PM2.5 chemical component vertical profile inversion system and method based on graph neural network and adaptive meta learning

Abstract

The invention discloses a PM2.5 chemical component vertical profile inversion system and method based on a graph neural network and adaptive meta-learning, and the system achieves the quick migration and high-precision inversion in a data scarce region through the construction of a dynamic space graph of a vertical height layer in combination with a meta-learning frame. The system utilizes the heterogeneous graph neural network to capture the component diffusion rule, reduces the annotation data demand through MAML optimization, solves the problems that a traditional model is insufficient in generalization ability, depends on a large amount of data and the like, and can be widely applied to air quality monitoring in a complex geographical environment.

Description

Technical Field

[0001] The present invention relates to the field of environmental monitoring, and in particular to a PM2.5 chemical component vertical profile inversion system and method based on graph neural network and adaptive meta-learning. Background Art

[0002] With the increasing severity of air pollution, especially the more significant impact of PM2.5 (fine particulate matter) on human health and the environment, accurate monitoring and forecasting of PM2.5 has become one of the key research topics in the field of environmental monitoring. In recent years, the application of deep learning technology in air quality prediction and inversion has made significant progress. Traditional models, such as convolutional neural networks (CNN) and long short-term memory networks (LSTM), have been widely used in the prediction of spatiotemporal distribution and concentration inversion of PM2.5. However, these traditional models have some limitations: Traditional deep learning models usually focus on processing two-dimensional plane data and fail to effectively capture the diffusion and transmission process of PM2.5 chemical components in the vertical direction of the atmosphere. In particular, the propagation of pollutants between vertical height layers is affected by meteorological conditions, topography, and atmospheric structure, but traditional models lack effective spatial modeling capabilities, resulting in low physical consistency of inversion results.

[0003] At the same time, existing deep learning models usually rely on a large amount of labeled data for training, and most of their training data comes from specific geographical areas. However, due to differences in geographical environments (such as plateaus and coastal areas) and climatic conditions, these models have weak generalization capabilities and are difficult to meet the needs of pollution prediction in different regions and environments. In addition, in scenarios where data is scarce, the performance of existing models is often unsatisfactory.

[0004] In sudden pollution events, the temporal and spatial evolution of pollutants changes dramatically, and traditional static deep learning models are often unable to respond and adjust quickly to adapt to this rapidly changing environment. This makes the existing models less capable of accurate prediction and response in a short period of time during sudden pollution events.

[0005] Therefore, how to design a model that can better capture the vertical diffusion process of PM2.5 and improve the generalization ability and adaptability of the model is a key issue that needs to be urgently solved in current PM2.5 monitoring and forecasting technology. Summary of the invention

[0006] The purpose of the present invention is to provide a PM2.5 chemical component vertical profile inversion system and method based on graph neural network and adaptive meta-learning, which solves the above-mentioned technical problems pointed out in the prior art.

[0007] The present invention provides a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning, including a multi-source data acquisition module, a graph structure construction module, an adaptive meta-learning module and an inversion output module; The multi-source data acquisition module is used to collect and acquire multi-source data on k vertical height layers of the current geographical area; pre-process the multi-source data to obtain pre-processed multi-source data; The graph structure component module is used to construct a graph structure by taking each of the vertical height layers as a node; and to calculate the edge weight between the i-th node and the j-th node in the graph structure through the meteorological variable data collected in a preset time period. ; The adaptive meta-learning module is used to Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; The inversion output module is used to process the preprocessed multi-source data of the current area at the current moment through the target adaptive meta-learning model, and output the inversion results of the predicted PM2.5 chemical component vertical profiles.

[0008] Preferably, the multi-source data includes ground-based lidar optical parameters, meteorological variable data and local component concentration data.

[0009] Preferably, the ground-based laser radar optical parameters include vertical profile data of 532nm aerosol backscattering coefficient, extinction coefficient and depolarization ratio; The meteorological variable data include the integrated ERA5 reanalysis data set; The local component concentration data include real-time monitoring values ​​of near-ground PM2.5 chemical component concentrations.

[0010] Preferably, the edge weight is calculated as follows: in, is the wind speed eigenvector of the ith node, is the wind speed eigenvector of the jth node, is the bandwidth parameter, which controls the weight decay speed.

[0011] Preferably, the adaptive meta-learning module is specifically used to construct and train an initial adaptive meta-learning model by using sample multi-source data collected from multiple regions; Based on the edge weight Updating the initial adaptive meta-learning model to obtain a first adaptive meta-learning model; Based on the multi-source data of the current area collected in n consecutive time periods, the parameters of the first adaptive meta-learning model are updated through the Adam optimizer to obtain the target adaptive meta-learning model.

[0012] Preferably, the adaptive meta-learning module is further used to collect and obtain sample multi-source data from multiple regions, wherein the sample multi-source data includes multi-source data from various regions monitored for one year; Preprocessing the sample multi-source data of each region respectively and splitting them into a support set and a query set; Construct a first initial adaptive meta-learning model; use the support set and query set corresponding to each region to train the first initial adaptive meta-learning model and calculate the loss function, and when the loss function is less than or equal to the minimum loss function threshold, the model converges, and the initial parameters θ of the first initial adaptive meta-learning model corresponding to each region are obtained; Based on the initial parameters θ, the initial parameters θ of the first initial adaptive meta-learning model are updated through double-layer optimization to obtain the target model parameters ; Preferably, the loss function is calculated as follows: ; In the formula, is the importance weight of each component concentration; is the predicted sample local component concentration data of the i-th sample multi-source data; is the sample local component concentration data of the i-th sample multi-source data; The target model parameters are calculated as follows: ; in, ; θ is the initial parameter of the first initial adaptive meta-learning model, is the inner layer learning rate, is the loss gradient, where Represents the training missions for each region, For the first initial adaptive meta-learning model, The loss dependence ratio.

[0013] Preferably, the adaptive meta-learning module is further used in the specific implementation process to calculate the edge weights of the graph structure. Get the k-th layer edge weight of the k-th vertical height layer ; Based on the k-th layer edge weight The neighbor node set of the i-th vertical height layer is iteratively updated and calculated by the GraphSAGE algorithm to obtain the embedding value of the i-th node in the k-th vertical height layer; The initial adaptive meta-learning model is updated based on the node embedding value to obtain a first adaptive meta-learning model.

[0014] Preferably, the node embedding value is calculated as follows: ; In the formula, is the edge weight of the kth layer; is the set of neighbor nodes; The embedding of node i at layer k-1, Indicates the jth node from the neighbor node set; is an aggregate function, is the connection function, is a non-linear activation function.

[0015] Accordingly, the present invention also proposes a PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning, comprising the following steps: Collect and obtain multi-source data on k vertical height layers of the current geographical area; pre-process the multi-source data to obtain pre-processed multi-source data; Each of the vertical height layers is used as a node to construct a graph structure; and the edge weight between the i-th node and the j-th node in the graph structure is calculated through the meteorological variable data collected in a preset time period. ; Based on the edge weight Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; The preprocessed multi-source data of the current area at the current moment is processed by the target adaptive meta-learning model, and the inversion result of the vertical profile of the predicted PM2.5 chemical components is output.

[0016] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages: From the analysis of the above-mentioned PM2.5 chemical component vertical profile inversion system and method based on graph neural network and adaptive meta-learning provided by the present invention, it can be known that in specific applications, firstly, 60 layers of data with a layer of 100m in the vertical range of 0-6km are obtained through ground-based lidar (optical parameters), meteorological monitoring stations (ERA5 data set) and ground monitoring stations (PM2.5 component concentrations), covering aerosol characteristics, meteorological conditions and near-ground component concentrations, providing multi-dimensional input for the model; then, by using 60 vertical height layers as nodes, a spatial topological network is constructed to explicitly express the correlation between different height layers (such as the vertical diffusion of aerosols is affected by the meteorological conditions of the upper and lower layers), thereby enhancing the model's ability to characterize physical processes; further, by utilizing Sample data from different geographical environments (such as cities, suburbs, and different climate zones) are used to learn common cross-regional characteristics (such as meteorological-component correlation patterns), improve model generalization, and quickly adapt to new regions with a small amount of labeled data through meta-learning processing, thereby solving the performance degradation problem of traditional methods caused by scarcity of regional data and reducing dependence on a large amount of labeled data in a single region. Finally, the real-time pre-processed data (optical parameters, meteorology, and near-ground components) of the current region are input into the model, and the features of nodes at different altitudes are aggregated through the graph neural network (GNN) (such as the spatial correlation between aerosol optical properties and wind speed), and the model parameters are dynamically adjusted in combination with meta-learning, and finally the 60-layer PM2.5 chemical component concentrations (such as sulfate, nitrate, etc.) are output through the fully connected layer mapping. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of the overall architecture of a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning; Figure 2 A schematic diagram of the main process of a PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning; Figure 3 A schematic diagram of the CORR comparison curve simulation after fine-tuning in different regions in a PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning; Figure 4 This is a schematic diagram of the resource usage comparison simulation of model training and inference in a PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning.

[0018] Reference numerals: multi-source data acquisition module 10 , graph structure construction module 20 , adaptive meta-learning module 30 , inversion output module 40 . DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0021] Embodiment 1

[0022] like Figure 1 As shown, the first embodiment of the present invention provides a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning, including a multi-source data acquisition module 10, a graph structure construction module 20, an adaptive meta-learning module 30 and an inversion output module 40; The multi-source data acquisition module 10 is used to collect and acquire multi-source data on k vertical height layers of the current geographical area; pre-process the multi-source data to obtain pre-processed multi-source data; The graph structure component module 20 is used to construct a graph structure by taking each of the vertical height layers as a node; and to calculate the edge weight between the i-th node and the j-th node in the graph structure through the meteorological variable data collected in a preset time period. ; The adaptive meta-learning module 30 is used to Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; The inversion output module 40 is used to process the preprocessed multi-source data of the current area at the current moment through the target adaptive meta-learning model, and output the inversion result of the predicted PM2.5 chemical component vertical profile.

[0023] The multi-source data include ground-based lidar optical parameters, meteorological variable data and local component concentration data.

[0024] The ground-based lidar optical parameters include vertical profile data of 532nm aerosol backscattering coefficient, extinction coefficient and depolarization ratio; The meteorological variable data include the integrated ERA5 reanalysis data set; The local component concentration data include real-time monitoring values ​​of near-ground PM2.5 chemical component concentrations.

[0025] The edge weight is calculated as follows: in, is the wind speed eigenvector of the ith node, is the wind speed eigenvector of the jth node, Bandwidth parameter, which controls the weight decay speed.

[0026] The adaptive meta-learning module 30 is specifically used to construct and train an initial adaptive meta-learning model using sample multi-source data collected from multiple regions; Based on the edge weight Updating the initial adaptive meta-learning model to obtain a first adaptive meta-learning model; Based on the multi-source data of the current area collected in n consecutive time periods, the parameters of the first adaptive meta-learning model are updated through the Adam optimizer to obtain the target adaptive meta-learning model.

[0027] The adaptive meta-learning module 30 is further used to collect and obtain sample multi-source data from multiple regions, wherein the sample multi-source data includes multi-source data from various regions monitored for one year; Preprocessing the sample multi-source data of each region respectively and splitting them into a support set and a query set; Construct a first initial adaptive meta-learning model; use the support set and query set corresponding to each region to train the first initial adaptive meta-learning model and calculate the loss function, and when the loss function is less than or equal to the minimum loss function threshold, the model converges, and the initial parameters θ of the first initial adaptive meta-learning model corresponding to each region are obtained; Based on the initial parameters θ, the initial parameters θ of the first initial adaptive meta-learning model are updated through double-layer optimization to obtain the target model parameters ; The loss function is calculated as follows: In the formula, is the importance weight of each component concentration; is the predicted sample local component concentration data of the i-th sample multi-source data; is the sample local component concentration data of the i-th sample multi-source data; The target model parameters are calculated as follows: ; in, ; θ is the initial parameter of the first initial adaptive meta-learning model, is the inner layer learning rate, is the loss gradient, where Represents the training missions for each region, For the first initial adaptive meta-learning model, The loss dependence ratio.

[0028] The adaptive meta-learning module 30 is also used in the specific implementation process to calculate the edge weights of the graph structure. Get the k-th layer edge weight of the k-th vertical height layer ; Based on the k-th layer edge weight The neighbor node set of the i-th vertical height layer is iteratively updated and calculated by the GraphSAGE algorithm to obtain the embedding value of the i-th node in the k-th vertical height layer; The initial adaptive meta-learning model is updated based on the node embedding value to obtain a first adaptive meta-learning model.

[0029] The node embedding value is calculated as follows: ; In the formula, is the edge weight of the kth layer; is the set of neighbor nodes; The embedding of node i at layer k-1, Indicates the jth node from the neighbor node set; is an aggregate function, is the connection function, is a non-linear activation function.

[0030] In summary, the first embodiment of the present application provides a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning, which constructs a graph structure by utilizing multi-source data of multiple vertical height layers in the current area, models the spatial topological relationship of the vertical height layers according to the graph structure, and improves the physical rationality of the inversion; trains a pre-constructed lightweight model framework (i.e., an adaptive meta-learning model) through multi-source data of samples from multiple different regions, reduces dependence on labeled data, supports rapid migration of a small number of samples, and thereby realizes dynamic adaptive optimization of the model in multiple geographical environments.

[0031] Embodiment 2 like Figure 2 As shown, the first embodiment of the present invention provides a PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning, including the following operating steps: Step S10: collecting and acquiring multi-source data on k vertical height layers of the current geographical area; preprocessing the multi-source data to obtain preprocessed multi-source data; The multi-source data include ground-based laser radar optical parameters, meteorological variable data and local component concentration data; The ground-based lidar optical parameters include vertical profile data of 532nm aerosol backscattering coefficient, extinction coefficient and depolarization ratio; The meteorological variable data include the integrated ERA5 reanalysis data set (the integrated ERA5 reanalysis data set includes parameters such as temperature, humidity, wind speed, vertical speed, etc.); The local component concentration data include real-time monitoring values ​​of PM2.5 chemical component concentrations near the ground; Specifically, the multi-source data is collected and acquired by (using ground-based laser radar, meteorological monitoring stations and ground monitoring stations) collecting multi-source data of 60 vertical spaces at a vertical height of 6 km with each layer of 100 meters at intervals (the time period is 1 hour); The number of the above-mentioned vertical height layers is 60; the vertical height of each of the above-mentioned vertical height layers is 100m; The preprocessing refers to cleaning the multi-source data, using Hampel filtering to remove abnormal values ​​from the lidar data (window width 12 hours, threshold 3 times the median absolute deviation); and filling missing values ​​in the meteorological data through linear interpolation; It should be noted that the above-mentioned embodiment of the present application collects multi-source data at a vertical height of 0-6 km, wherein the vertical height of 0-6 km is divided into 60 vertical height layers with each layer of 100 m; in each vertical height layer, multi-source data information is included, namely, ground-based laser radar optical parameters collected by ground-based laser radar, meteorological variable data collected by meteorological monitoring stations, and local component concentration data collected by ground monitoring stations; providing a data basis for subsequent PM2.5 monitoring and forecasting; Step S20: constructing a graph structure by taking each of the vertical height layers as a node; and calculating the edge weight between the i-th node and the j-th node in the graph structure by using the meteorological variable data collected in a preset time period. ; The edge weight is calculated as follows: in, is the wind speed eigenvector of the ith node, is the wind speed eigenvector of the jth node, is the bandwidth parameter, which controls the weight decay speed; The above-mentioned embodiment of the application improves the physical rationality of the inversion by modeling the spatial topological relationship of vertical height layers through a graph structure; Step S30: Based on the edge weight Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; It should be noted that the above-mentioned embodiments of the present application reduce the dependence on labeled data by constructing a lightweight model framework, support the rapid migration of a small number of samples, and jointly construct and train the target adaptive meta-learning model through multi-source data of samples from multiple regions, so as to realize dynamic adaptive optimization of the model in multiple geographical environments.

[0032] Step S40: Processing the preprocessed multi-source data of the current region at the current moment through the target adaptive meta-learning model, and outputting the inversion result of the vertical profile of the predicted PM2.5 chemical components; It should be noted that the above-mentioned processing of the preprocessed multi-source data of the current area at the current moment through the target adaptive meta-learning model to output the predicted PM2.5 chemical component vertical profile inversion result is obtained by inputting the preprocessed multi-source data of the current area at the current moment into the target adaptive meta-learning model, and then, after calculation and processing by the target adaptive meta-learning model, mapping is performed through the fully connected layer of the target adaptive meta-learning model to obtain the concentration value of each chemical component (the concentration value of each chemical component is the PM2.5 chemical component vertical profile inversion result); The above-mentioned embodiment of the present application first obtains 60 layers of data every 100m in the vertical range of 0-6km through ground-based lidar (optical parameters), meteorological monitoring stations (ERA5 data set) and ground monitoring stations (PM2.5 component concentration), covering aerosol characteristics, meteorological conditions and near-ground component concentrations, providing multi-dimensional input for the model; then, by using 60 vertical height layers as nodes, a spatial topological network is constructed to explicitly express the correlation between different height layers (such as the vertical diffusion of aerosols is affected by the meteorological conditions of the upper and lower layers), thereby enhancing the model's ability to characterize physical processes; further, by using sample data from different geographical environments (such as cities, suburbs, and different climatic zones), the model can be used to obtain a spatial topological network with a spatial topological network. ), learn common features across regions (such as meteorological-component correlation rules), improve model generalization, and quickly adapt to new regions with a small amount of labeled data through meta-learning processing, so as to solve the performance degradation problem of traditional methods caused by scarce regional data and reduce dependence on a large amount of labeled data in a single region; finally, the real-time pre-processed data (optical parameters, meteorology, near-ground components) of the current region are input into the model, and the features of nodes at different altitudes are aggregated through the graph neural network (GNN) (such as the spatial correlation between aerosol optical properties and wind speed), and the model parameters are dynamically adjusted in combination with meta-learning, and finally the PM2.5 chemical component concentrations (such as sulfate, nitrate, etc.) of 60 layers are output through the fully connected layer mapping.

[0033] Specifically, in step S30, based on the edge weight Combining the sample multi-source data collected from multiple regions to build and train the target adaptive meta-learning model includes the following steps: Step S31: constructing and training an initial adaptive meta-learning model by using sample multi-source data collected from multiple regions; It should be noted that the above-mentioned multiple regions refer to different geographical areas (such as cities and plateaus); Step S32: Based on the edge weight Updating the initial adaptive meta-learning model to obtain a first adaptive meta-learning model; Step S33: Based on the multi-source data of the current area collected for n consecutive time periods, the parameters of the first adaptive meta-learning model are updated through the Adam optimizer to obtain the target adaptive meta-learning model.

[0034] It should be noted that the above-mentioned embodiment of the present application jointly trains the model by utilizing heterogeneous data (different climates, pollution sources, and terrains) from multiple regions; further, through the meta-learning framework, the model is equipped with "meta-knowledge" that can quickly adapt to new tasks, laying the foundation for subsequent dynamic optimization and reducing dependence on annotated data from a single region; further, by introducing the edge weights calculated in step S20 (reflecting the similarity of wind speeds between different altitude layers), the physical correlation of vertical altitude layers (such as turbulent mixing and horizontal transport) is encoded into the model, and the model learning is constrained to conform to the spatial propagation laws of atmospheric dynamics. The model parameters are corrected through edge weights (such as the intensity of GNN inter-layer message transmission) to improve the vertical coherence and physical rationality of the inversion results; further, by utilizing real-time multi-source data for n consecutive time periods (such as 72 hours) in the current region (that is, using the multi-source data of the current region as annotated data to fine-tune the parameters of the first adaptive meta-learning model), the first model is lightweight fine-tuned through the Adam optimizer, so that the model can quickly adapt to the local unique environment (such as the emission characteristics of local pollution sources and the boundary layer structure differences caused by terrain).

[0035] Specifically, in step S31, an initial adaptive meta-learning model is constructed and trained by using sample multi-source data collected from multiple regions, including the following steps: Step S311: Collect and obtain sample multi-source data of multiple regions, wherein the sample multi-source data includes multi-source data of various regions monitored for one year; Step S312: pre-processing the sample multi-source data of each region and splitting them into a support set and a query set; Step S313: constructing a first initial adaptive meta-learning model; using the support set and query set corresponding to each region to respectively train the first initial adaptive meta-learning model and calculate the loss function, the model converges when the loss function is less than or equal to the minimum loss function threshold, and obtaining the initial parameters θ of the first initial adaptive meta-learning model corresponding to each region; The loss function is calculated as follows: ; In the formula, is the importance weight of each component concentration (e.g. black carbon has a higher weight); is the predicted sample local component concentration data of the i-th sample multi-source data; is the sample local component concentration data of the i-th sample multi-source data; Step S314: Based on the initial parameters θ, the initial parameters θ of the first initial adaptive meta-learning model are updated through double-layer optimization to obtain the target model parameters The target model parameters are calculated as follows: ; in, ; θ is the initial parameter of the first initial adaptive meta-learning model, is the inner layer learning rate, is the loss gradient, where represents the training tasks for each region (the training tasks are the training tasks for the first initial adaptive meta-learning model using each region respectively), For the first initial adaptive meta-learning model, The loss dependence ratio.

[0036] It should be noted that the above-mentioned embodiment of the present application collects multi-source data (ground-based lidar, meteorology, and ground component monitoring) from multiple regions for one year to ensure that the data covers different seasons, climate conditions (such as winter inversion, summer convection), and pollution events (such as dust and haze), thereby enhancing the model's adaptability to spatiotemporal heterogeneity, and capturing the periodic characteristics of the meteorological-pollution coupling relationship (such as diurnal boundary layer changes and monsoon transmission effects) through data throughout the year, providing the model with stable physical model prior knowledge; then, the data from each region is uniformly preprocessed (such as Hampel filtering and meteorological interpolation), Eliminate instrument errors or data format differences between regions to ensure the compatibility of cross-regional training; construct multiple "pseudo-tasks" (one task for each region) by splitting the support set and the query set to adapt to the training requirements of the meta-learning framework (such as MAML); through iterative training of the support set-query set of multiple regions, calculate the loss function, and gradually optimize the initial parameter θ, so that the model has preliminary cross-regional generalization capabilities; finally, through double-layer optimization, the model can retain the common laws across regions (such as the universal dynamic mechanism of vertical diffusion), and can quickly adapt to the local characteristics of the new region (such as the emission intensity of local pollution sources).

[0037] Specifically, in step S32, based on the edge weight Updating the initial adaptive meta-learning model to obtain a first adaptive meta-learning model includes the following steps: Step S321: Based on the edge weights of the graph structure Get the k-th layer edge weight of the k-th vertical height layer ; Step S322: Based on the edge weight of the kth layer The neighbor node set of the i-th vertical height layer is iteratively updated and calculated by the GraphSAGE algorithm to obtain the embedding value of the i-th node in the k-th vertical height layer; The node embedding value is calculated as follows: In the formula, is the edge weight of the kth layer; is the set of neighbor nodes, is the embedding of node i in the k-1th layer, Indicates the jth node from the neighbor node set; is an aggregate function, is the connection function, is a nonlinear activation function; Step S323: updating the initial adaptive meta-learning model based on the node embedding value to obtain a first adaptive meta-learning model.

[0038] It should be noted that the above-mentioned embodiment of the present application extracts the edge weight corresponding to the k-th vertical height layer based on the edge weight calculated in step S20 (reflecting the similarity of wind speeds between different height layers), and clarifies the interaction strength between this layer and the neighboring layer (such as strong mixing effect between high wind speed layers and relative independence between low wind speed layers); through the GraphSAGE algorithm, the neighbor set of the i-th node is weightedly aggregated using the edge weight of the k-th layer, and the characteristics of the neighboring layers (such as the meteorological conditions of the upper layer and the aerosol concentration of the lower layer) are integrated to simulate the physical interaction process in the vertical direction (such as the vertical diffusion of pollutants and the upward and downward transmission of meteorological elements); finally, the updated node embedding value (including vertical interaction information) is input into the initial adaptive meta-learning model, and the parameters of its graph neural network (GNN) part are adjusted to make the model more in line with the actual atmospheric dynamic process; through the embedding update guided by the edge weight, the model is constrained to learn the spatial propagation mode that conforms to the laws of meteorological dynamics (such as pollutants diffuse upward with rising air currents and are suppressed by sinking air currents and retained near the ground).

[0039] Based on the above embodiments of the present application, the researchers conducted field tests: By collecting PM2.5 chemical composition data from five typical regions in China (Beijing, Chengdu, Lhasa, Guangzhou, and Urumqi) from 2021 to 2023 as sample multi-source data; And use traditional models as comparison baselines: CNN-ATT-BiLSTM model and traditional WRF-Chem model; By calculating the accuracy: CORR (correlation coefficient), RMSE (μg / m³) as the analysis results; As shown in Table 1, the technical solution adopted in the embodiment of the present application is significantly better than the effect of the baseline technical solution; Table 1

[0040] like Figure 3 As shown, in terms of migration efficiency, compared with the baseline technical solution, the technical solution adopted in this application only requires 5% labeled data (36 hours) in the Urumqi area, and after fine-tuning for 1 hour, the CORR reaches 0.82, an improvement of 30% over the baseline.

[0041] like Figure 4 As shown in the figure, compared with the baseline technical solution, the computing resource consumption of the technical solution adopted in this application is: GPU memory usage is reduced by 45% during the training phase, and the inference speed is increased to real-time (<1 second / time) In summary, the technical solution adopted in the above-mentioned embodiment of the present application captures the diffusion and transmission laws of vertical height layers through a dynamic graph structure, and the physical consistency of the inversion result is improved by 25%; the meta-learning framework reduces the training data requirements of the model in data-scarce areas by 70%, and migration to new locations only requires 1 hour of fine-tuning; the local aggregation strategy of GraphSAGE reduces the computational complexity, which is 40% faster than traditional GCN.

[0042] In summary, the present invention proposes a PM2.5 chemical component vertical profile inversion system and method based on graph neural network and adaptive meta-learning. First, 60 layers of data are obtained every 100m in the vertical range of 0-6km through ground-based lidar, meteorological monitoring stations and ground monitoring stations, covering aerosol characteristics, meteorological conditions and near-ground component concentrations, providing multi-dimensional input for the model; then, by taking 60 vertical height layers as nodes, a spatial topological network is constructed to explicitly express the correlation between different height layers, thereby enhancing the model's ability to characterize physical processes; further, by utilizing sample data from different geographical environments, common features across regions are learned to improve the generalization of the model, and meta-learning processing is used to quickly adapt to new regions with a small amount of labeled data, thereby solving the performance degradation problem of traditional methods caused by scarcity of regional data and reducing dependence on a large amount of labeled data in a single region; finally, the real-time pre-processed data of the current region is input into the model, the features of nodes at different height layers are aggregated through the graph neural network, and the model parameters are dynamically adjusted in combination with meta-learning, and finally the PM2.5 chemical component concentrations of 60 layers are output through full-connection layer mapping; During the specific operation, sample data from different geographical areas are used to train the model with each area as a meta-task, and the model parameters are comprehensively trained and optimized. The heterogeneous graph neural network is used to capture the diffusion law of components, and MAML optimization is used to reduce the demand for labeled data, thus solving the problems of insufficient generalization ability of traditional models and dependence on large amounts of data.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace part or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning, characterized in that: It includes a multi-source data acquisition module, a graph structure construction module, an adaptive meta-learning module and an inversion output module; The multi-source data acquisition module is used to collect and acquire multi-source data on k vertical height layers of the current geographical area; pre-process the multi-source data to obtain pre-processed multi-source data; The graph structure component module is used to construct a graph structure by taking each of the vertical height layers as a node; and to calculate the edge weight between the i-th node and the j-th node in the graph structure through the meteorological variable data collected in a preset time period. ; The adaptive meta-learning module is used to Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; The inversion output module is used to process the preprocessed multi-source data of the current area at the current moment through the target adaptive meta-learning model, and output the inversion results of the predicted PM2.5 chemical component vertical profiles.

2. According to claim 1, a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning is characterized in that: The multi-source data include ground-based lidar optical parameters, meteorological variable data and local component concentration data.

3. According to claim 2, a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning is characterized in that: The ground-based lidar optical parameters include vertical profile data of 532nm aerosol backscattering coefficient, extinction coefficient and depolarization ratio; The meteorological variable data include the integrated ERA5 reanalysis data set; The local component concentration data include real-time monitoring values ​​of near-ground PM2.5 chemical component concentrations.

4. According to claim 3, a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning is characterized in that: The edge weight is calculated as follows: in, is the wind speed eigenvector of the ith node, is the wind speed eigenvector of the jth node, is the bandwidth parameter, which controls the weight decay speed.

5. According to claim 4, a PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning is characterized in that: The adaptive meta-learning module is specifically used to construct and train an initial adaptive meta-learning model using sample multi-source data collected from multiple regions; Based on the edge weight Updating the initial adaptive meta-learning model to obtain a first adaptive meta-learning model; Based on the multi-source data of the current area collected in n consecutive time periods, the parameters of the first adaptive meta-learning model are updated through the Adam optimizer to obtain the target adaptive meta-learning model.

6. A PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning according to claim 5, characterized in that: The adaptive meta-learning module is further used to collect and obtain sample multi-source data from multiple regions, wherein the sample multi-source data includes multi-source data from various regions monitored for one year; Preprocessing the sample multi-source data of each region respectively and splitting them into a support set and a query set; Construct a first initial adaptive meta-learning model; use the support set and query set corresponding to each region to train the first initial adaptive meta-learning model and calculate the loss function, and when the loss function is less than or equal to the minimum loss function threshold, the model converges, and the initial parameters of the first initial adaptive meta-learning model corresponding to each region are obtained. ; Based on the initial parameters Update the initial parameters of the first initial adaptive meta-learning model through two-level optimization , get the target model parameters .

7. A PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning according to claim 6, characterized in that: The loss function is calculated as follows: ; In the formula, is the importance weight of each component concentration; is the predicted sample local component concentration data of the i-th sample multi-source data; is the sample local component concentration data of the i-th sample multi-source data; The target model parameters are calculated as follows: ; ; are the initial parameters of the first initial adaptive meta-learning model, is the inner layer learning rate, is the loss gradient, where Represents the training missions for each region, For the first initial adaptive meta-learning model, The loss dependence ratio.

8. The PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning according to claim 7 is characterized in that: The adaptive meta-learning module is also used in the specific implementation process to calculate the edge weights of the graph structure. Get the k-th layer edge weight of the k-th vertical height layer ; Based on the k-th layer edge weight The neighbor node set of the i-th vertical height layer is iteratively updated and calculated by the GraphSAGE algorithm to obtain the embedding value of the i-th node in the k-th vertical height layer; The initial adaptive meta-learning model is updated based on the node embedding value to obtain a first adaptive meta-learning model.

9. A PM2.5 chemical component vertical profile inversion system based on graph neural network and adaptive meta-learning according to claim 8, characterized in that: The node embedding value is calculated as follows: ; In the formula, is the edge weight of the kth layer; is the set of neighbor nodes; is the embedding of node i in the k-1th layer, Indicates the jth node from the neighbor node set; is an aggregate function, is the connection function, is a non-linear activation function.

10. A PM2.5 chemical component vertical profile inversion method based on graph neural network and adaptive meta-learning, characterized in that: The steps are as follows: Collect and obtain multi-source data on k vertical height layers of the current geographical area; pre-process the multi-source data to obtain pre-processed multi-source data; Each of the vertical height layers is used as a node to construct a graph structure; The edge weight between the i-th node and the j-th node in the graph structure is calculated by using the meteorological variable data collected during a preset time period. ; Based on the edge weight Combining sample multi-source data collected from multiple regions to build and train a target adaptive meta-learning model; The preprocessed multi-source data of the current area at the current moment is processed by the target adaptive meta-learning model, and the inversion result of the vertical profile of the predicted PM2.5 chemical components is output.

Images