Air pollutant concentration prediction method and related equipment thereof

Abstract

The invention discloses an air pollutant concentration prediction method and related equipment thereof, and relates to the technical field of data processing. The method comprises the following steps: acquiring current pollutant concentration data, current meteorological data and target site geographic position data corresponding to a target environment monitoring site; taking the current pollutant concentration data, the current meteorological data and the geographic position data of the target station as target input data of the prediction model; wherein an input module of the prediction model is used for performing modal decomposition processing on input data to obtain multiple pieces of modal data; the time feature extraction module is used for extracting time feature data; the spatial feature extraction module is used for extracting spatial feature data; the space-time fusion convolutional network module is used for extracting space-time features; the output module is used for outputting a prediction result; determining a prediction result of the air pollutant concentration corresponding to the target environment monitoring station according to the output result; the method achieves the effects of improving the comprehensiveness and accuracy of a prediction result and expanding application scenes.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method for predicting air pollutant concentration and related equipment. Background Art

[0002] Air quality is directly related to people's health, quality of life, and sustainable social and economic development. Predicting the concentration of air pollutants can help take measures in advance to reduce pollution emissions and protect environmental quality; timely warning of high-pollution weather can help relevant departments take necessary protective measures to reduce health hazards.

[0003] The commonly used air pollutant concentration prediction methods in related technologies mainly include the following: (1) Statistical models, such as predicting by establishing linear relationships based on historical data; (2) Machine learning models, such as using random forests (RandomForest), to improve prediction accuracy by integrating multiple decision trees. (3) Deep learning models, such as processing time series data through long short-term memory networks (LSTMs) to capture long-term dependencies and achieve prediction. Although the above methods perform well in some aspects, they also have some limitations: most models only use data of a single type and a single site, ignoring multi-type, multi-modal, and multi-site data, and fail to fully consider the temporal and spatial correlations, resulting in incomplete prediction results and low accuracy of prediction results; and many models use fixed parameter settings, which are difficult to adapt to dynamically changing environmental conditions, limiting the applicable scenarios of air pollutant concentration prediction schemes; at the same time, important information may be lost or more unimportant data may be retained during the data preprocessing stage, affecting the prediction effect and the final prediction results.

[0004] With respect to the above-mentioned problems in the related technologies, no effective solution has been proposed yet. Summary of the invention

[0005] An air pollutant concentration prediction method and related equipment provided by an embodiment of the present invention at least solve the problem that the air pollutant concentration prediction method in the related art has poor prediction results comprehensiveness and accuracy and narrow applicable scenarios.

[0006] In order to solve the above problem, one aspect of an embodiment of the present invention provides a method for predicting air pollutant concentration, comprising: Obtain the current pollutant concentration data, current meteorological data and geographic location data of the target environmental monitoring site; The current pollutant concentration data, current meteorological data and target site geographic location data are used as target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results; The prediction results of air pollutant concentrations corresponding to the target environmental monitoring sites are determined based on the output results of the prediction model.

[0007] In some of the embodiments, the method further comprises the step of training the prediction model: Obtain historical pollutant concentration data, historical meteorological data and station geographic location data corresponding to multiple environmental monitoring stations; The variational modal decomposition algorithm is used to perform variational modal decomposition on the historical pollutant concentration data of each station through the input module of the prediction model, and the particle swarm optimization algorithm is used to optimize the variational modal decomposition results to obtain multiple modal data corresponding to each station, and the multiple modal data are normalized; Extracting time feature data from multiple modal data corresponding to each site through a time convolution network in a time feature extraction module; The historical pollutant concentration data, historical meteorological data and geographic location data of each station are input into the spatial feature extraction module to construct the adjacency matrix in the graph convolutional network, and the spatial feature data is extracted through the graph convolutional network; The temporal feature data and the spatial feature data are input into the convolution gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data and the spatial feature data, and the attention mechanism is introduced to obtain the spatiotemporal features corresponding to different modalities; wherein the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; The predicted outputs corresponding to all modes are aggregated through the output module to obtain predicted concentration output feature data, and the predicted concentration output feature data is denormalized to obtain the prediction result.

[0008] In some embodiments, after the step of obtaining historical pollutant concentration data, historical meteorological data, and station geographic location data corresponding to a plurality of environmental monitoring stations, the method further includes: The historical pollutant concentration data and historical meteorological data corresponding to multiple environmental monitoring stations are preprocessed through the input module; wherein the preprocessing includes missing value processing and outlier processing; Calculate the mutual information values ​​of historical pollutant concentration data and historical meteorological data, sort the mutual information values, and select historical pollutant concentration types and meteorological data types that meet the target quantity.

[0009] In some of the embodiments, the prediction model further includes a Mogrifier gating operation module. Before the step of extracting time feature data from the multiple modal data corresponding to each site respectively through the time convolution network in the time feature extraction module, the training step of the prediction model further includes: The Mogrifier gating operation module is used to optimize the multiple modal data corresponding to each site.

[0010] In some of the embodiments, the method further comprises the step of optimizing the prediction model: The particle swarm optimization algorithm is used to search for the optimal parameter configuration of the prediction model to optimize the prediction model.

[0011] In some of the embodiments, the step of optimizing the prediction model further includes: Determine the loss function based on the predicted output result, the actual output result and the predicted time step value; Determine the initialization parameters and particle swarm parameters of the prediction model; the particle swarm parameters include particle swarm size, maximum number of iterations, inertia weight, acceleration factor, initial velocity, and initial position; Iteratively update the particle swarm to obtain the speed and position of each particle, the individual extreme value of each particle, and the global extreme value of the particle swarm after each iterative update; The output parameters of the prediction model are updated according to the position of the particle corresponding to the final global extreme value of the particle swarm obtained when the maximum number of iterations is met, so as to complete the optimization of the prediction model.

[0012] In some of the embodiments, when the method is applied to predict the concentration of air pollutants in a target area, the target environmental monitoring site is a plurality of environmental monitoring sites in the target area; When the method is applied to predict the concentration of air pollutants at a single or multiple designated environmental monitoring sites, the target environmental monitoring sites are the single or multiple designated environmental monitoring sites.

[0013] In order to solve the above problem, one aspect of an embodiment of the present invention provides a device for predicting air pollutant concentration, comprising: A data acquisition module is used to obtain the current pollutant concentration data, current meteorological data and geographical location data of the target site corresponding to the target environmental monitoring site; A model processing module is used to use the current pollutant concentration data, the current meteorological data and the geographic location data of the target site as the target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data (time series features) from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results; The prediction result determination module is used to determine the prediction result of the air pollutant concentration corresponding to the target environmental monitoring site according to the output result of the prediction model.

[0014] In some of the embodiments, the device also includes a prediction model training module, which is used to: obtain historical pollutant concentration data, historical meteorological data and station geographic location data corresponding to multiple environmental monitoring stations respectively; use a variational mode decomposition algorithm to perform variational mode decomposition on the historical pollutant concentration data of each station through the input module of the prediction model, use a particle swarm optimization algorithm to optimize the variational mode decomposition results to obtain multiple modal data corresponding to each station, and perform data normalization on the multiple modal data; extract time feature data from the multiple modal data corresponding to each station respectively through the time convolution network in the time feature extraction module; input the historical pollutant concentration data, historical meteorological data and station geographic location data of each station into the prediction model input module; In the spatial feature extraction module, the adjacency matrix in the graph convolutional network is constructed, and the spatial feature data is extracted through the graph convolutional network; the temporal feature data and the spatial feature data are input into the convolutional gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data with the spatial feature data, and the attention mechanism is introduced to obtain the spatiotemporal features corresponding to different modalities; wherein, the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; the predicted outputs corresponding to all modalities are aggregated through the output module to obtain the predicted concentration output feature data, and the predicted concentration output feature data is denormalized to obtain the prediction result.

[0015] In order to solve the above problems, one aspect of an embodiment of the present invention provides an electronic device, including: a processor, and a memory for storing a program, wherein the program includes instructions, and when the instructions are executed by the processor, the processor executes any of the above-mentioned methods for predicting the concentration of air pollutants.

[0016] In order to solve the above problem, one aspect of an embodiment of the present invention provides a non-transitory machine-readable medium storing computer instructions, where the computer instructions are used to enable a computer to execute any of the above-mentioned methods for predicting the concentration of air pollutants.

[0017] The beneficial effects of the embodiments of the present invention are as follows: by obtaining the current pollutant concentration data, current meteorological data and geographical location data of the target site corresponding to the target environmental monitoring site; using the current pollutant concentration data, current meteorological data and geographical location data of the target site as the target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to output the predictions corresponding to multiple modalities. Aggregation is performed to output prediction results; the technical means of determining the prediction results of air pollutant concentrations corresponding to the target environmental monitoring site based on the output results of the prediction model overcomes the problems of poor comprehensiveness and accuracy of prediction results and narrow applicable scenarios in the prediction methods of air pollutant concentrations in related technologies. Through modal decomposition and feature extraction in the prediction model, key information is ensured to be retained in the prediction process, and the robustness of the model is enhanced. The spatiotemporal feature fusion module based on the prediction model enables the prediction method to adapt to more diverse application scenarios, improving its versatility and flexibility. Finally, by aggregating the prediction outputs of multiple modes, the deviation that may be caused by single-mode prediction is reduced, and the stability and reliability of the final prediction results are improved. Multimodal data fusion, time feature extraction, spatial feature extraction and spatiotemporal feature fusion are performed through the prediction model, achieving the technical effect of significantly improving the comprehensiveness and accuracy of the prediction results and expanding the applicable scenarios of the prediction method.

[0018] The details of one or more embodiments of the invention are set forth in the following drawings and description so that other features, objects, and advantages of the invention are more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0020] Figure 1 It is a main flow chart of a method for predicting air pollutant concentration according to an embodiment of the present invention.

[0021] Figure 2 It is a schematic diagram of the main process of prediction model training of an embodiment of the present invention.

[0022] Figure 3 It is a main framework schematic diagram of an apparatus for predicting air pollutant concentration according to an embodiment of the present invention.

[0023] Figure 4 It is a schematic structural diagram of the electronic device of the present invention. DETAILED DESCRIPTION

[0024] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0025] The commonly used air pollutant concentration prediction methods in related technologies mainly include the following:

[0026] (1) Statistical model. For example, CN109657842A discloses a method and device for predicting air pollutant concentrations, and electronic equipment. The method first obtains the location information, air pollutant concentrations, and historical air pollutant concentration collection time of several air detection sites adjacent to the target prediction site, and then predicts the air pollutant concentration of the target prediction site in the specified prediction time period based on the above-obtained data. However, CN109657842A mainly relies on the historical air pollutant concentration data of neighboring sites, only considers part of the spatial correlation, and fails to make full use of other relevant data (such as meteorological data, geographical location, data of all sites in the region, etc.). It is easily affected by a single data source, resulting in unstable prediction results.

[0027] (2) Machine learning models, such as CN115327041A, which discloses a method for predicting air pollutant concentrations based on correlation analysis, first sorts the air pollutant concentrations of adjacent stations and the pollutant concentrations of the target station by the Pearson coefficient, and then puts the data of the combination of several pollutant concentrations with the highest correlation and the historical pollutant concentrations of the target station into the neural network model for training. In order to explore the impact of the changes in the air pollutant concentrations of adjacent stations on the changes in the pollutant concentrations of the target station, the changes in the air pollutant concentrations of the target station can be predicted more accurately. However, CN109657842A relies on several pollutant concentration data with the highest correlation and the historical pollutant concentration data of the target station, and mainly focuses on optimizing the input data based on correlation analysis, without making full use of other relevant data (such as meteorological data, station geographical location, etc.), and fails to achieve deep integration in time and space dimensions. The prediction results are easily affected by a single data source, resulting in unstable prediction results. At the same time, the model parameter settings are relatively fixed, making it difficult to adapt to complex environmental changes.

[0028] (3) Deep learning models. For example, CN114741972A discloses a method for constructing a seasonal prediction model for air pollutant concentrations. The method divides the historical data on air pollutant concentrations into four seasonal historical data sets, namely spring, summer, autumn and winter, according to the seasons, and then determines the time series of the data sets. The long short-term memory network is used to learn the variation law of pollutant concentrations contained in the time series of air pollutant concentrations. At the same time, the Bayesian optimization algorithm is used to select the optimal hyperparameter combination for the long short-term memory network, and then a prediction model with the best prediction performance is obtained. CN113188968A discloses a method and system for predicting air pollutant concentrations based on a combined deep learning model. The method obtains the PM2.5 concentrations of air pollutants at multiple air quality stations and obtains the time series of PM2.5. The time series is used to train the constructed combined deep learning network to obtain a prediction model. The prediction model is then used to predict the PM2.5 concentration of the target station. However, CN114741972A and CN113188968A do not consider other relevant data such as meteorological data and geographical location. CN114741972A only considers feature extraction in the time dimension. Although CN113188968A extracts time and space features by combining deep learning networks, it fails to achieve deep-level spatiotemporal feature fusion, and also has the problem of poor comprehensiveness and accuracy of prediction results. For another example, CN118035777A discloses a method and device for predicting air pollutant concentrations based on spatiotemporal graph data. The method obtains monitoring site distribution data by acquiring historical data on air pollutant concentrations and corresponding meteorological data to construct a data set; initializes the MGATT-LSTM prediction model using a graph convolutional network, a long short-term memory network, and a fully connected network based on a multi-graph attention mechanism; trains and iteratively optimizes the MGATT-LSTM prediction model, and obtains the final prediction model when the prediction result meets the preset prediction conditions. CN118035777A considers the fusion of temporal features and spatial features, but does not consider the modal characteristics of the data, and there are still problems such as insufficient comprehensiveness and poor accuracy of the prediction results.

[0029] In summary, although the air pollutant concentration prediction methods of related technologies perform well in some aspects, they also have some limitations: most models only use a single type of data, ignore multi-type and multi-modal data, and fail to fully consider the correlation in time and space, resulting in incomplete prediction results; and many models use fixed parameter settings, which are difficult to adapt to dynamically changing environmental conditions, limiting the applicability of air pollutant concentration prediction schemes; important information may be lost or more unimportant data may be retained in the data preprocessing stage, affecting the prediction effect and the final prediction effect.

[0030] In order to solve the above problems, the present invention provides a method for predicting the concentration of air pollutants. Figure 1 As shown in Figure 2, the prediction method of air pollutant concentration mainly includes:

[0031] Step S101, obtaining current pollutant concentration data, current meteorological data and target site geographical location data corresponding to the target environmental monitoring site.

[0032] By simultaneously collecting pollutant concentrations, meteorological conditions (such as temperature, humidity, wind speed, etc.) and geographic information (such as site location), the integrity and diversity of the input data are ensured, providing a richer information basis for subsequent feature extraction and prediction.

[0033] According to the embodiments of the present invention, by acquiring multiple types of data, the correlation between different factors can be better analyzed through the prediction model, such as the impact of meteorological conditions on pollutant concentrations, thereby improving the accuracy of the prediction. At the same time, combining geographic location data helps capture the spatial correlation between different sites and further improve the spatial perception ability of the prediction model.

[0034] Among them, the pollutants in the embodiments of the present invention include but are not limited to sulfur dioxide, nitrogen dioxide, ozone, carbon monoxide, inhalable particulate matter (PM10, referring to particles with an aerodynamic equivalent diameter ≤ 10 microns), fine particulate matter (PM2.5, i.e. particles in the atmosphere with an aerodynamic equivalent diameter less than or equal to 2.5 microns, also known as respirable particles).

[0035] The data acquired in step S101 of the embodiment of the present invention should be the latest data, which can reflect the current environmental changes and help improve the timeliness and reliability of the prediction results.

[0036] In some of the embodiments, when the air pollutant concentration prediction method provided by the embodiments of the present invention is applied to predict the air pollutant concentration in a target area, the target environmental monitoring site is multiple environmental monitoring sites within the target area; when the prediction method is applied to predict the air pollutant concentration at a single or multiple designated environmental monitoring sites, the target environmental monitoring site is a single or multiple designated environmental monitoring sites.

[0037] When used to predict the concentration of air pollutants in a target area, by considering data from multiple stations at the same time, the target area can be fully covered and a wider range of air quality information can be provided. In addition, using data from multiple stations can better capture the spatial correlation between different locations and improve the accuracy of overall regional predictions. Data from multiple stations also provides more redundant information, enhances the robustness and reliability of the model, and reduces the impact of data anomalies at a single station.

[0038] When used to predict the concentration of air pollutants at a single or multiple designated environmental monitoring sites, it can provide more accurate and personalized prediction results based on the needs of specific sites, meeting the refined management requirements of local areas. Key sites can be selected for key monitoring according to actual needs, optimizing resource allocation and improving efficiency.

[0039] The prediction model adopted by the prediction method for air pollutant concentration provided in the embodiment of the present invention has wide applicability and high prediction accuracy, and can meet diverse prediction needs. By comprehensively utilizing data of different scales and types, the accuracy and comprehensiveness of the prediction results are ensured.

[0040] Step S102, taking the current pollutant concentration data, the current meteorological data and the geographic location data of the target site as the target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results.

[0041] Based on the above prediction model, through multimodal data processing and fusion, it is ensured that the data of each mode can be effectively utilized, avoiding the limitations brought by single-mode data and enhancing the robustness of the model. The temporal feature extraction module can accurately capture the changing law of pollutant concentration over time through methods such as TCN (Temporal Convolutional Network), enhance the prediction ability of short-term and long-term trends, and improve the accuracy of prediction. The spatial feature extraction module fully considers the spatial correlation between different sites by constructing an adjacency matrix (the adjacency matrix is ​​a parameter matrix in the graph convolutional network) and applying GCN (Graph Convolutional Network), improves the understanding of the overall pollution status of the region, and improves the prediction accuracy. The spatiotemporal convolutional network module achieves a more comprehensive and in-depth feature representation by integrating the features in the time and space dimensions, further improving the performance of the prediction model and adapting to more diverse application scenarios. The output module reduces the deviation that may be caused by single-mode prediction by aggregating the prediction outputs of multiple modes, improves the robustness and stability of the model, and ensures the high accuracy and reliability of the final prediction results.

[0042] Among them, the above-mentioned TCN is a neural network architecture for processing time series data. Unlike traditional recurrent neural networks (such as LSTM and GRU), TCN uses a one-dimensional convolutional layer to capture long-term dependencies in time series data while maintaining causality (that is, future information does not affect current predictions). TCN has the advantage of parallel computing and can process long sequence data more efficiently. The above-mentioned GCN (graph convolutional network) is a neural network architecture specifically designed for processing graph structured data. It extracts the relationship between node features and their neighbor nodes by applying convolution operations on the nodes and edges of the graph. GCN can effectively capture complex relationships and structural information in graph data.

[0043] Based on the above settings, the comprehensiveness and accuracy of air pollutant concentration prediction are significantly improved through multimodal data processing and fusion, temporal feature extraction, spatial feature extraction and spatiotemporal feature fusion. The robustness and reliability of the prediction results are also improved through a flexible output aggregation mechanism.

[0044] In some of the embodiments, the method further includes a step of training the prediction model: obtaining historical pollutant concentration data, historical meteorological data, and station geographic location data corresponding to a plurality of environmental monitoring stations; performing variational modal decomposition on the historical pollutant concentration data of each station using a variational modal decomposition algorithm through an input module of the prediction model, performing optimization processing on the variational modal decomposition results using a particle swarm optimization algorithm to obtain a plurality of modal data corresponding to each station, and performing data normalization processing on the plurality of modal data; extracting time feature data from the plurality of modal data corresponding to each station through a time convolutional network in a time feature extraction module; inputting the historical pollutant concentration data, historical meteorological data, and station geographic location data of each station into a spatial feature extraction module; and In the extraction module, the adjacency matrix in the graph convolutional network is constructed, and the spatial feature data is extracted through the graph convolutional network in the spatial feature extraction module; the temporal feature data and the spatial feature data are input into the convolutional gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data with the spatial feature data, and the attention mechanism is introduced to obtain the spatiotemporal features corresponding to different modalities; wherein, the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; the predicted outputs corresponding to all modalities are aggregated through the output module to obtain the predicted concentration output feature data, and the predicted concentration output feature data is denormalized to obtain the prediction result.

[0045] According to an embodiment of the present invention, data from multiple environmental monitoring sites are collected, and each site corresponds to different pollutant concentration data, meteorological data, and geographic location data. The prediction model trained accordingly has distributed monitoring and forecasting functions. Further, multimodal data processing and optimization are performed through the input module. Specifically, the historical pollutant concentration data of each site is subjected to variational mode decomposition through VMD (Variational Mode Decomposition) to obtain multiple modes with different frequency characteristics, ensuring that the data of each mode can be effectively utilized; at the same time, the VMD results are optimized through PSO (Particle Swarm Optimization), further optimizing the decomposition results and enhancing the robustness of the model; by normalizing multiple modal data, it is ensured that different modal data are processed on the same scale to avoid some modal data affecting the model performance due to excessive differences in magnitude.

[0046] According to an embodiment of the present invention, time feature data is extracted from multiple modal data corresponding to each site through a time feature extraction module. Specifically, TCN can efficiently capture long-term dependencies in time series, enhance the ability to predict short-term and long-term trends, and compared with the traditional RNN structure, TCN has the advantage of parallel computing, which improves the efficiency of processing long sequence data.

[0047] According to a specific implementation of an embodiment of the present invention, when VMD (variational mode decomposition) is performed on the historical pollutant concentration data of each station, each mode does represent a specific aspect or pattern of the original data. These modes can capture different frequency components, trends, periodic changes and other hidden features in the data. Specifically, different modes may represent the following specific aspects or patterns: long-term trends (i.e., low-frequency modes usually reflect long-term trends or slowly changing parts of the data. For pollutant concentration data, low-frequency modes may capture annual or seasonal trends, such as higher pollution levels in winter and lower pollution levels in summer), short-term fluctuations (i.e., high-frequency modes, which reflect the impact of rapid changes or instantaneous events. This may include a sharp increase or decrease in pollutant concentrations in a short period of time due to increased traffic flow, intensified industrial activities, or special weather phenomena (such as sandstorms, rainfall, etc.)), periodic changes (some modes may show obvious periodic behavior, which may be at the daily, weekly or monthly level. For pollutant concentration data, it may be the weekly traffic pattern. The impact of pollutant emissions), random noise or sudden events (some modes may be mainly composed of random noise or non-periodic sudden events. This may include measurement errors of monitoring equipment, sudden local pollution sources (such as fire, accidental emissions), etc.), intermediate frequency modes (in addition to extreme high and low frequencies, there are some intermediate frequency modes, which are between long-term trends and short-term fluctuations. These modes may capture fluctuations on longer time scales, but not as long as interannual changes, such as air quality trends within a few months), spatial correlation patterns (when considering data from multiple sites, some modes may reflect correlations between geographical locations. If two adjacent sites are affected by the same pollution source, then their common modes may exhibit similar time series characteristics).

[0048] According to the above specific implementation, the multiple modal data obtained by modal decomposition of historical pollutant concentration data by VDM can be characterized as follows: (1) Low-frequency mode: reflects long-term trends or slow changes, which may reveal the long-term improvement effect after the implementation of environmental policies, or the impact of natural climate cycles on air pollution. (2) High-frequency mode: reflects short-term fluctuations or instantaneous events, which can help identify the impact of traffic emissions during daily commuting peak hours, or short-term pollution events under specific weather conditions. (3) Periodic mode: reflects periodic changes, such as daily, weekly, and monthly cycles; helps to understand how factors such as day and night temperature differences and seasonal changes periodically affect air quality. (4) Random noise mode: reflects random noise or non-periodic events; provides a way to evaluate the stability and reliability of the monitoring system and exclude the influence of outliers. (5) Spatial correlation mode: reflects the influence of spatial location, which can show the pollution transmission path and interaction between different regions, which is very important for formulating cross-regional environmental protection measures.

[0049] Given the complexity of pollutant concentration data, pollutant concentration data usually have more complex time series characteristics and higher volatility. These data may contain multiple frequency components (such as long-term trends, short-term fluctuations, cyclical changes, etc.), so VMD decomposition is needed to extract different modal features. Through VMD decomposition, not only can complex data sets be simplified into a form that is easier to understand and process, but also potential important information can be mined, thereby significantly improving the performance and accuracy of the prediction model.

[0050] According to an embodiment of the present invention, an adjacency matrix integrating historical pollutant concentration data, historical meteorological data, and site location data in the input data is constructed through a spatial feature extraction module, and spatial feature data is extracted through GCN. Specifically, GCN fully considers the spatial correlation between different sites, improves the understanding of the overall pollution status of the region, and the extraction of spatial features enables the prediction model to better handle cross-site impacts, thereby improving the accuracy of the prediction results.

[0051] According to a specific embodiment of the embodiment of the present invention, the temporal feature data and the spatial feature data can be directly input into the convolutional gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data with the spatial feature data, and introduce the attention mechanism to obtain the spatiotemporal features corresponding to different modalities; wherein the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station.

[0052] According to another specific embodiment of the embodiment of the present invention, the output of the time convolution network can be connected to the input of the graph convolution network based on the time-space fusion convolution network module, and then the time feature data and the spatial feature data are fused through the convolution gated recurrent unit (ConvGRU) in the time-space fusion convolution network, and the attention mechanism is introduced to obtain the time-space features corresponding to different modalities. Specifically, by fusing the features in the time and space dimensions, a more comprehensive and in-depth feature representation is achieved, further improving the performance of the prediction model. The introduced attention mechanism can dynamically allocate weights according to importance, ensuring that key information receives more attention and improving the accuracy of the prediction. The fully connected layer of the time-space fusion convolution network module connects the time-space features of a single modality respectively, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station. The time-space features of different modalities are integrated through the fully connected layer to ensure that the information of each modality can be fully utilized.

[0053] According to a specific implementation of an embodiment of the present invention, a convolutional network module based on spatiotemporal fusion connects the output of the temporal convolutional network with the input of the graph convolutional network, and then performs spatiotemporal fusion through a convolutional gated recurrent unit (ConvGRU). It is ensured that the time series features can be integrated into the spatial features in an orderly manner. Specifically, TCN has extracted the local temporal features of each site from the time dimension, which can be regarded as the state representation of each node at different time points. When these state representations are passed to GCN as input, GCN can further capture the spatial dependencies between these states in the graph structure. Furthermore, by using a gated recurrent unit such as ConvGRU, spatial information can be effectively fused while maintaining temporal continuity. The update gate and reset gate in ConvGRU can control which temporal features should be retained or forgotten, and how to combine new spatial features. This method helps to dynamically adjust the importance of spatiotemporal features and enhance the model's understanding of complex spatiotemporal patterns. In addition, the introduction of an attention mechanism can further improve the effect of spatiotemporal feature fusion. The attention mechanism allows the model to automatically adjust the weights at each position or time step according to the needs of the current task, highlighting the information that is most helpful for prediction. By first letting TCN process the time series data, then letting GCN process the spatial association, and finally applying the attention mechanism, the whole process can be made more orderly and targeted.

[0054] According to the embodiment of the present invention, the predicted outputs corresponding to all modalities are aggregated through the output module to obtain the final predicted concentration output characteristic data, and the predicted concentration output characteristic data is denormalized to obtain the final prediction result, thereby improving the comprehensiveness and accuracy of the prediction result. Furthermore, the denormalization process restores the data to the original scale, ensuring the actual interpretability and application value of the prediction result.

[0055] Through the above training steps, multiple modal data are finely extracted and optimized through variational mode decomposition and particle swarm optimization algorithms, which enhances the robustness and accuracy of the model. The long-term dependencies in the time series are efficiently captured by the time convolution network, which enhances the prediction ability of short-term and long-term trends and improves the accuracy of the prediction. The spatial correlation between different sites is fully considered through the graph convolution network, which improves the understanding of the overall pollution status of the region and the accuracy of the prediction results. Through the convolutional gated recurrent unit and the attention mechanism, the feature fusion in the time and space dimensions is realized, which further improves the performance of the prediction model and adapts to more diverse application scenarios. Through the fully connected layer and anti-normalization processing, the comprehensiveness and practical application value of the prediction results are ensured, and the practicality and reliability of the model are enhanced. The technical effect of significantly improving the comprehensiveness and accuracy of the prediction of air pollutant concentrations and enhancing the robustness and practicality of the prediction model is achieved.

[0056] In some of the embodiments, after the step of obtaining historical pollutant concentration data, historical meteorological data and site geographic location data corresponding to multiple environmental monitoring sites, the method further includes: preprocessing the historical pollutant concentration data and historical meteorological data corresponding to multiple environmental monitoring sites through an input module; wherein the preprocessing includes missing value processing and outlier processing; calculating the mutual information value of the historical pollutant concentration data and the historical meteorological data, sorting the mutual information values, and selecting the historical pollutant concentration types and meteorological data types that meet the target quantity.

[0057] Among them, the missing value processing provided in the embodiment of the present invention refers to identifying and processing missing values ​​in data. Common methods include interpolation (such as mean interpolation, linear interpolation) or deleting data records containing missing values. Through missing value processing, data integrity is ensured, model deviation caused by missing values ​​is avoided, and the stability and accuracy of subsequent model training are improved.

[0058] Among them, the outlier processing provided by the embodiment of the present invention identifies and processes abnormal values ​​(outliers), and common methods include statistical detection (such as Z-score, IQR), machine learning methods (such as Isolation Forest), etc. Outliers are removed or corrected through outlier processing to ensure the consistency and rationality of data, reduce the interference of abnormal data on model training, and improve the robustness and generalization ability of the model.

[0059] According to an embodiment of the present invention, the mutual information method can be used to perform correlation analysis on historical pollutant concentration data and historical meteorological data to find the top N types of historical pollutant concentrations and historical meteorological data that are closely correlated with the air pollutant concentration to be predicted. Among them, mutual information (MI) is a statistic that measures the mutual dependence between two random variables. It quantifies the amount of information about another variable that can be obtained by observing one variable. The mutual information method is widely used in feature selection, pattern recognition, signal processing and other fields, especially when it is necessary to evaluate the correlation between different data sets.

[0060] Among them, the mutual information value provided by the embodiment of the present invention can quantify the correlation between different features and help identify the features that have the greatest influence on the target variable (air pollutant concentration). Then, by sorting the mutual information values, you can intuitively understand which features have the greatest impact on the prediction results, which improves the interpretability of the model. Then, according to the mutual information value sorting, select the historical pollutant concentration type and meteorological data type that meet the target quantity. The ability to select the most influential features reduces the data dimension, reduces the computational complexity, and improves the efficiency of model training. And while retaining the most useful information for prediction and removing redundant features, it helps to improve the accuracy and stability of the prediction.

[0061] In some of the embodiments, the above-mentioned prediction model also includes a Mogrifier gating operation module. Before the step of extracting time feature data from multiple modal data corresponding to each site through the time convolution network in the time feature extraction module, the training step of the prediction model also includes: performing data optimization on the multiple modal data corresponding to each site through the Mogrifier gating operation module.

[0062] Based on the above settings, the Mogrifier gating operation can dynamically select and retain the most useful information for prediction, avoiding the important information that may be lost in traditional preprocessing methods, and filtering out unimportant or redundant data through the gating mechanism, reducing the burden of subsequent processing and improving computational efficiency.

[0063] According to a specific implementation of an embodiment of the present invention, the Mogrifier gating operation module can also be used to perform adaptive adjustments according to the specific characteristics of the data. The Mogrifier gating mechanism automatically determines which parts of the data need to be retained and which can be appropriately modified by learning gating parameters. This helps to avoid information loss caused by over-simplification or fixed-rule data preprocessing. And the interaction between forward and backward gating enables the model to gradually optimize the data representation in multiple iterations, which not only ensures the integrity of the data, but also enhances the feature expression ability. That is, adaptive optimization is performed according to the data characteristics of different sites and different time periods, so that the prediction model can remain stable in the face of abnormal data or noise, which not only enhances the adaptability of the prediction model to complex environmental changes, but also enhances the reliability of the prediction results. In some of the embodiments, the above method also includes an optimization step of the prediction model: using a particle swarm optimization algorithm to search for the optimal parameter configuration of the prediction model to achieve optimization of the prediction model.

[0064] By adopting the particle swarm optimization algorithm (PSO) to optimize the prediction model, the present invention not only significantly improves the optimization level of the model parameter configuration, but also significantly improves the overall performance of the prediction model by means of global optimal search, precise parameter configuration, improved model performance and accelerated convergence.

[0065] In some of the embodiments, the optimization step of the above-mentioned prediction model also includes: determining a loss function based on the predicted output result, the actual output result and the predicted time step value; determining the initialization parameters and particle swarm parameters of the prediction model; wherein the particle swarm parameters include the particle swarm size, the maximum number of iterations, the inertia weight, the acceleration factor, the initial velocity, and the initial position; iteratively updating the particle swarm to obtain the velocity and position of each particle after each iterative update, the individual extreme value of each particle, and the global extreme value of the particle swarm; updating the output parameters of the prediction model according to the position of the particle corresponding to the final global extreme value of the particle swarm obtained when the maximum number of iterations is met, so as to complete the optimization of the prediction model.

[0066] Through the above settings, the difference between the predicted results and the actual results can be accurately quantified based on the loss function, providing a clear goal for model optimization; by introducing the prediction time step value, the loss function can dynamically reflect the prediction error in different time periods, ensuring that the optimization process takes into account the influence of the time dimension. The setting of initialization parameters and particle swarm parameters improves the convergence speed and stability of the optimization process. Then, through multiple iterative updates, the particle swarm gradually approaches the optimal solution, ensuring the gradualness and effectiveness of the optimization process and avoiding falling into the local optimal solution. Finally, by updating the output parameters of the prediction model, the model performance is ensured to reach the optimal state, which significantly improves the prediction accuracy and robustness and enhances the generalization ability.

[0067] Step S103, determining the prediction result of the air pollutant concentration corresponding to the target environmental monitoring site according to the output result of the prediction model.

[0068] Through the processing of input data by the above-mentioned prediction model, multiple factors such as pollutant concentration, meteorological conditions and geographical location are comprehensively considered, providing prediction results with high prediction accuracy, high reliability and high timeliness. The obtained prediction results have high practical application value and can be used as data support for environmental protection and public health management.

[0069] The above-mentioned air pollutant concentration prediction method provided by the embodiment of the present invention adopts the method of obtaining the current pollutant concentration data, current meteorological data and geographical location data of the target site corresponding to the target environmental monitoring site; the current pollutant concentration data, current meteorological data and geographical location data of the target site are used as the target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each mode; the output module is used to convert the multiple modal corresponding The prediction outputs of the prediction model are aggregated to output the prediction results; the technical means for determining the prediction results of the air pollutant concentrations corresponding to the target environmental monitoring sites according to the output results of the prediction model overcomes the problems of poor comprehensiveness and accuracy of the prediction results and narrow applicable scenarios in the prediction methods of air pollutant concentrations in related technologies. The modal decomposition and feature extraction in the prediction model are achieved to ensure that key information is retained in the prediction process and enhance the robustness of the model. The spatiotemporal feature fusion module based on the prediction model enables the prediction method to adapt to more diverse application scenarios and improves its versatility and flexibility. Finally, by aggregating the prediction outputs of multiple modes, the deviation that may be caused by single-mode prediction is reduced, and the stability and reliability of the final prediction results are improved. Multimodal data fusion, time feature extraction, spatial feature extraction and spatiotemporal feature fusion are performed through the prediction model to significantly improve the comprehensiveness and accuracy of the prediction results and expand the applicable scenarios of the prediction method.

[0070] The present invention also provides a specific embodiment of prediction model training and optimization. Figure 2 As shown in Figure 1, the prediction model training and optimization mainly includes the following steps:

[0071] Step S201, establish an input module of the prediction model, collect hourly historical pollutant concentration data, historical meteorological data and station geographical location data from environmental monitoring stations in various places.

[0072] Step S202, pre-processing the historical pollutant concentration data and the historical meteorological data; using the mutual information method to perform correlation analysis on the historical pollutant concentration data and the historical meteorological data, and finding the top N types of historical pollutant concentration data and historical meteorological data that are closely correlated with the predicted air pollutant concentration; using the variational mode decomposition algorithm and the particle swarm optimization algorithm to perform modal decomposition and optimization processing on the historical pollutant concentration data of each station, to obtain multiple modal data, and standardize each modal data.

[0073] Among them, the role of PSO-VMD (variational mode decomposition-particle swarm optimization algorithm) is to reduce the complexity of pollutant concentration data, improve the prediction accuracy of the prediction model, and enhance the adaptability of input data.

[0074] The specific steps of using PSO-VMD (variational mode decomposition-particle swarm optimization method) to perform modal decomposition and optimization processing on the historical pollutant concentration data of each station include: using VMD (variational mode decomposition algorithm) through the input module to perform variational mode decomposition on the historical pollutant concentration data of each station, and using PSO (particle swarm optimization algorithm) to optimize the variational mode decomposition results to obtain multiple modal data corresponding to each station.

[0075] Among them, the characteristics of multiple modal data are expressed as: , k is the modal number; after standard normalization of each modal data, the normalized modal data is recorded as: , k is the mode number.

[0076] According to a specific implementation of an embodiment of the present invention, PSO-VMD (variational mode decomposition-particle swarm optimization algorithm) is used to perform mode decomposition and optimization processing on the historical pollutant data of each station to obtain feature representations of N different modes. , k is the modal number, specifically including the following sub-steps:

[0077] Step S2021, establish a VMD optimization problem and construct the following expression: , (1) In the above formula (1), It is the modal component of historical pollutant data after VMD decomposition; is the center frequency of each modal component; is the impact function; is the gradient calculation; is the original signal.

[0078] Step S2022: Introducing a secondary penalty factor , transform the constrained variational problem into an unconstrained variational problem, expressed as (2): (2).

[0079] Step S2023, using PSO to find the optimal solution, , .

[0080] Step S2024, initializing particle swarm parameters and optimization parameters, , .

[0081] Step S2025, starting from k=1, the particle swarm is updated and iterated to obtain the updated speed and position of each particle, and the individual extreme value of the particle and the global extreme value of the particle swarm are updated according to step S2027 until the residual is optimized. , end the update iteration and obtain the final global extreme value of the particle swarm.

[0082] Step S2026, using the position of the particle corresponding to the global extreme value of the final particle swarm to update the input parameters of the reconstructed VMD model, and inputting the updated input parameters into the VDM model. At this time, the output of the VMD model is recorded as the optimal hierarchical mode for the input original signal.

[0083] Step S2027: The process of updating and iterating the particle swarm in step S2025 is as follows: k is the current iteration number of the particle swarm, and the current position of the i-th particle after the k-th iteration is recorded as , , the current velocity of the i-th particle after the k-th iteration is recorded as , ; The particle updates its velocity and position according to the following equations (3) and (4): (3) (4) In the above formulas (3) and (4), is the inertia weight; is the velocity of the particle; and is a non-negative constant, called the acceleration factor; and is a random number distributed between [0,1]; ; ; represents the minimum velocity of the ith particle, Represents the maximum velocity of the ith particle.

[0084] Step S2028: When the residual index is met, the optimization is terminated, and finally the layered signals of different modes are output. , k is the mode number.

[0085] Step S2029: Under the premise that the number of decomposition levels K is determined, the residual index The calculation process is as follows: (5) In the above formula (5), For each decomposition mode, is the original signal.

[0086] Step S203: establish a Mogrifier gating mechanism for the input data, and optimize the multiple modal data corresponding to each site through the Mogrifier gating operation module.

[0087] By establishing a Mogrifier mechanism for the input data information and establishing a Mogrifier gating operation, it is possible to prevent information loss caused by the data preprocessing in step S202.

[0088] According to a specific implementation of an embodiment of the present invention, the specific implementation steps of establishing a Mogrifier gating mechanism for input data include: For input and the output of the previous hidden layer Perform Mogrifier gating operation as follows: , if t is an odd number (6) , if t is an even number (7) In the above formulas (6) and (7), and They are the current input value and the intermediate value in the Mogrifier structure iteration, , ; and is the intermediate value in the hidden layer output and Mogrifier structure, ; is the sigmoid operation; t is the hyperparameter in the Mogrifier structure; , To help , An additional matrix is ​​created for interaction. It is Hadamard.

[0089] Step S204: establish a time feature extraction module, and extract time feature data from multiple modal data corresponding to each site through a time convolution network in the time feature extraction module.

[0090] According to a specific implementation of the embodiment of the present invention, for different modes k, the step of extracting the temporal features of the input data using a multi-scale TCN operation includes: (8) (9) In the above formulas (8) and (9), is the tth element of the input sequence; is the expansion rate, which varies with the mode; is the convolution kernel size; is a trainable parameter; is the tanh activation function; is the output of the temporal feature extraction module of the kth mode; Represents a one-dimensional convolution layer with a convolution kernel of 1×1.

[0091] Step S205, establish a spatial feature extraction module, input the historical pollutant concentration data, historical meteorological data and corresponding site geographic location data of the first N types obtained in step S202 into the spatial feature extraction module, construct an adjacency matrix in the graph convolutional network, and extract spatial feature data through the graph convolutional network.

[0092] Adjacency Matrix is ​​a very important concept in graph theory and graph convolutional network (GCN). It is used to represent the connection relationship between nodes in the graph.

[0093] According to a specific implementation of an embodiment of the present invention, the steps of constructing an adjacency matrix and extracting spatial feature data through a graph convolutional network include:

[0094] Step S2051, construct an adaptive graph convolutional network GCN and an adjacency matrix A according to the distribution of monitoring stations.

[0095] Step S2052: All sites and their pairwise relationships are represented as a weighted graph G=(V,E,A), where V is a site set including N sites; E is an edge set; is the adjacency matrix, where the weights Representative Site V i and Site V j The spatiotemporal correlation strength between them, the edges are connected according to the weights, , .

[0096] Step S2053, establish the adjacency matrix A ( ), matrix A integrates pollutant concentration data, meteorological data and site location data.

[0097] Step S2054, define the time series data adjacency matrix of pollutant concentration data as: (10) In the above formula (10), an adjacency matrix representing the temporal similarity graph of particle contaminants; Environmental Monitoring Station Time series data of N pollutant concentrations; Environmental Monitoring Station The average DTW distance between the time series data of pollutant concentrations; represents the variance of Gaussian distribution of pollutant data at the site; Indicates the distance threshold of the set pollutant concentration set time mode.

[0098] Step S2055, define the adjacency matrix of meteorological information between sites as: (11) In the above formula (11), The adjacency matrix representing the temporal similarity graph of meteorological information; Environmental Monitoring Station Meteorological data; Environmental Monitoring Station The average DTW distance between the time series data of the weather; Indicates the variance of Gaussian distribution of meteorological data at the site; Indicates the distance threshold of the set weather data setting time mode.

[0099] Step S2056, define the adjacency matrix of spatial information: Site The geographical location is represented by , site The geographical location is represented by ,in, Respectively represent sites The latitude and longitude of Respectively represent sites Latitude and longitude of the site The geographical distance between is: (12) (13) (14) (15) (16) In the above formulas (12)-(16), Represents the conversion function between angle and radian, where R is the equatorial radius of the earth.

[0100] The spatiotemporal adaptive matrix E that integrates the temporal information of pollutant concentration data, meteorological data, and spatial information between stations: (17) In the above formula (17), , , , is a trainable parameter, is the activation function.

[0101] Take the adjacency matrix A as: (18) In the above formula (18), E represents the spatiotemporal adaptive matrix, and A represents the spatiotemporal adaptive adjacency matrix.

[0102] Step S2057, for each modality k, a graph convolutional network GCN is established and represented as: (19) It can be decomposed into the following operating rules: (20) (twenty one) (twenty two) In the above formulas (20)-(22), A is the adjacency matrix of the graph structure; is the identity matrix; is an adjacency matrix with self-connection; Degree matrix representing the adjacency matrix with self-connection; The feature vector of the kth mode and the lth layer of the graph node; That is, it is the feature vector of the node in the l+1 layer after the k-mode is convolved; when l=0, ; Represents the learnable parameters of the l-th layer convolution; is the sigmoid activation function.

[0103] Step S206, input the time feature data and the space feature data into the convolution gated recurrent unit in the time-space fusion convolutional network module to fuse the time feature data with the space feature data, and introduce the attention mechanism to obtain the time-space features corresponding to different modalities; wherein the fully connected layer of the time-space fusion convolutional network module respectively connects the time-space features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station.

[0104] Specifically, the above steps require the establishment of a convolutional network model TSNN model that integrates time and space. That is, the temporal convolutional network (TCN) and the graph convolutional network (GCN) are integrated by updating the gating and resetting the gating, and the spatiotemporal features of a single modality are connected through the attention transmission mechanism and the fully connected layer FC to obtain the convolutional network TSNN model for predicting air pollutant concentrations with spatiotemporal feature coupling.

[0105] According to a specific implementation of an embodiment of the present invention, a spatiotemporal convolutional network model TSNN model is established, which specifically includes the following sub-steps: In step S2061, the spatiotemporal convolutional network TSNN model of air quality pollutant concentration obtains the output of the temporal feature extraction convolutional network TCN and the graph convolutional network GCN, and inputs it into the convolutional gated recurrent unit ConvGRU model for spatiotemporal fusion. According to formula (23), the update gate in the convolutional gated recurrent unit ConvGRU model structure is used to process the data output by GCN to extract the temporal features of historical data: (twenty three) In the above formula (23), , are trainable weights; is the bias term; is the sigmoid function; is the state of a hidden layer on the graph convolutional network. When t = 0, It is the output data of the input module after the Mogrifier gating operation.

[0106] Step S2062, according to formula (24), the reset gate in the convolutional gated recurrent unit ConvGRU model structure is used to further process the data transmitted at the previous moment, and further extract the time series features of the historical data: (twenty four) In the above formula (24), , are trainable weights; is the bias term; is the sigmoid function; is the output of the k-th modal time feature extraction module; is the state of a hidden layer on the graph convolutional network. When t = 0, It is the output data of the input module after the Mogrifier gating operation.

[0107] Step S2063, the data transmitted at the last moment and The current memory content in ConvGRU is obtained by processing, and the calculation formula is as follows: (25) In the above formula (25), , , are trainable weights; is the state of a hidden layer on the graph convolutional network. When t = 0, It is the output data of the input module after the Mogrifier gate operation; is the Hadamard product; tanh is a hyperbolic tangent activation function.

[0108] Step S2064: Integrate the calculation results of equations (24) to (25) to obtain the final hidden layer state information at the current moment: , which is calculated as follows: (26) In the above formula (26), is a trainable parameter; is the output feature of spatiotemporal fusion.

[0109] In step S2065, the global spatiotemporal characteristics of multiple environmental monitoring stations are obtained by introducing the following attention mechanism as shown in equations (27)-(29): (27) (28) (29) In the above formulas (27)-(29), , is the weight vector; is the attention coefficient; is the global spatiotemporal feature of the kth mode.

[0110] Step S2066: The fully connected output module in step S206 is to connect the global spatiotemporal features output by TSNN through the fully connected layer FC, and output the predicted output of the pollutant concentration of the kth mode of the output station. .

[0111] Step S207, aggregate the predicted outputs corresponding to all modes through the output module to obtain predicted concentration output feature data, and perform denormalization on the predicted concentration output feature data to obtain a prediction result.

[0112] Specifically, the above steps need to first construct the output layer of the prediction model, then aggregate the prediction outputs of k modes, predict the concentration output feature data of all stations for T time steps (for example, one time step per hour in the next 24 hours, that is, T=24), and denormalize the output feature data to obtain the prediction results.

[0113] According to a specific implementation of an embodiment of the present invention, an output layer of a prediction model is established. The prediction outputs of k modes are aggregated: (30) In the above formula (30), represents the prediction results of air pollutant concentrations of k modes, is the weight aggregation function.

[0114] right Perform inverse normalization to obtain the predicted output of air pollutant concentration .

[0115] Step S208, using a particle swarm optimization algorithm to search for the optimal parameter configuration of the prediction model to optimize the prediction model.

[0116] The particle swarm optimization algorithm (PSO) is used to search for uncertain parameters in the prediction model, which helps to improve the prediction accuracy, prevent falling into the local optimum, and obtain a prediction model that meets the requirements.

[0117] According to a specific implementation of an embodiment of the present invention, a particle swarm optimization algorithm (PSO) is used to train the uncertain parameters of the prediction model composed of an input module, a TCN module, a GCN module and a TSNN module, including the following steps:

[0118] Step S2081, determine the loss function: (31) In the above formula (31), Represents the predicted output; Indicates the actual output; Represents the forecast time step.

[0119] Step S2082, determine the optimization initialization parameters and the PSO initialization parameters.

[0120] Step S2083, starting from k=1, the particle swarm is updated and iterated to obtain the updated speed and position of each particle, and the individual extreme value of the particle and the global extreme value of the particle swarm are updated until , end the update iteration and obtain the final global extreme value of the particle swarm.

[0121] Step S2084, using the position of the particle corresponding to the global extreme value of the final particle group to update the input parameters of the reconstructed prediction model, the updated input parameters are input into the prediction model, and the output of the prediction model is the predicted output of the input pollutant concentration.

[0122] The specific process of updating and iterating the particle swarm in step S2083 is as follows: k is the current iteration number of the particle swarm, and the current position of the i-th particle after the k-th iteration is recorded as , , the current velocity of the i-th particle after the k-th iteration is recorded as , ; The particle updates its velocity and position according to the following formulas (32) and (33): (32) (33) In the above formulas (32) and (33), is the inertia weight; the maximum , is the velocity of the particle; , is a non-negative constant, called the acceleration factor; , is a random number distributed between [0,1]; ; ; represents the minimum velocity of the ith particle, Represents the maximum velocity of the ith particle.

[0123] In order to solve the above problems, the embodiment of the present invention also provides a device for predicting the concentration of air pollutants, such as Figure 3 As shown, the air pollutant concentration prediction device 300 mainly includes:

[0124] The data acquisition module 301 is used to acquire the current pollutant concentration data, current meteorological data and the geographical location data of the target site corresponding to the target environmental monitoring site.

[0125] By simultaneously collecting pollutant concentrations, meteorological conditions (such as temperature, humidity, wind speed, etc.) and geographic information (such as site location), the integrity and diversity of the input data are ensured, providing a richer information basis for subsequent feature extraction and prediction.

[0126] In some of the embodiments, when the air pollutant concentration prediction device provided by the embodiments of the present invention is applied to predict the air pollutant concentration in a target area, the target environmental monitoring site is a plurality of environmental monitoring sites within the target area; when the prediction device is applied to predict the air pollutant concentration at a single or multiple designated environmental monitoring sites, the target environmental monitoring site is a single or multiple designated environmental monitoring sites.

[0127] The prediction model adopted by the prediction device for air pollutant concentration provided by the embodiment of the present invention has wide applicability and high prediction accuracy, can meet diverse prediction needs, and ensures the accuracy and comprehensiveness of the prediction results by comprehensively utilizing data of different scales and types.

[0128] The model processing module 302 is used to use the current pollutant concentration data, the current meteorological data and the geographic location data of the target site as the target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data (time series features) from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results.

[0129] In some of the embodiments, the above-mentioned air pollutant concentration prediction device 300 also includes a prediction model training module, which is used to: obtain historical pollutant concentration data, historical meteorological data and station geographical location data corresponding to multiple environmental monitoring stations respectively; use a variational mode decomposition algorithm to perform variational mode decomposition on the historical pollutant concentration data of each station through the input module of the prediction model, use a particle swarm optimization algorithm to optimize the variational mode decomposition results to obtain multiple modal data corresponding to each station, and perform data normalization on the multiple modal data; extract time feature data from the multiple modal data corresponding to each station respectively through the time convolution network in the time feature extraction module; and combine the historical pollutant concentration data, historical meteorological data and station location data of each station into a single unit; The geographic location data is input into the spatial feature extraction module to construct the adjacency matrix in the graph convolutional network, and the spatial feature data is extracted through the graph convolutional network; the temporal feature data and the spatial feature data are input into the convolutional gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data with the spatial feature data, and introduce the attention mechanism to obtain the spatiotemporal features corresponding to different modalities; wherein, the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; the predicted outputs corresponding to all modalities are aggregated through the output module to obtain the predicted concentration output feature data, and the predicted concentration output feature data is denormalized to obtain the prediction result.

[0130] Through the above prediction model training module, multiple modal data are finely extracted and optimized through variational mode decomposition and particle swarm optimization algorithms, which enhances the robustness and accuracy of the model. The long-term dependencies in the time series are efficiently captured by the time convolution network, which enhances the prediction ability of short-term and long-term trends and improves the accuracy of the prediction. The spatial correlation between different sites is fully considered through the graph convolution network, which improves the understanding of the overall pollution status of the region and the accuracy of the prediction results. Through the convolutional gated recurrent unit and attention mechanism, feature fusion in the time and space dimensions is achieved, which further improves the performance of the prediction model and adapts to more diverse application scenarios. Through the fully connected layer and anti-normalization processing, the comprehensiveness and practical application value of the prediction results are ensured, and the practicality and reliability of the model are enhanced. The technical effect of significantly improving the comprehensiveness and accuracy of the prediction of air pollutant concentrations and enhancing the robustness and practicality of the prediction model is achieved.

[0131] In some of the embodiments, the air pollutant concentration prediction device 300 further includes a data processing module. After the steps of obtaining historical pollutant concentration data, historical meteorological data and station geographic location data respectively corresponding to a plurality of environmental monitoring stations, the data processing module is used to: pre-process the historical pollutant concentration data and historical meteorological data respectively corresponding to the plurality of environmental monitoring stations through an input module; wherein the pre-processing includes missing value processing and outlier processing; calculating the mutual information values ​​of the historical pollutant concentration data and the historical meteorological data, sorting the mutual information values, and selecting the historical pollutant concentration types and meteorological data types that meet the target quantity.

[0132] Among them, the mutual information value provided by the embodiment of the present invention can quantify the correlation between different features and help identify the features that have the greatest influence on the target variable (air pollutant concentration). Then, by sorting the mutual information values, you can intuitively understand which features have the greatest impact on the prediction results, which improves the interpretability of the model. Then, according to the mutual information value sorting, select the historical pollutant concentration type and meteorological data type that meet the target quantity. The ability to select the most influential features reduces the data dimension, reduces the computational complexity, and improves the efficiency of model training. And while retaining the most useful information for prediction and removing redundant features, it helps to improve the accuracy and stability of the prediction.

[0133] In some of the embodiments, the above-mentioned prediction model also includes a Mogrifier gating operation module. Before the step of extracting time feature data from multiple modal data corresponding to each site through the time convolution network in the time feature extraction module, the above-mentioned prediction model training module is also used to: perform data optimization on the multiple modal data corresponding to each site through the Mogrifier gating operation module.

[0134] Based on the above settings, the Mogrifier gating operation can dynamically select and retain the most useful information for prediction, avoiding the important information that may be lost in traditional preprocessing methods, and filtering out unimportant or redundant data through the gating mechanism, reducing the burden of subsequent processing and improving computational efficiency.

[0135] In some of the embodiments, the air pollutant concentration prediction device 300 further includes a prediction model optimization module for searching for the optimal parameter configuration of the prediction model using a particle swarm optimization algorithm to optimize the prediction model.

[0136] By adopting the particle swarm optimization algorithm (PSO) to optimize the prediction model, the present invention not only significantly improves the optimization level of the model parameter configuration, but also significantly improves the overall performance of the prediction model by means of global optimal search, precise parameter configuration, improved model performance and accelerated convergence.

[0137] In some of the embodiments, the optimization module of the above-mentioned prediction model is also used to: determine the loss function according to the predicted output result, the actual output result and the predicted time step value; determine the initialization parameters and particle swarm parameters of the prediction model; wherein the particle swarm parameters include the particle swarm size, the maximum number of iterations, the inertia weight, the acceleration factor, the initial speed, and the initial position; iteratively update the particle swarm to obtain the speed and position of each particle after each iterative update, the individual extreme value of each particle, and the global extreme value of the particle swarm; update the output parameters of the prediction model according to the position of the particle corresponding to the final global extreme value of the particle swarm obtained when the maximum number of iterations is met, so as to complete the optimization of the prediction model.

[0138] Through the above settings, the difference between the predicted results and the actual results can be accurately quantified based on the loss function, providing a clear goal for model optimization; by introducing the prediction time step value, the loss function can dynamically reflect the prediction error in different time periods, ensuring that the optimization process takes into account the influence of the time dimension. The setting of initialization parameters and particle swarm parameters improves the convergence speed and stability of the optimization process. Then, through multiple iterative updates, the particle swarm gradually approaches the optimal solution, ensuring the gradualness and effectiveness of the optimization process and avoiding falling into the local optimal solution. Finally, by updating the output parameters of the prediction model, the model performance is ensured to reach the optimal state, which significantly improves the prediction accuracy and robustness and enhances the generalization ability.

[0139] The prediction result determination module 303 is used to determine the prediction result of the air pollutant concentration corresponding to the target environmental monitoring site according to the output result of the prediction model.

[0140] Through the processing of input data by the above-mentioned prediction model, multiple factors such as pollutant concentration, meteorological conditions and geographical location are comprehensively considered, providing prediction results with high prediction accuracy, high reliability and high timeliness. The obtained prediction results have high practical application value and can be used as data support for environmental protection and public health management.

[0141] The above-mentioned air pollutant concentration prediction device provided by the embodiment of the present invention adopts a data acquisition module for obtaining the current pollutant concentration data, current meteorological data and target site geographical location data corresponding to the target environmental monitoring site; a model processing module is used to use the current pollutant concentration data, current meteorological data and target site geographical location data as target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network The module is used to extract the spatiotemporal features corresponding to each mode; the output module is used to aggregate the prediction outputs corresponding to multiple modes to output the prediction results; the prediction result determination module is used to determine the prediction results of the air pollutant concentrations corresponding to the target environmental monitoring site according to the output results of the prediction model. Through the modal decomposition and feature extraction in the prediction model, it is ensured that key information is retained in the prediction process and the robustness of the model is enhanced. The spatiotemporal feature fusion module based on the prediction model enables the prediction method to adapt to more diverse application scenarios and improves its versatility and flexibility. Finally, by aggregating the prediction outputs of multiple modes, the deviation that may be caused by single modal prediction is reduced, and the stability and reliability of the final prediction results are improved. Through the prediction model, multimodal data fusion, time feature extraction, spatial feature extraction and spatiotemporal feature fusion are performed, which significantly improves the comprehensiveness and accuracy of the prediction results and expands the technical effect of the applicable scenarios of the prediction method.

[0142] An embodiment of the present invention further provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present invention.

[0143] The embodiment of the present invention further provides a computer program product, including a computer program, wherein the computer program is used to make the computer perform the method of the embodiment of the present invention when executed by a processor of the computer. The computer program product should be understood as a software product that mainly implements the above method of the present application through a computer program.

[0144] An embodiment of the present invention further provides an electronic device, comprising: at least one processor; and a memory in communication with the at least one processor. The memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the electronic device executes the method of the embodiment of the present invention.

[0145] refer to Figure 4 , a block diagram of an electronic device that can be used as a server or client of an embodiment of the present invention will now be described, which is an example of a hardware device that can be applied to various aspects of the present invention. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0146] like Figure 4 As shown, the electronic device includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory ROM402 or a computer program loaded from a storage unit 408 into a random access memory RAM403. In RAM403, various programs and data required for the operation of the electronic device can also be stored. The computing unit 401, ROM402 and RAM403 are connected to each other via a bus 404. An input / output I / O interface 405 is also connected to the bus 404.

[0147] Multiple components in the electronic device are connected to the I / O interface 405, including: an input unit 406, an output unit 407, a storage unit 408, and a communication unit 409. The input unit 406 can be any type of device that can input information to the electronic device, and the input unit 406 can receive input digital or character information, and generate key signal input related to user settings and / or function control of the electronic device. The output unit 407 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 408 can include but is not limited to a disk, an optical disk. The communication unit 409 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, and / or a wireless communication transceiver, such as a Bluetooth device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0148] The computing unit 401 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a CPU, a graphics processing unit (GPU), various special artificial intelligence (AI) computing units, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 401 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present invention may be implemented as a computer program, which is tangibly contained in a machine-readable medium, such as a storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via ROM 402 and / or a communication unit 409. In some embodiments, the computing unit 401 may be configured to perform the above-described method in any other appropriate manner (e.g., by means of firmware).

[0149] The computer programs for implementing the methods of the embodiments of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer programs are executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer programs may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0150] In the context of an embodiment of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, or infrared systems, devices, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0151] It should be noted that the term "including" and its variations used in the embodiments of the present invention are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless the context clearly indicates otherwise, it should be understood as "one or more".

[0152] The information / data involved in the embodiments of the present invention (including but not limited to information / data used for analysis, information / data stored, information / data displayed, etc.) are all information / data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant information / data must comply with relevant laws, regulations and standards of relevant countries and regions, and corresponding operation entrances are provided for users to choose to authorize or refuse.

[0153] The various steps described in the method implementation methods provided in the embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method implementation methods may include additional steps and / or omit the steps shown. The scope of protection of the present invention is not limited in this respect.

[0154] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0155] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the attached claims.

Claims

1. A method for predicting air pollutant concentration, characterized in that: include: Obtain the current pollutant concentration data, current meteorological data and geographic location data of the target environmental monitoring site; The current pollutant concentration data, the current meteorological data and the target site geographic location data are used as target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from the multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results; The prediction result of the air pollutant concentration corresponding to the target environmental monitoring site is determined according to the output result of the prediction model.

2. The method for predicting air pollutant concentration according to claim 1, characterized in that: The method also includes a step of training a prediction model: Obtain historical pollutant concentration data, historical meteorological data and station geographic location data corresponding to multiple environmental monitoring stations; The input module of the prediction model uses a variational mode decomposition algorithm to perform variational mode decomposition on the historical pollutant concentration data of each station, uses a particle swarm optimization algorithm to optimize the variational mode decomposition results, obtains multiple mode data corresponding to each station, and performs data normalization on the multiple mode data; Extracting time feature data from a plurality of modal data corresponding to each site respectively through a time convolution network in the time feature extraction module; Inputting historical pollutant concentration data, historical meteorological data and station geographic location data of each station into the spatial feature extraction module, constructing an adjacency matrix in a graph convolutional network, and extracting spatial feature data through the graph convolutional network; The temporal feature data and the spatial feature data are input into the convolution gated recurrent unit in the spatiotemporal fusion convolutional network module to fuse the temporal feature data with the spatial feature data, and an attention mechanism is introduced to obtain spatiotemporal features corresponding to different modalities; wherein the fully connected layer of the spatiotemporal fusion convolutional network module respectively connects the spatiotemporal features of a single modality, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; The predicted outputs corresponding to all modes are aggregated through the output module to obtain predicted concentration output characteristic data, and the predicted concentration output characteristic data are denormalized to obtain a prediction result.

3. The method for predicting air pollutant concentration according to claim 2, characterized in that: After the step of obtaining historical pollutant concentration data, historical meteorological data, and station geographic location data corresponding to a plurality of environmental monitoring stations, the method further includes: Preprocessing the historical pollutant concentration data and historical meteorological data corresponding to the plurality of environmental monitoring sites respectively obtained by the input module; wherein the preprocessing includes missing value processing and outlier processing; The mutual information values ​​of the historical pollutant concentration data and the historical meteorological data are calculated, the mutual information values ​​are sorted, and the historical pollutant concentration types and meteorological data types that meet the target quantity are selected.

4. The method for predicting air pollutant concentration according to claim 3, characterized in that: The prediction model further includes a Mogrifier gating operation module. Before the step of extracting time feature data from a plurality of modal data corresponding to each site respectively through the time convolution network in the time feature extraction module, the training step of the prediction model further includes: The Mogrifier gating operation module is used to optimize the multiple modal data corresponding to each site.

5. The method for predicting air pollutant concentration according to claim 2, characterized in that: The method also includes the step of optimizing the prediction model: A particle swarm optimization algorithm is used to search for the optimal parameter configuration of the prediction model to optimize the prediction model.

6. A method for predicting air pollutant concentration according to claim 5, characterized in that: The optimization step of the prediction model also includes: Determine the loss function based on the predicted output result, the actual output result and the predicted time step value; Determine the initialization parameters and particle swarm parameters of the prediction model; wherein the particle swarm parameters include particle swarm size, maximum number of iterations, inertia weight, acceleration factor, initial velocity, and initial position; Iteratively update the particle swarm to obtain the speed and position of each particle, the individual extreme value of each particle, and the global extreme value of the particle swarm after each iterative update; The output parameters of the prediction model are updated according to the position of the particle corresponding to the final global extreme value of the particle swarm obtained when the maximum number of iterations is met, so as to complete the optimization of the prediction model.

7. The method for predicting air pollutant concentration according to claim 1, characterized in that: When the method is applied to predict the concentration of air pollutants in a target area, the target environmental monitoring site is a plurality of environmental monitoring sites in the target area; When the method is applied to predict the concentration of air pollutants at a single or multiple designated environmental monitoring sites, the target environmental monitoring site is the single or multiple designated environmental monitoring sites.

8. A device for predicting air pollutant concentration, characterized in that: include: A data acquisition module is used to obtain the current pollutant concentration data, current meteorological data and geographical location data of the target environmental monitoring site; A model processing module, used to use the current pollutant concentration data, the current meteorological data and the target site geographic location data as target input data of the prediction model; wherein the prediction model includes an input module, a time feature extraction module, a spatial feature extraction module, a time-space fusion convolutional network module and an output module; the input module is used to perform modal decomposition processing on the pollutant concentration data in the input data to obtain multiple modal data; the time feature extraction module is used to extract time feature data from the multiple modal data; the spatial feature extraction module is used to construct an adjacency matrix that integrates the pollutant concentration data, meteorological data and site location data in the input data, and extracts spatial feature data by a graph convolutional network; the time-space fusion convolutional network module is used to extract the time-space features corresponding to each modality; the output module is used to aggregate the prediction outputs corresponding to multiple modalities to output the prediction results; The prediction result determination module is used to determine the prediction result of the air pollutant concentration corresponding to the target environmental monitoring site according to the output result of the prediction model.

9. The device for predicting air pollutant concentration according to claim 8, characterized in that: The device also includes a prediction model training module, which is used to: obtain historical pollutant concentration data, historical meteorological data and station geographical location data corresponding to multiple environmental monitoring stations respectively; use a variational mode decomposition algorithm to perform variational mode decomposition on the historical pollutant concentration data of each station through the input module of the prediction model, use a particle swarm optimization algorithm to optimize the variational mode decomposition results to obtain multiple modal data corresponding to each station, and perform data normalization on the multiple modal data; extract time feature data from the multiple modal data corresponding to each station respectively through the time convolution network in the time feature extraction module; The historical pollutant concentration data, historical meteorological data and station geographic location data of each station are input into the spatial feature extraction module to construct an adjacency matrix in the graph convolutional network, and the spatial feature data is extracted through the graph convolutional network; the temporal feature data and the spatial feature data are input into the convolutional gated recurrent unit in the spatiotemporal convolutional network module to fuse the temporal feature data with the spatial feature data, and an attention mechanism is introduced to obtain the spatiotemporal features corresponding to different modalities; wherein the fully connected layer of the spatiotemporal convolutional network module connects the spatiotemporal features of a single modality respectively, and outputs the predicted output of the air pollutant concentration corresponding to each modality of each station; the predicted outputs corresponding to all modalities are aggregated through the output module to obtain predicted concentration output feature data, and the predicted concentration output feature data is denormalized to obtain a prediction result.

10. An electronic device comprising: A processor, and a memory storing a program, wherein the program comprises instructions, which, when executed by the processor, cause the processor to perform the method for predicting the concentration of air pollutants according to any one of claims 1 to 7.

11. A non-transitory machine-readable medium storing computer instructions, characterized in that: The computer instructions are used to enable the computer to execute the method for predicting the air pollutant concentration according to any one of claims 1-7.

Images