[0082] It can be seen that the present application provides a specific distance quantification formula for the distance calculation involved in the solution, so that the required distance between sites can be quantified conveniently and accurately in the application scenario of the present application.

[0083] The matching results can be stored in the form of lists or the like.

[0084] As an example, one could have:

[0085] Table 4 - Site matching result example 1

[0086]

[0087] After obtaining the site matching results (such as the records in Table 4 above), the monitoring data of the sites not involved can be ignored / discarded from the two types of monitoring data, and the monitoring data of the corresponding sites can be extracted to obtain the target sample data as training samples.

[0088] In this process, further data preprocessing operations may be involved to obtain higher quality target sample data.

[0089] Specifically, as an exemplary embodiment, in the previous step S103, the data of the corresponding site is extracted from the sample air pollution data and the sample aerosol data according to the site matching result to obtain the target sample data as the training sample, which may specifically include:

[0090] 1) extracting first data of the corresponding site from the sample air pollution data and the sample aerosol data according to the site matching result, wherein the first data includes first sub-data of the corresponding site extracted from the sample aerosol data according to the site matching result and second sub-data of the corresponding site extracted from the sample air pollution data according to the site matching result;

[0091] It can be understood that the first data extracted from the original monitoring data according to the matching results are recorded as the first sub-data corresponding to the aerosol and the second sub-data corresponding to the air pollution.

[0092] 2) In the first data, the first sub-data is adjusted to the date format used by the second sub-data, and the average is calculated based on the data collection time point corresponding to the date format, so that the format corresponds to the second sub-data;

[0093] It can be understood that the present application notes that the air pollution data provided by the air pollution monitoring stations are recorded hourly (0 to 23 o'clock) and do not contain precise hour, minute and second information, that is, they have larger time intervals (corresponding to longer monitoring periods). Therefore, it is necessary to align the aerosol data provided by the aerosol monitoring stations with smaller time intervals.

[0094] In this regard, the present application can adjust the first sub-data corresponding to aerosol to the date format used by the second sub-data corresponding to air pollution.

[0095] For example, change the hour, minute, and second format to hour, extract the hour information from the time, and adjust the date format to the standard YYYYMMDD format.

[0096] In addition, the monitoring data can also be unified into Beijing time.

[0097] For the monitoring data grouped by hours, continue to calculate the average value to ensure that the hourly aerosol data can be aligned with the second sub-data corresponding to the air pollution in terms of particle size.

[0098] 3) Sort the first data by time to obtain second data as target sample data.

[0099] After completing the alignment of the first sub-data corresponding to aerosols and the second sub-data corresponding to air pollution in the first data to further ensure the consistency and accuracy of the data, you can continue to sort and save them according to time (data collection time point or monitoring time point) to obtain the second data (target sample data) that has completed the particle size alignment and can be used more conveniently.

[0100] At the same time, for the target sample data, this application takes into account the application needs of more efficient data retrieval and analysis in the subsequent model training phase. This application can also store it separately by site, thereby improving the overall processing efficiency.

[0101] In this regard, as an exemplary embodiment, before the aerosol data prediction model is trained based on the target sample data in the previous step S104, the configuration method of the aerosol data prediction model of the present application may further include:

[0102] The data involved in the target sample data is stored separately by site unit.

[0103] In addition, data from the same site can be further stored separately according to time units such as years, quarters, months or hours to further improve the convenience of data retrieval / analysis.

[0104] As an example, one could have:

[0105] Table 5 - Site matching result example 2

[0106]

[0107] Furthermore, the present application also considers the introduction of meteorological factors for aerosol prediction, and uses meteorological factors to assist in the prediction process, so as to achieve a more accurate prediction effect.

[0108] In this regard, as another exemplary embodiment, referring to the above processing of sample aerosol data, the configuration method of the aerosol data prediction model of the present application may further include:

[0109] 1) Obtain sample meteorological data monitored by different ground meteorological monitoring stations within a selected time range;

[0110] It can be understood that how to obtain sample meteorological data can refer to how to obtain sample aerosol data, which will not be repeated here.

[0111] As an example, the sample meteorological data may be meteorological data provided by the ERA5 data product.

[0112] In addition, with regard to the specific content of the sample meteorological data, as an exemplary embodiment, the sample meteorological data may specifically include several types of data, including 2m dew point temperature (2m_dewpoint_temperature), 2m temperature (2m_temperature), 10m U-direction wind speed component (10m_u_component_of_wind), 10m V-direction wind speed component (10m_v_component_of_wind) and boundary layer height (boundary_layer_height).

[0113] Taking ERA5 data products as an example, there are:

[0114] Table 6 - Sample meteorological data examples

[0115]

[0116] 2) Matching is performed on two types of sites, namely, different air pollution monitoring sites and different ground meteorological monitoring sites, to obtain new site matching results. Based on the new site matching results, new data of the corresponding sites are extracted from the sample air pollution data and the sample period data, and then added to the target sample data.

[0117] It can be understood that for the newly introduced meteorological factors, their sample meteorological data and sample aerosol data are the same, involving site matching processing between sample atmospheric pollution data, and on this basis, simplified and more effective monitoring data are extracted from the two to serve as training samples for specific model training. The specific processing in this process can refer to the previous processing of sample aerosol data, such as matching processing based on distance d based on coordinate data configured by site number, alignment of date format, and separate storage by site unit, etc., which will not be elaborated here.

[0118] As an example, one could have:

[0119] Table 7-Site matching result example 3

[0120]

[0121] For the matching results mentioned above, in specific operations, you can also convert the xlsx file into a csv file for the model to process.

[0122] In this case, the aerosol data prediction model is specifically used to predict the corresponding aerosol data based on the atmospheric pollution data and meteorological data input into the model.

[0123] In the subsequent model training process, it can be understood that the present application may also involve further detailed configuration work so that the model training can be carried out conveniently and efficiently.

[0124] For example, among the neural network model schemes that can be used in the aerosol data prediction model, the LightGBM model can be used specifically. The LightGBM model is a gradient boosting framework optimized on XGBoost. It uses decision trees and is known for its ability to handle large-scale data sets and its significant advantages in accelerating model training. It introduces gradient-based one-side sampling (GOSS), which can significantly improve training time and efficiency.

[0125] During the training process, the application can also be configured to train four types of aerosol data: aerosol optical thickness (AOD), Angstrom index (AE), single scattering albedo (SSA) and fine mode fraction (FMF). The model configuration has the following contents:

[0126] 1) Data import and preprocessing:

[0127] The pandas library is used for data reading and preliminary processing. A data set including features such as pollutant concentration, date, and time is imported from the specified path. The features and target variables (aerosol data) in the data are separated and normalized using MinMaxScaler. The main purpose of this step is to eliminate the dimensional differences between different features and improve the convergence speed and prediction accuracy of the model.

[0128] 2) K-fold cross-validation: K-fold cross-validation is used to evaluate the performance of the model on different data set partitions. Its basic principle is to divide the data set into K equal-sized subsets (folds), and then perform K training and validation, each time selecting a different subset as the validation set, and the remaining K-1 subsets as the training set. Set 10-fold cross-validation (KFold). The purpose of selecting 10-fold cross-validation is to obtain lower variance and more accurate results when evaluating model performance. By splitting the data set multiple times and performing training and validation, the performance of the model on different data sets can be comprehensively evaluated, providing more reliable model evaluation results.

[0129] 3) Dataset division: Use train_test_split to divide the normalized dataset into a training set and a test set, where the training set accounts for 90% and the test set accounts for 10%. In addition, in each iteration of the 10-fold cross-validation, 90% of the training set is used for training and 10% is used for validation.

[0130] 4) Hyperparameter optimization: Hyperopt is applied to the model for hyperparameter optimization, and the parameter search space is defined, including max_depth, learning_rate, bagging_fraction, and num_leaves. Among them, max_depth controls the maximum number of layers of the tree to avoid overfitting, and the parameter is set to hp.randint("max_depth", 222); learning_rate is the learning rate that controls the update step size of each iteration, and the parameter is set to hp.uniform('learning_rate', 0.1, 0.9); bagging_fraction is used to control the data ratio of the training model at each iteration, and the parameter is set to hp.uniform("bagging_fraction", 0.8, 0.9); num_leaves controls the complexity of each tree, and the parameter is set to hp.randint("num_leaves", 111).

[0131] 5) Model training and prediction: According to the results of hyperparameter optimization, the LightGBM model is trained using the optimal parameter combination. During the training process, the model sets an early stopping mechanism to monitor the performance changes on the validation set for 20 rounds to ensure that the model stops training at the optimal number of iterations.

[0132] 6) Model saving and result output: Save the trained model to the specified path for subsequent use and reproduction of research results. At the same time, denormalize the prediction results, restore them to the dimension of the original data and save them for subsequent analysis and application.

[0133] In order to verify the training accuracy of the model, this application specifically selected R 2 A total of seven evaluation indicators, including MAE, RMSE, MB, GCOS, EE_SSA1 and EE_SSA2, are used to evaluate the matching conversion accuracy of atmospheric pollution data and various aerosol data, that is, the prediction accuracy. The results show that the aerosol data prediction model trained by the present application scheme has high prediction accuracy and can well meet the high-quality prediction requirements for aerosol data at different locations based on atmospheric pollution data.

[0134] The above is an introduction to the configuration method of the aerosol data prediction model provided in the present application. In order to facilitate better implementation of the configuration method of the aerosol data prediction model provided in the present application, the present application also provides a configuration device of an aerosol data prediction model from the perspective of functional modules.

[0135] See also Figure 2 , Figure 2 This is a structural diagram of a configuration device for an aerosol data prediction model of the present application. In the present application, the configuration device 200 for an aerosol data prediction model may specifically include the following structure:

[0136] The first acquisition unit 201 is used to acquire sample air pollution data monitored by different air pollution monitoring stations within a selected time range;

[0137] The second acquisition unit 202 is used to acquire sample aerosol data monitored by different aerosol monitoring stations within a selected time range;

[0138] The matching unit 203 is used to match the two types of sites, namely, different air pollution monitoring sites and different aerosol monitoring sites, to obtain site matching results, and to extract data of corresponding sites from the sample air pollution data and the sample aerosol data according to the site matching results, to obtain target sample data as training samples;

[0139] The training unit 204 is used to train an aerosol data prediction model based on target sample data, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

[0140] As an exemplary embodiment, the matching unit 203 is specifically configured to:

[0141] Configure site numbers for different air pollution monitoring sites and different aerosol monitoring sites;

[0142] Based on the site number, the coordinate data corresponding to different sites are merged;

[0143] In the coordinate data, the distance d between the two types of sites is calculated based on the distance quantification formula. If the distance d ≤ 50 km, it is considered a match, and all matching sites form the site matching result. The distance quantification formula is:

[0144]

[0145] Where R represents the radius of the earth, φ i represents the latitude of station i in radians, i = 1 or 2, with λ i represents the longitude of station i in radians, Δφ=φ 2 ―φ 1, Δλ=λ 2 -λ 1 .

[0146] As another exemplary embodiment, the matching unit 203 is specifically configured to:

[0147] Extracting first data corresponding to the site from the sample air pollution data and the sample aerosol data according to the site matching result, wherein the first data includes first sub-data corresponding to the site extracted from the sample aerosol data according to the site matching result and second sub-data corresponding to the site extracted from the sample air pollution data according to the site matching result;

[0148] In the first data, the first sub-data is adjusted to the date format used by the second sub-data, and the average is calculated based on the data collection time point corresponding to the date format, so that the format corresponds to the second sub-data;

[0149] The first data is sorted according to time to obtain second data as target sample data.

[0150] As another exemplary embodiment, the matching unit 203 is further configured to:

[0151] The data involved in the target sample data is stored separately by site unit.

[0152] As another exemplary embodiment, the sample air pollution data specifically includes data of PM2.5, PM10, SO2, NO2, O3 and CO;

[0153] The sample aerosol data specifically include aerosol optical depth AOD, Angstrom index AE, single scattering albedo SSA and fine mode fraction FMF.

[0154] As another exemplary embodiment, the device further includes:

[0155] The third acquisition unit 205 is used to acquire sample meteorological data monitored by different ground meteorological monitoring stations within a selected time range;

[0156] The matching unit 203 is also used to match the two types of sites, namely, different air pollution monitoring sites and different ground meteorological monitoring sites, to obtain new site matching results, and extract new data of the corresponding sites from the sample air pollution data and the sample period data according to the new site matching results, and then add them to the target sample data;

[0157] The aerosol data prediction model is specifically used to predict corresponding aerosol data based on the atmospheric pollution data and meteorological data input into the model.

[0158] As another exemplary embodiment, the sample meteorological data specifically includes data of 2-meter dew point temperature, 2-meter temperature, 10-meter U-direction wind speed component, 10-meter V-direction wind speed component and boundary layer height.

[0159] This application also provides a processing device from the perspective of hardware structure, see Figure 3 , Figure 3 301, a memory 302, and an input / output device 303. The processor 301 is used to execute the computer program stored in the memory 302 to implement the following Figure 1 The steps of the configuration method of the aerosol data prediction model in the corresponding embodiment; or, the processor 301 is used to execute the computer program stored in the memory 302 to implement the following Figure 2 Corresponding to the functions of each unit in the embodiment, the memory 302 is used to store the processor 301 executing the above Figure 1 The computer program required for the configuration method of the aerosol data prediction model in the corresponding embodiment.

[0160] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 302, and executed by the processor 301 to complete the present application. One or more modules / units may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0161] The processing device may include, but is not limited to, a processor 301, a memory 302, and an input / output device 303. Those skilled in the art will appreciate that the illustration is merely an example of a processing device and does not constitute a limitation on the processing device, and may include more or fewer components than shown in the illustration, or a combination of certain components, or different components. For example, the processing device may also include a network access device, a bus, etc., and the processor 301, the memory 302, the input / output device 303, etc. are connected via a bus.

[0162] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the processing device, and uses various interfaces and lines to connect various parts of the entire device.

[0163] The memory 302 can be used to store computer programs and / or modules. The processor 301 implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 302 and calling the data stored in the memory 302. The memory 302 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the processing device, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0164] When the processor 301 is used to execute the computer program stored in the memory 302, the following functions can be implemented:

[0165] Obtain sample air pollution data monitored by different air pollution monitoring stations within a selected time range;

[0166] Obtain sample aerosol data monitored by different aerosol monitoring stations within a selected time range;

[0167] Matching processing is performed on two types of sites, namely, different air pollution monitoring sites and different aerosol monitoring sites, to obtain site matching results, and data of corresponding sites are extracted from sample air pollution data and sample aerosol data according to the site matching results to obtain target sample data as training samples;

[0168] Based on the target sample data, an aerosol data prediction model is trained, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

[0169] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the configuration device, processing equipment and specific working process of the above-described aerosol data prediction model can refer to the following. Figure 1 The description of the configuration method of the aerosol data prediction model in the corresponding embodiment will not be repeated here.

[0170] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0171] To this end, the present application provides a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the present application as follows: Figure 1 The steps of the configuration method of the aerosol data prediction model in the corresponding embodiment, the specific operation can refer to the following Figure 1 The description of the configuration method of the aerosol data prediction model in the corresponding embodiment will not be repeated here.

[0172] The computer-readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0173] Due to the instructions stored in the computer-readable storage medium, the present application can be executed. Figure 1 The steps of the configuration method of the aerosol data prediction model in the corresponding embodiment, therefore, the present application can be implemented as follows Figure 1 The beneficial effects that can be achieved by the configuration method of the aerosol data prediction model in the corresponding embodiment are detailed in the previous description and will not be repeated here.

[0174] The configuration method, device, processing equipment and computer-readable storage medium of the aerosol data prediction model provided by the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for configuring an aerosol data prediction model, characterized in that: The method comprises: Obtain sample air pollution data monitored by different air pollution monitoring stations within a selected time range; Acquire sample aerosol data monitored by different aerosol monitoring stations within the selected time range; Performing matching processing on the two types of sites, namely, the different air pollution monitoring sites and the different aerosol monitoring sites, to obtain site matching results, and extracting data of corresponding sites from the sample air pollution data and the sample aerosol data according to the site matching results to obtain target sample data as training samples; Based on the target sample data, an aerosol data prediction model is trained, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

2. The method according to claim 1, characterized in that The matching process of the two types of sites is performed to obtain a site matching result, including: Configuring site numbers of the different air pollution monitoring sites and the different aerosol monitoring sites; Based on the site number, merging the coordinate data corresponding to different sites; In the coordinate data, the distance d between the two types of sites is calculated based on a distance quantization formula. If the distance d≤50km, it is considered a match, and all matching sites form the site matching result, wherein the distance quantization formula is: Where R is the radius of the earth, φi is the latitude of station i in radians, i = 1 or 2, and λ i Represents the longitude of the site i in radians, Aφ=φ2-φ1, Δλ=λ2-λ1.

3. The method according to claim 1, characterized in that The step of extracting data of corresponding sites from the sample air pollution data and the sample aerosol data according to the site matching result to obtain target sample data as training samples includes: Extracting first data of a corresponding site from the sample air pollution data and the sample aerosol data according to the site matching result, wherein the first data includes first sub-data of the corresponding site extracted from the sample aerosol data according to the site matching result and second sub-data of the corresponding site extracted from the sample air pollution data according to the site matching result; In the first data, the first sub-data is adjusted to the date format adopted by the second sub-data, and the average is calculated based on the data collection time point corresponding to the date format, so as to correspond to the second sub-data in format; The first data are sorted according to time to obtain second data as the target sample data.

4. The method according to claim 1, characterized in that: Before training the aerosol data prediction model based on the target sample data, the method further includes: The data related to the target sample data are stored separately in site units.

5. The method according to claim 1, characterized in that: The sample air pollution data specifically includes data of PM2.5, PM10, SO2, NO2, O3 and CO; The sample aerosol data specifically include several types of data, including aerosol optical thickness AOD, Angstrom index AE, single scattering albedo SSA and fine mode fraction FMF.

6. The method according to claim 1, characterized in that The method further comprises: Acquire sample meteorological data monitored by different ground meteorological monitoring stations within the selected time range; Matching the two types of sites, namely, the different air pollution monitoring sites and the different ground meteorological monitoring sites, is performed to obtain a new site matching result, and according to the new site matching result, new data of the corresponding site is extracted from the sample air pollution data and the sample period data, and then added to the target sample data; The aerosol data prediction model is specifically used to predict the corresponding aerosol data according to the atmospheric pollution data and meteorological data input into the model.

7. The method according to claim 6, characterized in that The sample meteorological data specifically include several types of data, including 2-meter dew point temperature, 2-meter temperature, 10-meter U-direction wind speed component, 10-meter V-direction wind speed component and boundary layer height.

8. A configuration device for an aerosol data prediction model, characterized in that: The device comprises: A first acquisition unit is used to acquire sample air pollution data monitored by different air pollution monitoring stations within a selected time range; A second acquisition unit is used to acquire sample aerosol data monitored by different aerosol monitoring stations within the selected time range; A matching unit, used for performing matching processing on the two types of sites, namely, the different air pollution monitoring sites and the different aerosol monitoring sites, to obtain a site matching result, and extracting data of corresponding sites from the sample air pollution data and the sample aerosol data according to the site matching result, to obtain target sample data as training samples; A training unit is used to train an aerosol data prediction model based on the target sample data, wherein the aerosol data prediction model is used to predict corresponding aerosol data according to the atmospheric pollution data input into the model.

9. A processing device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor executes the method according to any one of claims 1 to 7 when calling the computer program in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor to execute the method according to any one of claims 1 to 7.