Method for identifying transmission path of ozone precursor and related equipment

Abstract

The invention relates to the technical field of ozone precursor transmission, and provides an ozone precursor transmission path identification method and related equipment, and the method comprises the steps: constructing an atmospheric pollutant transmission graph network based on all grids and ozone precursor data at all moments; predicting the ozone concentration of each grid by using the graph convolutional network according to the atmospheric pollutant transmission graph network and all ozone precursor data to obtain the predicted ozone concentration of each grid at each moment, and training the graph convolutional network by using all the predicted ozone concentrations to obtain a trained graph convolutional network; performing feature extraction on the atmospheric pollutant transmission graph network by using the trained graph convolutional network to obtain an ozone precursor transmission intensity matrix at each moment; and according to the transmission intensity matrix of all the ozone precursors and all the grids, carrying out path identification to obtain a key transmission path of the ozone precursors in the target area. According to the method, the identification accuracy of the ozone precursor transmission path can be improved.

Description

Technical Field

[0001] This application relates to the technical field of ozone precursor transmission, and particularly to a method for identifying the transmission path of ozone precursors and related equipment. Background Art

[0002] With the increasing severity of the atmospheric ozone pollution problem in China, ozone pollution has become an important factor affecting urban air quality and public health. Ozone is not directly emitted into the atmosphere, but is formed by the photochemical reaction of precursors such as nitrogen oxides (NOx, Nitrogen Oxides) and volatile organic compounds (VOCs, Volatile Organic Compounds) under specific meteorological conditions. Therefore, understanding the propagation path, concentration change and spatial distribution law of ozone precursors in the atmosphere is of great significance for formulating effective ozone pollution prevention and control measures.

[0003] Most of the existing transmission path analysis methods focus on the diffusion path of known pollutants and their influence range, such as the method and system for identifying key nodes of air pollution transmission based on complex networks, and a method for analyzing regional air pollutants based on complex networks. However, less attention has been paid to the propagation characteristics of ozone precursors (such as nitrogen oxides and volatile organic compounds) in the atmosphere. The transmission and reaction processes of these precursors are the key factors in the generation of ozone pollution. The current methods fail to effectively consider the influence of the propagation of precursors on secondary pollution (i.e., ozone generation), resulting in large deviations in the prediction results.

[0004] In addition, traditional complex network models usually focus on the direct connection and propagation between nodes. Especially in the process of atmospheric pollutant propagation, meteorological factors such as wind speed, temperature, and humidity have an important impact on the diffusion of pollutants. The complex network algorithm fails to fully consider these factors that change with time and space. It can be seen that there is currently a problem of inaccurate identification of the transmission path of ozone precursors. Summary of the Invention

[0005] This application provides a method for identifying the transmission path of ozone precursors and related equipment, which can solve the problem of inaccurate identification of the transmission path of ozone precursors.

[0006] In a first aspect, an embodiment of this application provides a method for identifying the transmission path of ozone precursors, and the identification method includes: Obtain ozone precursor data of a target area at multiple moments; the ozone precursor data includes multiple data that affect the transmission of ozone precursors; Divide the target area into multiple grids, and construct an atmospheric pollutant transmission graph network based on all grids and ozone precursor data at all times; multiple nodes of the atmospheric pollutant transmission graph network correspond to multiple grids one by one, and the edges between nodes are the adjacent transmission relationships between the corresponding two grids; Use a graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission graph network and all ozone precursor data, obtain the predicted ozone concentration of each grid at each time, and use all the predicted ozone concentrations to train the graph convolutional network to obtain a trained graph convolutional network; Use the trained graph convolutional network to extract features from the atmospheric pollutant transmission graph network to obtain the ozone precursor transmission intensity matrix at each time; Perform path recognition according to all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of ozone precursors in the target area.

[0007] Optionally, constructing an atmospheric pollutant transmission graph network based on all grids and ozone precursor data at all times includes: For each pair of grids at each time, determine whether the two grids satisfy the adjacent transmission condition at that time. If so, it is considered that there is an adjacent transmission relationship between the two grids at that time; Generate a corresponding node for each grid, and generate edges between nodes according to all adjacent transmission relationships to obtain an atmospheric pollutant transmission graph network.

[0008] Optionally, the adjacent transmission condition is: Among them, represents the row difference between the th grid and the th grid at the th time, represents the column difference between the th grid and the th grid at the th time, represents the horizontal wind speed from the th grid to the th grid at the th time, represents the angle between the horizontal wind speed and the horizontal transmission direction of the line connecting the th grid to the th grid, represents the terrain correction angle between the th grid and the th grid, , represents the number of times, , represents a set of grid numbers.

[0009] Optionally, using a graph convolutional network, the ozone concentration of each grid is predicted based on the atmospheric pollutant transport map network and all ozone precursor data, and the predicted ozone concentration of each grid at each moment is obtained, including: Construct a node feature matrix for each moment based on the atmospheric pollutant transport map network and all ozone precursor data; the elements in the node feature matrix are the ozone precursor data corresponding to each node in the atmospheric pollutant transport map network; Construct a dynamic adjacency matrix for each moment based on the ozone precursor data at each moment; For each moment respectively, using a graph convolutional network, update the node feature matrix of the moment based on the dynamic adjacency matrix of the moment to obtain the new node feature matrix of the moment, and map the new node feature matrix to obtain the predicted ozone concentration of each grid at the moment.

[0010] Optionally, constructing a dynamic adjacency matrix for each moment includes: Through the formula: Calculate the adjacency parameter between the th grid and the th grid at the th moment; where , are both weight coefficients, represents the terrain-corrected wind speed from the th grid to the th grid at the th moment, , represents the horizontal wind speed from the th grid to the th grid at the th moment, represents the angle between the horizontal wind speed and the horizontal transmission direction of the line connecting the th grid to the th grid, represents the terrain correction angle between the th grid and the th grid, represents the minimum value of the terrain-corrected wind speed, represents the maximum value of the terrain-corrected wind speed, represents the ozone concentration of the th grid in the ozone precursor data at the th moment, Denotes the ozone concentration of the th grid in the ozone precursor data at the th moment, denotes the maximum value of the ozone concentration, denotes the minimum value of the ozone concentration, , denotes the number of moments, , denotes the set of grid numbers; Integrate the adjacency parameters corresponding to each moment into a matrix to obtain the initial adjacency matrix for each moment;

[0011] Through the formula: Calculate the dynamic adjacency matrix of the th moment ; Among them, denotes the initial adjacency matrix of the th moment, denotes the initial adjacency matrix of the th moment; Use the graph convolutional network to update the node feature matrix of the moment based on the dynamic adjacency matrix of the moment to obtain the new node feature matrix of the moment, including: Through the formula: ; Calculate the output of the th layer ; Among them, denotes 's degree matrix, denotes the dynamic adjacency matrix of the th moment with self-loops added , , denotes the output of the th layer, denotes the trainable weight matrix of the th layer, , denotes the last layer. When , denotes the dynamic adjacency matrix of the th moment. When , denotes the new node feature matrix of the th moment.

[0012] Optionally, use the trained graph convolutional network to extract features from the atmospheric pollutant transport graph network to obtain the ozone precursor transport intensity matrix, including: For each moment, perform the following steps: Use the trained graph convolutional network to update the node feature matrix at the moment based on the dynamic adjacency matrix at the moment to obtain the final node feature matrix at the moment; Extract and integrate the features related to ozone precursors in the final node feature matrix to obtain the ozone precursor transport features at this moment; Calculate the ozone precursor transport intensity matrix at the moment based on the ozone precursor transport features.

[0013] Optionally, calculate the ozone precursor transport intensity matrix at the moment based on the ozone precursor transport features, including: Through the formula: Calculate the elements of the first sub-matrix in the ozone precursor transport intensity matrix and the elements of the second sub-matrix ; where, represents the transport intensity of nitrogen oxides from the th grid to the th grid at the th moment, represents the transport intensity of volatile organic compounds from the th grid to the th grid at the th moment, represents the terrain-corrected wind speed from the th grid to the th grid at the th moment, represents the nitrogen oxide-related feature of the th grid in the ozone precursor transport features at the th moment, represents the nitrogen oxide-related feature of the th grid in the ozone precursor transport features at the th moment, represents the minimum value of the terrain-corrected wind speed, represents the maximum value of the terrain-corrected wind speed, represents the maximum value of the nitrogen oxide-related features, represents the minimum value of the nitrogen oxide-related features, 、 are both weight coefficients, Indicates the characteristics related to volatile organic compounds in the th grid among the ozone precursor transport characteristics at the th moment, Indicates the characteristics related to volatile organic compounds in the th grid among the ozone precursor transport characteristics at the th moment, , Indicates the number of moments, , Indicates the set of grid numbers.

[0014] Optionally, based on all ozone precursor transport intensity matrices and all grids, path recognition is performed to obtain the key transport paths of ozone precursors in the target area, including: Regarding the grids corresponding to the potential pollution source areas in all grids as source grids; For each moment respectively, according to the ozone precursor transport intensity matrix of the moment, starting from each source grid, search for the transport paths of the source grid among all grids; Aggregate all the transport paths of all moments to obtain a set of multiple spatio-temporal trajectories; Calculate the similarity between every two spatio-temporal trajectories; Cluster all the spatio-temporal trajectories according to all the similarities to obtain the key transport paths.

[0015] In a second aspect, an embodiment of the present application provides an apparatus for identifying ozone precursor transport paths, including: An acquisition module, configured to acquire ozone precursor data of the target area at multiple moments; the ozone precursor data includes multiple data affecting the transport of ozone precursors; A division module, configured to divide the target area into multiple grids and construct an atmospheric pollutant transport graph network based on all grids and ozone precursor data of all moments; multiple nodes of the atmospheric pollutant transport graph network correspond to multiple grids one by one, and the edges between the nodes are the adjacent transport relationships between the corresponding two grids; A prediction module, configured to use a graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transport graph network and all ozone precursor data, obtain the predicted ozone concentration of each grid at each moment, and use all the predicted ozone concentrations to train the graph convolutional network to obtain the trained graph convolutional network; A feature extraction module, configured to extract features from the atmospheric pollutant transmission graph network by using the trained graph convolutional network, so as to obtain the ozone precursor transmission intensity matrix at each moment; A path recognition module, configured to perform path recognition according to all ozone precursor transmission intensity matrices and all grids, so as to obtain the key transmission paths of ozone precursors in the target area.

[0016] In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned method for identifying the ozone precursor transmission path is implemented.

[0017] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned method for identifying the ozone precursor transmission path is implemented.

[0018] The above solution of the present application has the following beneficial effects: In the embodiment of the present application, by obtaining ozone precursor data in the target area at multiple moments, then dividing the target area into multiple grids, constructing an atmospheric pollutant transmission graph network based on all grids and ozone precursor data at all moments, and then using the graph convolutional network, predicting the ozone concentration of each grid according to the atmospheric pollutant transmission graph network and all ozone precursor data, obtaining the predicted ozone concentration of each grid at each moment, and using all the predicted ozone concentrations to train the graph convolutional network to obtain the trained graph convolutional network, then using the trained graph convolutional network to extract features from the atmospheric pollutant transmission graph network to obtain the ozone precursor transmission intensity matrix at each moment, and finally performing path recognition according to all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of ozone precursors in the target area. Among them, constructing an atmospheric pollutant transmission graph network based on ozone precursor data including multiple data takes into account multiple factors affecting the transmission of ozone precursors, improves the comprehensiveness of information in the atmospheric pollutant transmission graph network, performs feature extraction and path recognition on the atmospheric pollutant transmission graph network with comprehensive information, can effectively capture the transmission paths of ozone precursors in space, identify dynamic propagation characteristics, improve the accuracy of the key transmission paths of ozone precursors, and further improve the accuracy of identifying the ozone precursor transmission path.

[0019] Other beneficial effects of the present application will be described in detail in the subsequent specific implementation part. Description of the Drawings

[0020] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a flowchart of a method for identifying the ozone precursor transmission path provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of a device for identifying the ozone precursor transmission path provided by an embodiment of the present application; Figure 3 It is a schematic structural diagram of a terminal device provided by an embodiment of the present application. Detailed implementation manners

[0022] In the following description, for the purpose of illustration rather than limitation, specific details such as specific system structures and technologies are proposed to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, the detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0023] It should be understood that when used in the specification and the appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0024] It should also be understood that the term "and / or" used in the specification and the appended claims of the present application refers to any combination and all possible combinations of one or more of the related listed items, and includes these combinations.

[0025] As used in the specification and the appended claims of the present application, the term "if" can be interpreted as "when", "once", "in response to determining", or "in response to detecting" according to the context. Similarly, the phrase "if determined" or "if detecting [the described condition or event]" can be interpreted as meaning "once determined", "in response to determining", "once detecting [the described condition or event]", or "in response to detecting [the described condition or event]" according to the context.

[0026] In addition, in the description of the specification and the appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0027] As used in the specification of this application, references to "one embodiment" or "some embodiments" etc. mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of this application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized.

[0028] In view of the problem of inaccurate identification of the ozone precursor transmission path in the prior art, the embodiment of this application provides a method for identifying the ozone precursor transmission path. The identification method obtains the ozone precursor data of the target area at multiple moments, then divides the target area into multiple grids, and constructs an air pollutant transmission graph network based on all grids and the ozone precursor data at all moments. Then, using a graph convolutional network, based on the air pollutant transmission graph network and all the ozone precursor data, the ozone concentration of each grid is predicted to obtain the predicted ozone concentration of each grid at each moment, and all the predicted ozone concentrations are used to train the graph convolutional network to obtain a trained graph convolutional network. Then, the trained graph convolutional network is used to extract features from the air pollutant transmission graph network to obtain the ozone precursor transmission intensity matrix at each moment. Finally, based on all the ozone precursor transmission intensity matrices and all the grids, path identification is performed to obtain the key transmission path of the ozone precursor in the target area. Among them, constructing an air pollutant transmission graph network based on ozone precursor data including multiple data takes into account multiple factors affecting the transmission of ozone precursors, improves the comprehensiveness of information in the air pollutant transmission graph network, and performs feature extraction and path identification on the air pollutant transmission graph network with comprehensive information, which can effectively capture the transmission path of ozone precursors in space, identify dynamic propagation characteristics, improve the accuracy of the key transmission path of ozone precursors, and thus improve the accuracy of identifying the ozone precursor transmission path.

[0029] Next, an exemplary description will be given of the method for identifying the ozone precursor transmission path provided by this application.

[0030] As Figure 1 shown, the method for identifying the ozone precursor transmission path provided by this application includes the following steps: Step 11, obtain the ozone precursor data of the target area at multiple moments.

[0031] The above ozone precursor data includes multiple data that affect the transmission of ozone precursors, such as surface ozone concentration data, precursor emission data (including NOx emission data and VOCs emission data), meteorological data (including surface temperature, atmospheric humidity, solar radiation intensity, wind speed and direction, boundary layer height, etc.), geographical information data (including regional terrain elevation), etc. The above target area is an area where the transmission path of ozone precursors needs to be identified, such as a certain county or a certain township. Multiple moments are multiple moments within the time period for analyzing the transmission path of ozone precursors. For example, if it is necessary to identify and analyze the transmission path of ozone precursors in the target area from January 1st to February 1st, then the multiple moments can be January 1st, January 10th, January 20th, and February 1st.

[0032] In some embodiments of the present application, ozone precursor data can be obtained through devices such as sensors and theodolites.

[0033] Step 12: Divide the target area into multiple grids, and construct an atmospheric pollutant transmission graph network based on the ozone precursor data of all grids and all moments.

[0034] Multiple nodes of the above atmospheric pollutant transmission graph network correspond one-to-one with multiple grids, and the edges between the nodes are the adjacent transmission relationships between the corresponding two grids. After dividing the target area into grids, the obtained multiple grids are arranged in rows and columns. The rows are the horizontal division of the target area, and the columns are the vertical division of the target area. The horizontal and vertical directions of the target area are set according to the actual geographical location of the target area. For example, the latitude direction is set as the horizontal direction, and the longitude direction is set as the vertical direction.

[0035] In some embodiments of the present application, the step of dividing the target area into multiple grids and constructing an atmospheric pollutant transmission graph network based on the ozone precursor data of all grids and all moments includes: The first step: For every two grids at each moment, determine whether the two grids satisfy the adjacent transmission condition at that moment. If so, it is considered that there is an adjacent transmission relationship between the two grids at that moment. If the two grids do not satisfy the adjacent transmission condition at that moment, it is considered that there is no adjacent transmission relationship between the two grids at that moment.

[0036] Specifically, the adjacent transmission condition is: Among them, represents the row difference between the th grid and the th grid at the th moment, represents at the th moment, the The column difference between the th grid and the th grid at the th moment, from the th grid to the th grid, represents the horizontal wind speed, which represents the angle between the horizontal wind speed and the horizontal transmission direction of the line connecting the th grid and the th grid, which represents the terrain correction angle between the th grid and the th grid, where represents the number of moments, and

[0037] It should be noted that the row difference is the difference between the row numbers of the two grids (for example, if one grid is in the 2nd row and the other grid is in the 3rd row, the row difference is 1), and the column difference is the difference between the column numbers of the two grids (for example, if one grid is in the 6th column and the other grid is in the 2nd column, the column difference is 4). Since the horizontal wind speed from the th grid to the th grid is unidirectional, when analyzing the adjacent transmission relationship through the above process, it is necessary to consider the other propagation direction between the two grids. For example, when considering the horizontal wind speed from the th grid to the th grid, the adjacent transmission condition is not met, but when considering the horizontal wind speed from the th grid to the th grid, the adjacent propagation condition is met. In this case, it is still considered that there is an adjacent propagation relationship between the th grid and the th grid.

[0038] In the second step, a corresponding node is generated for each grid, and edges between the nodes are generated according to all adjacent transmission relationships to obtain the air pollutant transmission graph network.

[0039] It should be noted that for each moment, if there is a neighboring transmission relationship between the two grids corresponding to the two nodes at this moment, an edge is generated, and the edge is directed (determined according to the direction of the horizontal wind speed considered when there is a neighboring propagation relationship in the previous step. The direction of the horizontal wind speed determines the direction of the directed edge. If there is a neighboring propagation relationship in both directions, the edge is bidirectional). If there is no neighboring transmission relationship between the two grids corresponding to the two nodes at this moment, no edge is generated. For example, at the 1st moment and the 3rd moment, there is a neighboring relationship between the two grids, so there are two edges between the two nodes corresponding to the two grids, respectively representing the neighboring transmission relationship of the two grids at the 1st moment and the neighboring transmission relationship at the 3rd moment. The data corresponding to the nodes in all the ozone precursor data is used as the attributes of the nodes.

[0040] Step 13: Use the graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission graph network and all the ozone precursor data, obtain the predicted ozone concentration of each grid at each moment, and use all the predicted ozone concentrations to train the graph convolutional network to obtain the trained graph convolutional network.

[0041] In some embodiments of the present application, the above step of using the graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission graph network and all the ozone precursor data, obtain the predicted ozone concentration of each grid at each moment, and use all the predicted ozone concentrations to train the graph convolutional network to obtain the trained graph convolutional network includes: The first step: Construct the node feature matrix of each moment based on the atmospheric pollutant transmission graph network and all the ozone precursor data.

[0042] The elements in the above node feature matrix are the ozone precursor data corresponding to each node in the atmospheric pollutant transmission graph network.

[0043] Specifically, for each moment respectively, extract the data corresponding to each node in the ozone precursor data of this moment, and integrate the data of all nodes into a matrix to obtain the node feature matrix.

[0044] Such as the expression: Among them, 、 are the NOx emissions and VOCs emissions of node at the th moment, 、 、 、 、 、 respectively represent node Wind speed, temperature, humidity, solar radiation intensity, boundary layer height, and terrain elevation data at the th moment.

[0045] Second, construct a dynamic adjacency matrix for each moment based on the ozone precursor data at each moment.

[0046] First, through the formula: Calculate the adjacency parameter between the th grid and the th grid at the th moment.

[0047] Among them, , are both weight coefficients, represents the terrain-corrected wind speed transmitted from the th grid to the th grid at the th moment, , represents the horizontal wind speed from the th grid to the th grid at the th moment, represents the angle between the horizontal wind speed and the horizontal transmission direction of the line connecting the th grid to the th grid, represents the terrain correction angle between the th grid and the th grid, represents the minimum value of the terrain-corrected wind speed, represents the maximum value of the terrain-corrected wind speed, represents the ozone concentration of the th grid in the ozone precursor data at the th moment, represents the ozone concentration of the th grid in the ozone precursor data at the th moment, represents the maximum value of the ozone concentration, represents the minimum value of the ozone concentration, , represents the number of moments, , represents the set of grid numbers.

[0048] Then, integrate the adjacency parameters corresponding to each moment into a matrix to obtain the initial adjacency matrix for each moment.

[0049] Finally, through the formula: calculate the dynamic adjacency matrix at the th moment.

[0050] Among them, represents the initial adjacency matrix at the th moment, represents the initial adjacency matrix at the th moment.

[0051] In the third step, for each moment respectively, use the graph convolutional network to update the node feature matrix at that moment based on the dynamic adjacency matrix at that moment, obtain the new node feature matrix at that moment, and map the new node feature matrix to obtain the predicted ozone concentration of each grid at that moment.

[0052] Specifically, through the formula: calculate the output of the th layer.

[0053] Among them, represents 's degree matrix, represents adding a self-loop to the dynamic adjacency matrix at the th moment, , represents the output of the th layer, represents the trainable weight matrix of the th layer, , represents the last layer. When , represents the dynamic adjacency matrix at the th moment. When , represents the new node feature matrix at the th moment.

[0054] Exemplarily, a fully connected layer can be used to map the new node feature matrix to obtain the predicted ozone concentration of each grid at the moment.

[0055] It can be understood that the formula for calculating the output of the th layer above is the expression of the th layer graph convolution in the graph convolutional network.

[0056] In the fourth step, use all the predicted ozone concentrations to train the graph convolutional network to obtain the trained graph convolutional network.

[0057] Specifically, a loss function is constructed based on all predicted ozone concentrations, and then the parameters of the graph convolutional network are optimized through the backpropagation algorithm to minimize the loss function , and the weight matrix is updated using the gradient descent method .

[0058] The above loss function is as follows: where and are the true concentration and predicted concentration of ozone at the grid at the -th moment respectively, is the total number of grids, is the length of the time series.

[0059] Step 14: Use the trained graph convolutional network to extract features from the air pollutant transport graph network to obtain the ozone precursor transport intensity matrix at each moment.

[0060] In some embodiments of the present application, the step of using the trained graph convolutional network to extract features from the air pollutant transport graph network to obtain the ozone precursor transport intensity matrix at each moment includes: For each moment respectively, the following steps are performed: The first step: Use the trained graph convolutional network to update the node feature matrix at the moment based on the dynamic adjacency matrix at the moment to obtain the final node feature matrix at the moment.

[0061] Specifically, substitute the dynamic adjacency matrix into the trained graph convolutional network for calculation to obtain the final node feature matrix.

[0062] The second step: Extract and integrate the features related to ozone precursors in the final node feature matrix to obtain the ozone precursor transport features at this moment.

[0063] The final node feature matrix includes multiple features of each node (such as features of surface ozone concentration, precursor emission data, meteorological data, and geographic information data). All features related to ozone precursors are selected and integrated from them to obtain ozone precursor transport features.

[0064] The third step: Calculate the ozone precursor transport intensity matrix at this moment based on the ozone precursor transport features.

[0065] The above ozone precursor transmission intensity matrix includes a first sub - matrix and a second sub - matrix. The first sub - matrix is used to describe the transmission intensity of nitrogen oxides between every two grids, and the second sub - matrix is used to describe the transmission intensity of ozone precursors between every two grids.

[0066] Specifically, through the formula: Calculate the elements of the first sub - matrix and the elements of the second sub - matrix in the ozone precursor transmission intensity matrix and ; wherein, represents the transmission intensity of nitrogen oxides from the th grid to the th grid at the th moment, represents the transmission intensity of volatile organic compounds from the th grid to the th grid at the th moment, represents the terrain - corrected wind speed from the th grid to the th grid at the th moment, represents the nitrogen - oxide - related feature of the th grid in the ozone precursor transmission characteristics at the th moment, represents the nitrogen - oxide - related feature of the th grid in the ozone precursor transmission characteristics at the th moment, represents the minimum value of the terrain - corrected wind speed, represents the maximum value of the terrain - corrected wind speed, represents the maximum value of the nitrogen - oxide - related feature, represents the minimum value of the nitrogen - oxide - related feature, and are both weight coefficients, represents the volatile - organic - compound - related feature of the th grid in the ozone precursor transmission characteristics at the th moment, represents the volatile - organic - compound - related feature of the th grid in the ozone precursor transmission characteristics at the th moment, represents the maximum value of the volatile - organic - compound - related feature, represents the minimum value of the volatile - organic - compound - related feature, , represents the number of moments, , represents the set of grid numbers.

[0067] Step 15: Perform path recognition based on all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of ozone precursors in the target area.

[0068] In some embodiments of the present application, the above step of performing path recognition based on all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of ozone precursors in the target area includes: The first step: Use the grids corresponding to the potential pollution source areas in all grids as source grids.

[0069] Exemplarily, use industrial areas, natural source emission areas, etc. as potential pollution source areas, and use the grids corresponding to the potential pollution source areas as source grids.

[0070] The second step: For each moment, respectively, according to the ozone precursor transmission intensity matrix at that moment, use each source grid as the starting point to search for the transmission paths of the source grid in all grids.

[0071] Specifically, use the source grid as the current grid, obtain all the elements related to the current grid in the ozone precursor transmission intensity matrix, and determine whether there are elements greater than 0 among all the elements; If so, use the grid corresponding to the element with the largest value as the transmission grid, and use this transmission grid as the current grid, and return to the step of obtaining all the elements related to the current grid in the ozone precursor transmission intensity matrix; Otherwise, end the search, and integrate the source grid and all transmission grids into a transmission path , where represents the source grid, represents the transmission grid determined according to the source grid, represents the last propagation grid.

[0072] It should be noted that the grid corresponding to the ozone pollution core area in the target area can also be used as the target grid, use the target grid as the starting point, and perform reverse tracking according to the above search process of the source grid to gradually find the possible transmission sources of the precursors and obtain the propagation path , where represents the target grid, represents the grid where the precursors start to propagate traced back reversely.

[0073] The third step: Aggregate all the transmission paths at all moments to obtain a set of multiple spatio-temporal trajectories.

[0074] Specifically, the set of multiple spatio-temporal trajectories is as follows: Among them, is the geographical coordinate (i.e., the coordinate of the grid center point in the transmission path), is the spatio-temporal trajectory 's length, represents the number of spatio-temporal trajectories (i.e., the total number of all transmission paths at all moments in a cycle).

[0075] Exemplarily, the spatio-temporal trajectory corresponds to a certain transmission path among all propagation paths at all moments. The geographical coordinate in the spatio-temporal trajectory corresponds to the grid in this transmission path. The length of the spatio-temporal trajectory is the number of grids in this transmission path.

[0076] Fourth step, calculate the similarity between every two spatio-temporal trajectories.

[0077] Exemplarily, the dynamic time warping algorithm (DTW, Dynamic Time Warping) can be used to calculate the similarity between spatio-temporal trajectories. The expression is: Among them, represents the similarity between the spatio-temporal trajectory and the spatio-temporal trajectory , represents the cumulative distance between the spatio-temporal trajectory and the spatio-temporal trajectory . The recurrence formula is: ; ; ; Among them, represents 's node and 's node's cumulative distance, represents 's node and 's node's distance, represents 's node and 's node's cumulative distance, represents The node and the cumulative distance of the node, represents the node and the cumulative distance of the node, represents the cumulative distance between the 1st node of and the node, represents the cumulative distance between the node and the 1st node of node.

[0078] Step 5: Cluster all spatio-temporal trajectories according to all similarities to obtain the key transmission paths.

[0079] Exemplarily, the K-medoids algorithm can be used to cluster all spatio-temporal trajectories according to all similarities to obtain multiple clusters, and a spatio-temporal trajectory is randomly selected from each cluster as the key transmission path of the cluster (the key transmission path of the cluster can be updated). The process of the K-medoids algorithm is as follows: Randomly select K representative trajectories (medoids), which represents the medoids set. Assign each trajectory to the nearest medoid: Then for each cluster select a new medoid such that: Repeat the assignment and update until the medoids are stable or the maximum number of iterations is reached. At this time, the final clusters are obtained: , the clustered trajectories: , each cluster contains trajectories: wherein, is the sample size of the cluster , is the total number of trajectories identified within the period, is the indicator function, when the trajectory belongs to the cluster , the function value is 1, otherwise it is 0.

[0080] It is worth mentioning that an atmospheric pollutant transmission map network is constructed based on ozone precursor data including multiple data, taking into account multiple factors affecting the transmission of ozone precursors, improving the comprehensiveness of information in the atmospheric pollutant transmission map network. Feature extraction and path recognition are performed on the atmospheric pollutant transmission map network with comprehensive information, which can effectively capture the spatial transmission path of ozone precursors, identify dynamic propagation characteristics, improve the accuracy of the key transmission paths of ozone precursors, and further improve the accuracy of identifying the transmission paths of ozone precursors.

[0081] The following provides an exemplary description of the apparatus for identifying the transmission path of ozone precursors provided in this application.

[0082] As Figure 2 shown, an embodiment of this application provides an apparatus for identifying the transmission path of ozone precursors. The apparatus 200 for identifying the transmission path of ozone precursors includes: An acquisition module 201, configured to acquire ozone precursor data of a target area at multiple moments; the ozone precursor data includes multiple data affecting the transmission of ozone precursors; A division module 202, configured to divide the target area into multiple grids, and construct an atmospheric pollutant transmission map network based on all grids and ozone precursor data at all moments; multiple nodes of the atmospheric pollutant transmission map network correspond to the multiple grids one by one, and the edges between the nodes are the neighboring transmission relationships between the corresponding two grids; A prediction module 203, configured to use a graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission map network and all ozone precursor data, obtain the predicted ozone concentration of each grid at each moment, and use all the predicted ozone concentrations to train the graph convolutional network to obtain a trained graph convolutional network; A feature extraction module 204, configured to use the trained graph convolutional network to perform feature extraction on the atmospheric pollutant transmission map network to obtain an ozone precursor transmission intensity matrix at each moment; A path recognition module 205, configured to perform path recognition according to all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of ozone precursors in the target area.

[0083] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units, due to being based on the same concept as the method embodiment of this application, for their specific functions and the technical effects brought, please refer to the method embodiment part specifically, and will not be elaborated here.

[0084] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0085] As Figure 3 shown, an embodiment of the present application provides a terminal device. The terminal device D10 in this embodiment includes: at least one processor D100 ( Figure 3 only one processor is shown in the figure), a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100. When the processor D100 executes the computer program D102, it implements the steps in any of the foregoing method embodiments.

[0086] Specifically, when the processor D100 executes the computer program D102, it obtains the ozone precursor data of the target area at multiple moments, then divides the target area into multiple grids, constructs an atmospheric pollutant transmission map network based on all grids and the ozone precursor data at all moments, and then uses a graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission map network and all ozone precursor data, obtaining the predicted ozone concentration of each grid at each moment, and uses all the predicted ozone concentrations to train the graph convolutional network to obtain a trained graph convolutional network. Then, the trained graph convolutional network is used to extract features from the atmospheric pollutant transmission map network to obtain the ozone precursor transmission intensity matrix at each moment. Finally, path recognition is performed based on all the ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of the ozone precursors in the target area. Among them, constructing an atmospheric pollutant transmission map network based on the ozone precursor data including multiple data takes into account multiple factors affecting the transmission of ozone precursors, improves the comprehensiveness of information in the atmospheric pollutant transmission map network, performs feature extraction and path recognition on the atmospheric pollutant transmission map network with comprehensive information, can effectively capture the transmission paths of ozone precursors in space, identify dynamic propagation characteristics, improve the accuracy of the key transmission paths of ozone precursors, and thus improve the accuracy of identifying the transmission paths of ozone precursors.

[0087] The so-called processor D100 may be a central processing unit (CPU, Central Processing Unit), and this processor D100 may also be other general-purpose processors, digital signal processors (DSP, Digital Signal Processor), application specific integrated circuits (ASIC, Application Specific Integrated Circuit), field-programmable gate arrays (FPGA, Field-Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0088] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In some other embodiments, the memory D101 may also be an external storage device of the terminal device D10, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device D10. Further, the memory D101 may also include both the internal storage unit and the external storage device of the terminal device D10. The memory D101 is used to store an operating system, application programs, a boot loader (BootLoader), data, and other programs, such as program codes of the computer program. The memory D101 may also be used to temporarily store data that has been output or is to be output.

[0089] An embodiment of the present application further provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments can be implemented.

[0090] An embodiment of the present application provides a computer program product, and when the computer program product runs on a terminal device, the terminal device is enabled to implement the steps in the above method embodiments.

[0091] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, all or part of the processes in the above method embodiments of the present application can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps in the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable medium may at least include: any entity or device capable of carrying the computer program code to the ozone precursor transport path identification method device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc.

[0092] In the above embodiments, the descriptions of the various embodiments each have their own emphasis. For parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0093] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0094] The above is the preferred embodiment of this application. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle described in this application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of this application.

Claims

1. A method for identifying an ozone precursor transmission path, characterized in that: include: Obtain ozone precursor data in the target area at multiple times; The ozone precursor data includes a plurality of data affecting the transmission of the ozone precursor; The target area is divided into a plurality of grids, and an atmospheric pollutant transmission graph network is constructed based on the ozone precursor data of all grids and all times; the plurality of nodes of the atmospheric pollutant transmission graph network correspond one-to-one to the plurality of grids, and the edges between the nodes are the adjacent transmission relations between the corresponding two grids; Using a graph convolutional network, predicting the ozone concentration of each of the grids according to the atmospheric pollutant transmission graph network and all ozone precursor data, obtaining the predicted ozone concentration of each of the grids at each moment, and training the graph convolutional network using all the predicted ozone concentrations to obtain a trained graph convolutional network; Using the trained graph convolutional network to extract features from the atmospheric pollutant transmission graph network, and obtaining an ozone precursor transmission intensity matrix at each moment; Path identification is performed based on all ozone precursor transmission intensity matrices and all grids to obtain key transmission paths of ozone precursors in the target area.

2. The identification method according to claim 1, characterized in that: The method of constructing an atmospheric pollutant transmission map network based on ozone precursor data of all grids and all times includes: For each two grids at each moment, determine whether the two grids at the moment meet the adjacent transmission condition, and if so, consider that there is an adjacent transmission relationship between the two grids at the moment; A corresponding node is generated for each grid, and the edges between nodes are generated according to all adjacent transmission relationships to obtain the atmospheric pollutant transmission graph network.

3. The identification method according to claim 2, characterized in that: The adjacent transmission condition is: in, Indicated in moment, The grid and The row difference between the grids, Indicated in moment, The grid and The column difference between the grids, Indicated in moment, Grid to The horizontal wind speed of the grid, Indicates the horizontal wind speed and Grid to The angle between the horizontal transmission directions of the lines connecting the grids, Indicates The grid and The terrain correction angle between grids, , represents the number of moments, , Represents a collection of grid numbers.

4. The identification method according to claim 1, characterized in that: The method of using a graph convolutional network to predict the ozone concentration of each grid according to the atmospheric pollutant transmission graph network and all ozone precursor data to obtain the predicted ozone concentration of each grid at each moment includes: Constructing a node feature matrix at each moment based on the atmospheric pollutant transmission graph network and all ozone precursor data; the elements in the node feature matrix are the ozone precursor data corresponding to each node in the atmospheric pollutant transmission graph network; A dynamic adjacency matrix is ​​constructed at each moment based on the ozone precursor data at each moment; For each moment respectively, the graph convolutional network is used to update the node feature matrix at the moment based on the dynamic adjacency matrix at the moment to obtain a new node feature matrix at the moment, and the new node feature matrix is ​​mapped to obtain the predicted ozone concentration of each grid at the moment.

5. The identification method according to claim 4, characterized in that: The construction of the dynamic adjacency matrix at each moment includes: By formula: Calculate in Moment The grid and Adjacency parameters between meshes ; in, , are weight coefficients, Indicated in Moment The grid is transferred to the The terrain-corrected wind speed for each grid cell, , Indicated in moment, Grid to The horizontal wind speed of the grid, Indicates the horizontal wind speed and Grid to The angle between the horizontal transmission directions of the lines connecting the grids, Indicates The grid and The terrain correction angle between grids, Indicates the minimum value of terrain-corrected wind speed, represents the maximum value of the terrain-corrected wind speed, Indicated in The ozone precursor data at the moment The ozone concentration of each grid, Indicated in The ozone precursor data at the moment The ozone concentration of each grid, Indicates the maximum value of ozone concentration, Indicates the minimum value of ozone concentration, , represents the number of moments, , Represents a set of grid numbers; Integrate the adjacency parameters corresponding to each moment into a matrix to obtain an initial adjacency matrix at each moment; By formula: Calculate the The dynamic adjacency matrix at each moment ; in, Indicates The initial adjacency matrix at time instant is Indicates The initial adjacency matrix at time instant; The method of using the graph convolutional network to update the node feature matrix at the moment based on the dynamic adjacency matrix at the moment to obtain a new node feature matrix at the moment includes: By formula: ； Calculate the Output of the layer ; in, express The degree matrix of Indicates joining the self-loop No. The dynamic adjacency matrix at each moment is , Indicates The output of the layer, Indicates The trainable weight matrix of the layer, , Represents the last layer, when hour, Indicates The dynamic adjacency matrix at the moment, when hour, Indicates The new node feature matrix at the moment.

6. The identification method according to claim 4, characterized in that: The method of using the trained graph convolutional network to extract features from the atmospheric pollutant transmission graph network to obtain an ozone precursor transmission intensity matrix includes: For each moment, perform the following steps: Using the trained graph convolutional network, updating the node feature matrix at the moment based on the dynamic adjacency matrix at the moment to obtain the final node feature matrix at the moment; Extracting and integrating the features related to the ozone precursors in the final node feature matrix to obtain the ozone precursor transmission features at that moment; An ozone precursor transmission intensity matrix at the moment is calculated based on the ozone precursor transmission characteristics.

7. The identification method according to claim 6, characterized in that: The calculating of the ozone precursor transmission intensity matrix at the moment based on the ozone precursor transmission characteristics includes: By formula: Calculate the elements of the first submatrix in the ozone precursor transport intensity matrix and the elements of the second submatrix ; in, Indicates that nitrogen oxides The moment from The grid is transferred to the The transmission strength of the grid, Indicates that volatile organic compounds are The moment from The grid is transferred to the The transmission strength of the grid, Indicated in Moment The grid is transferred to the The terrain-corrected wind speed for each grid cell, Indicates The transport characteristics of ozone precursors at the moment NOx-related features of the grid, Indicates The transport characteristics of ozone precursors at the moment NOx-related features of the grid, Indicates the minimum value of terrain-corrected wind speed, represents the maximum value of the terrain-corrected wind speed, represents the maximum value of the NOx-related features, represents the minimum value of the NOx-related characteristics, , are weight coefficients, Indicates The transport characteristics of ozone precursors at the moment The VOC-related features of the grid, Indicates The transport characteristics of ozone precursors at the moment The VOC-related features of the grid, represents the maximum value of the VOC-related features, Indicates the minimum value of VOC-related features, , represents the number of moments, , Represents a collection of grid numbers.

8. The identification method according to claim 1, characterized in that: The path identification is performed according to all ozone precursor transmission intensity matrices and all grids to obtain the key transmission paths of the ozone precursors in the target area, including: The grids corresponding to the potential pollution source areas in all grids are taken as source grids; For each moment, respectively, according to the ozone precursor transmission intensity matrix at the moment, taking each of the source grids as a starting point, searching for the transmission path of the source grid in all grids; Aggregate all transmission paths at all times to obtain a collection of multiple spatiotemporal trajectories; Calculate the similarity between every two space-time trajectories; All spatiotemporal trajectories are clustered according to all similarities to obtain the key transmission path.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for identifying the ozone precursor transmission path according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying the ozone precursor transmission path according to any one of claims 1 to 8 is implemented.

Images