Atmospheric pollution diffusion path tracing method and system based on meteorological data

Abstract

The invention discloses an atmospheric pollution diffusion path tracing method and system based on meteorological data, and relates to the technical field of pollution path tracing, and the method comprises the steps: building a high-density monitoring network; obtaining a time sequence point monitoring data set of the high-density monitoring network, and distributing pollution hot spot areas; configuring a high-precision sensor and a monitoring unmanned aerial vehicle in a pollution hot spot area, establishing three-dimensional data of a pollution field, and synchronously obtaining a wind power data set; establishing a state variable; performing observation variable-based wind power data set prediction fitting on a group of initial parameters by using an atmospheric diffusion model, generating pollutant distribution, and establishing a candidate solution; and performing global particle search on the candidate solution to establish a pollution diffusion path tracing result. The technical problems that in the prior art, pollution hot spot areas are difficult to accurately recognize, and the accuracy and reliability of atmospheric pollution diffusion path tracing are low are solved, and the technical effect of improving the accuracy and reliability of atmospheric pollution diffusion path tracing is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of pollution path tracing, and in particular to a method and system for tracing the diffusion path of atmospheric pollution based on meteorological data. Background Art

[0002] With the acceleration of industrialization and urbanization, the problem of air pollution is becoming increasingly serious, and it is crucial to accurately trace the diffusion path of air pollution. In the existing technology, there are many difficulties in air pollution monitoring. On the one hand, the monitoring range of traditional monitoring methods is limited and the monitoring points are sparsely distributed, making it difficult to obtain comprehensive and accurate pollution data. For example, relying only on a few fixed monitoring stations, it is impossible to grasp the detailed spatial distribution and dynamic changes of pollution in a timely manner, and it is easy to miss local areas with serious pollution. On the other hand, the existing pollution diffusion path tracing methods are not accurate. Due to the lack of deep fusion analysis of meteorological data and pollution data, it is not comprehensive and accurate when considering key parameters such as the location of pollution sources, emission intensity and emission time, which makes it difficult to accurately determine the source and diffusion trajectory of pollution.

[0003] Existing technologies have technical problems such as difficulty in accurately identifying pollution hotspots and low accuracy and reliability in tracing the diffusion paths of atmospheric pollution. Summary of the invention

[0004] The present application provides a method and system for tracing the diffusion path of atmospheric pollution based on meteorological data, which is used to solve the technical problems in the prior art that it is difficult to accurately identify pollution hotspots and the accuracy and reliability of tracing the diffusion path of atmospheric pollution are low.

[0005] In view of the above problems, the present application provides a method and system for tracing the diffusion path of atmospheric pollution based on meteorological data.

[0006] In a first aspect of the present application, a method for tracing the diffusion path of atmospheric pollution based on meteorological data is provided, the method comprising:

[0007] NB-IoT and LoRa are used to connect small air quality sensors to establish a high-density monitoring network; the time-series point monitoring data set of the high-density monitoring network is obtained, the anomalies of the time-series point monitoring data set are screened by a fixed threshold, and the pollution hotspot areas are distributed based on the spatial interpolation and point anomaly recognition results; high-precision sensors and monitoring drones are configured in the pollution hotspot areas to perform pollution collection, establish three-dimensional data of the pollution field, and simultaneously obtain wind data sets; state variables are established, and observation variables are constructed with the wind data set and the three-dimensional data of the pollution field; a set of initial parameters are predicted and fitted based on the wind data set of the observation variables using the atmospheric diffusion model to generate pollutant distribution, and inversion fitting is performed by minimizing the error function between the pollutant distribution and the three-dimensional data of the pollution field to establish candidate solutions, wherein the initial parameters are parameters set according to the state variables; a global particle search is performed on the candidate solutions, and a double-layer optimization of local inversion fitting is performed based on the global particle search results to establish the pollution diffusion path tracing results.

[0008] The second aspect of the present application provides an atmospheric pollution diffusion path tracing system based on meteorological data, the system comprising:

[0009] A high-density monitoring network establishment module is used to establish a high-density monitoring network by connecting small air quality sensors using NB-IoT and LoRa; a pollution hotspot area distribution module is used to obtain the time-series point monitoring data set of the high-density monitoring network, screen the anomalies of the time-series point monitoring data set through a fixed threshold, and distribute the pollution hotspot areas based on spatial interpolation and point anomaly recognition results; a pollution field three-dimensional data establishment module is used to configure high-precision sensors and monitoring drones in pollution hotspot areas, perform focused pollution collection, establish three-dimensional data of pollution fields, and simultaneously obtain wind data sets; a state variable establishment module is used to establish state variables, and construct observation variables with the wind data set and the pollution field three-dimensional data; a candidate solution establishment module is used to use an atmospheric diffusion model to predict and fit a set of initial parameters based on the wind data set of observation variables, generate pollutant distribution, and perform inversion fitting by minimizing the error function between the pollutant distribution and the pollution field three-dimensional data to establish candidate solutions, wherein the initial parameters are parameters set according to the state variables; a pollution diffusion path tracing result establishment module is used to perform global particle search on the candidate solutions, and perform double-layer optimization of local inversion fitting based on the global particle search results to establish pollution diffusion path tracing results.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Use NB-IoT and LoRa to connect small air quality sensors and establish a high-density monitoring network; obtain the time-series point monitoring data set of the high-density monitoring network, and distribute the pollution hotspots based on the spatial interpolation and point anomaly recognition results; configure high-precision sensors and monitoring drones in the pollution hotspots, perform pollution collection, establish three-dimensional data of the pollution field, and simultaneously obtain wind data sets; establish state variables; use the atmospheric diffusion model to predict and fit a set of initial parameters based on the wind data set of observed variables, generate pollutant distribution, and establish candidate solutions; perform global particle search on the candidate solutions, and perform double-layer optimization of local inversion fitting based on the global particle search results to establish the pollution diffusion path tracing results. The technical effect of improving the accuracy and reliability of atmospheric pollution diffusion path tracing has been achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of a method for tracing the diffusion path of atmospheric pollution based on meteorological data provided in an embodiment of the present application;

[0014] Figure 2 A schematic diagram of the structure of an atmospheric pollution diffusion path tracing system based on meteorological data provided in an embodiment of the present application.

[0015] Explanation of the reference numerals: high-density monitoring network establishment module 10, pollution hotspot area distribution module 20, pollution field three-dimensional data establishment module 30, state variable establishment module 40, candidate solution establishment module 50, pollution diffusion path tracing result establishment module 60. DETAILED DESCRIPTION

[0016] The present application provides a method and system for tracing the diffusion path of atmospheric pollution based on meteorological data, which is used to solve the technical problems in the prior art that it is difficult to accurately identify pollution hotspots and the accuracy and reliability of tracing the diffusion path of atmospheric pollution are low.

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0018] Embodiment 1, as Figure 1 As shown, the present application provides a method for tracing the diffusion path of atmospheric pollution based on meteorological data, the method comprising:

[0019] Step S100: Use NB-IoT and LoRa to connect small air quality sensors and establish a high-density monitoring network.

[0020] Specifically, first of all, we need to build a broad and sophisticated monitoring network. Two low-power wide area network communication technologies, NB-IoT (narrowband Internet of Things) and LoRa (long-range radio), are used. They have the advantages of wide coverage, strong penetration and low power consumption. Small air quality sensors have become the basic units of the monitoring network because of their small size and easy deployment. With the help of NB-IoT and LoRa technology, a large number of small air quality sensors are connected to each other. These sensors are dispersed in the target monitoring area. They are distributed according to a certain density and layout to form a high-density monitoring network. In this way, air quality data at different locations in the monitoring area can be collected comprehensively and in real time, providing rich and reliable data support for the subsequent accurate analysis of air pollution conditions, identification of pollution hotspots, and tracing of pollution diffusion paths.

[0021] Step S200: Acquire a time-series point monitoring data set of the high-density monitoring network, filter anomalies in the time-series point monitoring data set by a fixed threshold, and distribute pollution hotspot areas based on spatial interpolation and point anomaly identification results.

[0022] Specifically, after the high-density monitoring network is established, a time-series point monitoring data set is obtained from the established high-density monitoring network. The data set records the air quality data collected by the monitoring network at different time points, covering various pollution index information at many monitoring points. These data are screened using a pre-set fixed threshold, and data points that exceed or fall below the threshold are determined as abnormal data, which may indicate the emergence or change of pollution conditions. Then, using spatial interpolation technology, based on the data of known monitoring points, the air quality data of other locations that are not directly monitored in the monitoring area are inferred, so as to have a more comprehensive estimate of the pollution distribution in the entire area. At the same time, combined with the point anomaly identification results, the areas where abnormal data frequently appear and the areas where the pollution concentration is found to be high through spatial interpolation are comprehensively analyzed, and finally the pollution hotspots are determined and marked. These areas are places where air pollution is more serious and need to be focused on, providing a key basis for subsequent more accurate monitoring and analysis.

[0023] Step S300: deploy high-precision sensors and monitoring drones in pollution hotspots, perform pollution collection, establish three-dimensional data of the pollution site, and simultaneously obtain wind data sets.

[0024] Specifically, in order to obtain more accurate and comprehensive pollution information, high-precision sensors such as FTIR spectrometers and GC-MS gas analyzers are deployed in these pollution hotspots. At the same time, monitoring drones are arranged to work together. Using the pollution hotspots to establish independent collection attention, the focus and frequency of collection are determined based on factors such as the severity of pollution and the size of the hotspot area to ensure that key pollution data can be effectively captured. Then, the independent collection attention is compensated for the adjacent correlation of attention, and the possible pollution transmission or influencing factors around the hotspot area are considered to optimize the collection strategy. After that, high-precision sensors and monitoring drones begin to perform the collection of attention pollution. High-precision sensors collect detailed data of different pollutants from the ground, and monitoring drones collect data from multiple angles and heights in the air. The two are combined to construct three-dimensional data of the pollution field containing pollution information at different heights and locations. In the process of collecting pollution data, the wind data set of the area, including wind speed, wind direction and other information, is obtained synchronously with the help of meteorological monitoring equipment or related data interfaces. These data provide a key basis for the subsequent analysis of the diffusion of atmospheric pollution, and help to more accurately understand the diffusion trend of pollutants under different meteorological conditions.

[0025] Step S400: establishing state variables, and constructing observation variables using the wind data set and the pollution field three-dimensional data.

[0026] Specifically, the state variables are defined and established. These variables are quantitative representations of the sources and initial conditions of atmospheric pollution diffusion, specifically covering the location of the pollution source, the emission intensity and the emission time. The location of the pollution source is accurately marked by determining its precise spatial coordinates (such as longitude, latitude and altitude). The emission intensity is measured by the mass of pollutants released per unit time, and the emission time is accurate to a specific time point. These three variables together build a basic framework for describing the initial state of pollution. Next, the observation variables are constructed based on the obtained wind data set and the three-dimensional data of the pollution field. The wind data set contains dynamic information such as wind speed and wind direction, which directly affect the diffusion direction and speed of pollutants; the three-dimensional data of the pollution field records in detail the distribution of pollutant concentrations at different spatial locations and heights. When constructing, the two are deeply integrated. For example, the wind speed of a certain area at a specific time is combined with the pollutant concentration at the corresponding position to form a wind speed-pollutant concentration correlation variable, which is used to reflect the immediate impact of wind on pollutant concentration; according to the change of wind direction, the change of pollutant concentration at different locations is associated, and the wind direction-pollutant concentration change variable is constructed to show the diffusion trend of pollutants when the wind direction changes. By integrating these data in a variety of ways, we construct observation variables that can comprehensively reflect the real-time dynamics of atmospheric pollution diffusion, providing strong support for subsequent precise analysis using atmospheric diffusion models.

[0027] Step S500: Use the atmospheric diffusion model to predict and fit a set of initial parameters to the wind data set based on the observed variables to generate pollutant distribution, and perform inversion fitting by minimizing the error function between the pollutant distribution and the three-dimensional data of the pollution field to establish a candidate solution, wherein the initial parameters are parameters set according to the state variables.

[0028] Specifically, a set of initial parameters is set according to the previously established state variables. These initial parameters contain key information such as the location of the pollution source, emission intensity, and emission time, which preliminarily describe the starting conditions for the diffusion of atmospheric pollution. Subsequently, the atmospheric diffusion model is used to perform prediction fitting based on the wind data set in the observed variables. The atmospheric diffusion model simulates the diffusion process of pollutants in the atmosphere based on these initial parameters and wind data, and then generates the distribution of pollutants in different spatial locations. Since the pollutant distribution generated by the model may be different from the actual three-dimensional data of the pollution site, it is necessary to optimize the model by minimizing the error function between the two, calculate the difference between the generated pollutant distribution and the pollutant concentration at each corresponding position in the three-dimensional data of the pollution site, construct the error function, and then use the optimization algorithm to continuously adjust the initial parameters so that the value of the error function gradually decreases. This process is inversion fitting. After multiple iterative calculations, a set of parameter combinations that minimize the error function are obtained, and these parameter combinations constitute candidate solutions. These candidate solutions represent possible pollution source parameter settings, providing a variety of potential options for subsequent determination of the pollution diffusion path that best meets the actual situation.

[0029] Step S600: performing a global particle search on the candidate solution, and performing a double-layer optimization of local inversion fitting based on the global particle search results to establish a pollution diffusion path tracing result.

[0030] Specifically, with the candidate solution as the center, particles are evenly distributed around it. These particles constitute a candidate particle set, and each particle represents a pollution source parameter containing information such as the location of the pollution source, emission intensity, and emission time. Then the global particle search begins. During the search process, it is constantly judged whether the particle update stage meets the preset stage threshold. If it does, the position of the particle in the candidate particle set is updated according to the distance constraint. Before updating the position, the pollutant distribution of each particle must be predicted first, and the particle weight is calculated according to the prediction result. The particle weight is used as the particle fitness value, and the position update is completed by combining the particle shrinkage factor, random perturbation amplitude and update interval, and the update result is recorded in the regional exploration window. When the particle update stage does not meet the preset stage threshold, the exploration result of the regional exploration window is called for regional elimination identification, and the particles that do not meet the requirements are eliminated from the candidate particle set. The retained particles are used to continue the update search, and finally the global particle search result is obtained. Based on the global particle search result, a double-layer optimization of local inversion fitting is performed. Through the fitness value of the particles in the global particle search result, the candidate solution is updated and tracked to find the particles or particle combinations that are most likely to reflect the actual pollution situation, and the update tracking identification results are established. Then, the result is used to optimize the location of the candidate solution, further adjust the pollution source parameters, and complete the double-layer optimization of the local inversion fitting. After this series of operations, the global and local optimization information are comprehensively considered to finally determine the pollution diffusion path tracing result that best meets the actual situation, clearly presenting the diffusion trajectory of pollutants from the source, the diffusion time, and the diffusion degree at different stages and other key information.

[0031] In a possible implementation, step S600 further includes:

[0032] Step S610: Taking the candidate solution as the center, uniformly distribute particles to establish a candidate particle set, wherein each particle in the candidate particle set represents a pollution source parameter, and the pollution source parameter includes the pollution source location, emission intensity, and emission time.

[0033] Step S620: Determine whether the update stage of the particle meets a preset stage threshold.

[0034] Step S630: If the update stage of the particle meets the preset stage threshold, the position of the candidate particle set is updated through the distance constraint and recorded in the area exploration window.

[0035] Step S640: When the update stage of the particle does not meet the preset stage threshold, the exploration result of the regional exploration window is called to perform regional elimination identification, and the corresponding particles of the candidate particle set are eliminated using the regional elimination identification result, and the update search is continued based on the retained particles to establish a global particle search result.

[0036] Specifically, after completing a series of operations in the early stage to obtain a candidate solution, the particles are evenly distributed in the multidimensional space with the candidate solution as the center. Assume that the pollution source position represented by the candidate solution is a specific spatial coordinate point, and the emission intensity and emission time also have corresponding values. Based on these values, the properties of each particle are determined according to specific rules within a certain range around the candidate solution. For the pollution source position, in three-dimensional space, the position coordinates of different particles are generated by making slight offsets along each coordinate axis direction according to equal spacing or specific distribution patterns based on the position of the candidate solution; similar operations are also performed for the emission intensity and emission time. Around the corresponding values ​​of the candidate solution, a series of different values ​​are generated according to certain intervals or distribution rules, and assigned to each particle. Each particle is assigned a unique set of pollution source parameters, which include the precise pollution source location, specific emission intensity, and accurate emission time. In this way, the constructed candidate particle set covers many potential pollution source parameter combinations around the candidate solution.

[0037] In the process of searching and optimizing particles, each particle has its own specific update stage, which records the number of updates or iterations that the particle has experienced from the initial state to the current state. The preset stage threshold is a standard value set in advance based on the research experience of atmospheric pollution diffusion problems, computing resources, and expected computing accuracy. The current update stage value of each particle is obtained in real time and compared with the preset stage threshold. Through this judgment operation, it is determined whether the particle has reached the stage node where a specific operation needs to be performed.

[0038] When the update stage of the particle reaches the preset stage threshold, the position of the candidate particle set is updated according to the distance constraint. As a restriction rule, the distance constraint ensures that the updated position of the particle conforms to the actual physical laws and logic of atmospheric pollution diffusion, and avoids unreasonable large jumps of particles. During the update process, each particle in the candidate particle set is calculated one by one, and its new position is determined by comprehensively considering the current position of the particle, the surrounding environmental factors, and the relationship with other particles. After the update is completed, the new position information of these particles will be recorded in the regional exploration window, which stores the interim results of the particles in the exploration process and provides data support for subsequent analysis.

[0039] When the particle update stage does not meet the preset stage threshold, at this time, the exploration results stored in the regional exploration window are called to perform regional elimination identification. Regional elimination identification is a screening process. By analyzing factors such as the position changes of particles in the previous exploration process, the association with other particles, and the impact on the overall pollution diffusion model, it is determined which particles have deviated from the possible optimal solution and which particles have low exploration direction value. Based on the results of regional elimination identification, the corresponding low-value particles in the candidate particle set are eliminated, and only those particles that are more likely to be close to the actual pollution source parameters are retained. Then, these retained particles are used to continue the update search, constantly adjusting the position of the particles and the pollution source parameters they represent, and continuously exploring better solutions. After many such iterative operations, a global particle search result is finally established, providing a key basis for accurately tracing the diffusion path of atmospheric pollution.

[0040] In a possible implementation, step S630 further includes:

[0041] Step S631: perform pollutant distribution prediction on each particle in the candidate particle set and establish a prediction result.

[0042] Step S632: Calculate the weight of each particle according to the prediction result, as follows:

[0043] ;

[0044] in, Characterizing particles The weight of Characterize the actual observed pollutant concentration data obtained from the three-dimensional data of the pollution site, Characterizing particles The simulated pollutant concentration is Characterizes the standard deviation of the observation noise.

[0045] Step S633: Using the particle weight as the particle fitness value to perform position update of the candidate particle set.

[0046] Specifically, in order to predict the pollutant distribution of each particle in the candidate particle set and establish the prediction results, the following specific implementation methods are adopted: First, extract the pollution source parameters represented by each particle, including the precise pollution source location (such as latitude and longitude and altitude determined based on the geographic coordinate system), emission intensity (amount of pollutant emissions per unit time) and emission time (accurate to a specific time point). Then, combined with the meteorological data collected in the early stage, especially the wind data set, the wind speed and wind direction information are obtained, and other meteorological monitoring equipment is used to obtain atmospheric temperature, humidity, air pressure and other data to determine the atmospheric stability. Based on this information, a suitable atmospheric diffusion model, such as the AERMOD model, is selected. In the model, the pollution source position of the particle is taken as the starting point, and the initial emission flux of the pollutant is determined according to the emission intensity. The dominant direction of pollutant diffusion is set according to the wind direction, and the transmission speed of the pollutant in this direction is calculated using the wind speed. Combined with the atmospheric stability, the parameters used in the model to describe the degree of diffusion of pollutants in the horizontal and vertical directions are determined. For example, when the stability is high, the diffusion of pollutants is relatively limited, corresponding to a smaller diffusion parameter; conversely, when it is unstable, the diffusion parameter is larger. Using these parameters, the model calculates the pollutant concentrations at various spatial points within a certain range centered on the pollution source at different times. The calculated pollutant concentration data at different times and spatial points are integrated to form the pollutant distribution prediction results for each particle. These results are presented in the form of three-dimensional data, covering spatial position (x, y, z coordinates) and time dimensions, and intuitively show the diffusion of pollutants in space over time, providing key data support for the subsequent calculation of particle weights and updating of particle positions, and facilitating the accurate tracing of the diffusion path of atmospheric pollution.

[0047] Then, the weight of each particle is calculated based on the prediction results. The calculation formula is: , in this formula, Represents particles The weight of the particle directly reflects the degree of fit between the pollution source parameter combination represented by the particle and the actual pollution situation. It is the actual observed pollutant concentration data obtained from the three-dimensional data of the pollution site, which is a direct reflection of the real pollution situation; Representative particles Pollutant concentrations simulated by the model; Represents the standard deviation of observation noise, which measures the uncertainty of observation data. According to the formula principle, the smaller the difference between the actual observed concentration and the simulated concentration, the larger the value of the exponential part and the higher the weight of the particle, which means that the pollution source parameter combination represented by the particle is closer to the actual situation.

[0048] The weight of the particle is used as the particle fitness value, and the position update operation of the candidate particle set is performed based on this as the core basis. The particle fitness value represents the degree of fit between the pollution source parameter combination represented by the particle and the actual atmospheric pollution situation. The larger the weight, the higher the fitness value, which means that the parameter combination represented by the particle is closer to the actual pollution source situation. First, the particle contraction factor is determined according to the particle fitness value. For particles with high fitness values, the contraction factor setting will be relatively small, the purpose is to make the particle move less when the position is updated, so as to maintain the exploration in the area close to the real solution; while for particles with low fitness values, the contraction factor is relatively large, so that it has a larger range of movement to explore new areas and increase the possibility of finding a better solution. Create an update interval based on the current particle position and distance constraint. The distance constraint is set according to the actual physical laws and relevant experience of atmospheric pollution diffusion to ensure that the updated position of the particle is within a reasonable range and avoid jumps that do not conform to the actual situation. This update interval limits the range boundary that can be changed when the particle position is updated. At the same time, configure the random perturbation amplitude. The random perturbation amplitude introduces a certain degree of randomness to the particle position update to prevent the particle from falling into the local optimal solution. When updating the position, the three elements of particle shrinkage factor, random disturbance amplitude and update interval are comprehensively considered. For each particle, the moving step is adjusted according to its own shrinkage factor, and then a direction is randomly selected within the update interval to move in combination with the random disturbance amplitude, thereby completing the position update of the candidate particle set. In this way, the particles continuously adjust their positions in the search space, gradually approaching the real pollution source parameters, laying the foundation for accurately tracing the diffusion path of atmospheric pollution.

[0049] In a possible implementation, step S633 further includes:

[0050] Step S6331: configuring the particle shrinkage factor according to the particle fitness value.

[0051] Step S6332: Create an update interval based on the current particle position and the distance constraint.

[0052] Step S6333: configure the random perturbation amplitude, and update the position of the candidate particle set according to the random perturbation amplitude, the particle shrinkage factor, and the update interval.

[0053] Specifically, the particle fitness value is determined based on the previously calculated particle weight, which reflects the degree of match between the pollution source parameter combination represented by the particle and the actual atmospheric pollution situation. The particle shrinkage factor is configured according to this fitness value. For particles with higher fitness values, this means that the pollution source parameter combination it represents is closer to the actual situation. In order to prevent particles from deviating from this possible correct area in subsequent searches, a smaller particle shrinkage factor will be set so that the particle moves relatively less when the position is updated, and can more finely explore a better solution near the current area; conversely, for particles with lower fitness values, it means that their current pollution source parameter combination deviates greatly from the actual situation. At this time, a larger particle shrinkage factor will be configured to allow particles to move in a larger range, increasing the chance of finding a better solution.

[0054] Get the specific location information of each particle in space. This location information accurately identifies the coordinates of the particle in the current search space. For example, in a three-dimensional space model, the coordinate values ​​are composed of longitude, latitude and altitude. The distance constraint is set according to the actual physical laws of atmospheric pollution diffusion, past research experience and the specific needs of this task. It specifies the maximum distance that a particle can move when updating its position. For example, considering the diffusion speed and range limitations of pollutants in the atmosphere, and the fact that pollution sources will not appear instantly in distant areas in real scenes, a reasonable distance threshold is set. Take the current particle position as the reference point and define the range in all directions according to the distance constraint. On a two-dimensional plane, take the particle position as the center of the circle and the distance constraint value as the radius, and expand to the surroundings to form a circular or square area (depending on the specific setting method); in three-dimensional space, take the particle position as the center of the sphere and the distance constraint value as the radius to construct a sphere, or expand according to the distance constraint value in the three coordinate axis directions to form a rectangular area. This area is the update interval created for particle position updates. It limits the range that particles can reach during subsequent position updates, ensuring that particle movement is both exploratory and in line with actual conditions, avoiding large jumps that are inconsistent with physical laws, and laying the foundation for more accurate determination of particle positions and tracing of pollution diffusion paths.

[0055] Configure the random perturbation amplitude. The random perturbation amplitude introduces a certain degree of randomness to the particle position update to prevent the particle from falling into the local optimal solution during the search process. In actual operation, the position of the candidate particle set is updated according to the particle contraction factor and the update interval, combined with the configured random perturbation amplitude. For each particle, first adjust its moving step size according to the particle contraction factor, and then randomly select a direction to move according to the random perturbation amplitude within the update interval. In this way, particles continuously adjust their positions in the search space, continuously explore pollution source parameters that are more in line with the actual situation, and gradually approach the real pollution diffusion path, providing strong support for accurately tracing the atmospheric pollution diffusion path.

[0056] In a possible implementation, step S640 further includes:

[0057] Step S641: Obtain the flight trajectory of the monitoring drone, and calibrate the particle with the maximum fitness value among the retained particles as the standard particle.

[0058] Step S642: Establish trajectory impact compensation according to the flight trajectory and the standard particles.

[0059] Step S643: updating the search constraints of the retained particles through the trajectory impact compensation.

[0060] Specifically, the flight trajectory of the monitoring drone is obtained. The monitoring drone will record its own flight path during the mission. This trajectory contains the spatial position information of the drone at different time points, reflecting its range of activities and path changes over the polluted area. At the same time, the retained particles are evaluated, and the particles with the largest fitness value are found and calibrated as standard particles. This standard particle represents the particle that is most likely to be close to the parameters of the real pollution source determined in the current search process. Its high fitness value means that the combination of parameters such as the pollution source location, emission intensity and emission time represented by it has the best match with the actual pollution situation.

[0061] First, extract the spatial position information of the monitoring drone at each time point from the flight trajectory data. This information accurately records the movement path of the drone over the polluted area. At the same time, clarify the key parameters such as the pollution source location, emission intensity and emission time represented by the standard particles. Based on the atmospheric diffusion model, combined with the currently acquired meteorological data, such as wind speed, wind direction, temperature, humidity, etc., the pollution source parameters of the standard particles are used as the starting conditions to simulate the diffusion process of pollutants in the atmosphere from the emission time, so as to obtain the concentration distribution of pollutants at different time and space points. For each point on the flight trajectory of the drone, calculate its distance from the pollution source position, determine its relative position in the direction of pollution diffusion, and obtain the simulated pollutant concentration corresponding to the point. According to these data, an impact factor is assigned to each point. The closer the distance to the pollution source, the point located in the main direction of pollution diffusion and the higher the simulated pollutant concentration, the greater its impact factor. Integrate the impact factors of all points on the flight trajectory, and construct a model that describes the impact of the flight trajectory on particle search through data analysis and processing. This is trajectory impact compensation. When the retained particles are subsequently updated and searched, this trajectory impact compensation mechanism is used to adjust the particle search direction, step size and other parameters according to the spatial position and time of the particles, so that the particle search process is more in line with the actual pollution diffusion situation, thereby improving the accuracy and efficiency of tracing the atmospheric pollution diffusion path.

[0062] When the retained particles are ready for update search, firstly, according to the current spatial position and time information of each retained particle, a matching query is performed in the trajectory impact compensation model to obtain the corresponding impact parameters. These impact parameters reflect the degree to which the area where the particle is located is affected by the UAV flight trajectory and the standard particle (representing the current optimal pollution source parameter combination). If the particle is in an area that is greatly affected by pollution in the UAV flight trajectory, and the trajectory impact compensation shows that this area is of great significance to finding the real pollution source, then when updating the search, the particle's movement direction will tend to be closer to the pollution source position represented by the standard particle, and the moving step size may be appropriately reduced to explore the area more finely and increase the probability of finding a better solution. On the contrary, if the particle is in an area that is less affected, the adjustment range of its movement direction and step size is relatively small, maintaining a certain exploration range to avoid excessive deviation from the possible optimal solution area. In this way, trajectory impact compensation sets reasonable rules and restrictions for the update search of retained particles, guides particles to explore more efficiently in the search space, avoids blind search, and makes the particle update search more in line with the actual pollution diffusion situation, which helps to determine the pollution source parameters faster and more accurately, and realizes the accurate tracing of the atmospheric pollution diffusion path.

[0063] In a possible implementation, step S200 further includes:

[0064] Step S210: Acquire spatial data of the space covered by the high-density monitoring network, and read backtracking wind data.

[0065] Step S220: performing spatial correlation analysis based on the spatial data and the retrospective wind data.

[0066] Step S230: performing abnormality clustering on the point abnormality recognition results to establish abnormality clustering results.

[0067] Step S240: performing spatial interpolation using the abnormal clustering results and spatial correlation analysis results to distribute pollution hotspot areas.

[0068] Specifically, the detailed spatial data of the space covered by the high-density monitoring network is obtained. These data contain information such as the geographic coordinates and topographic features of the area covered by the monitoring network, accurately depicting the spatial form of the monitoring area. At the same time, the retrospective wind data is read. This part of the data records the wind force and wind direction changes in the monitoring area over a period of time in the past, providing basic meteorological data for subsequent analysis.

[0069] The geographical location information in the spatial data is matched with each time node in the retrospective wind data. With longitude and latitude as coordinate reference, the wind conditions at different geographical locations at different times are studied. For example, the wind speed changes in mountainous areas and urban plains in the same time period are compared to analyze the impact of terrain on wind speed. For wind direction, visualization methods such as drawing wind rose diagrams are used to observe the differences in dominant wind directions in different regions and the changing patterns of wind direction with geographical location. Statistical analysis methods are used to calculate the correlation coefficient between spatial position and wind elements and quantify the degree of correlation between the two. For example, the Pearson correlation coefficient between wind speed and altitude is calculated to determine the changing trend of wind speed with altitude. At the same time, with the help of the spatial analysis function of the geographic information system (GIS), the spatial distribution characteristics of wind power and the possible impact of wind power on the diffusion of pollutants are analyzed. By combining the results of various aspects of analysis, the intrinsic connection between spatial data and retrospective wind power data is sorted out, providing important data support and analysis basis for the subsequent determination of pollution hotspots.

[0070] Process the previous point anomaly identification results to establish anomaly clustering results. In the early stage, a series of point anomaly data have been identified by screening the time-series point monitoring data set of the high-density monitoring network with a fixed threshold. In this step, a clustering algorithm is used, such as the density-based spatial clustering algorithm DBSCAN. The algorithm scans all point anomaly identification results and clusters them according to the distribution density of data points in space and the distance relationship between them. First, a core point is determined, that is, a point with a density reaching a certain threshold. Around the core point, other points within a specific range are classified into the same category. If a point is not a core point, but its distance from a core point is within the specified range, it will also be classified into the corresponding category. In this way, the clustering range is continuously expanded, and abnormal points with similar spatial locations and similar characteristics are clustered together. For example, in the monitoring area, if multiple abnormal points are concentrated in geographical location and the pollutant concentration changes reflected by them are similar, they will be clustered into one category. Finally, after being processed by the algorithm, the originally scattered point anomaly identification results were summarized into multiple different clusters. These clustering results clearly showed the spatial aggregation pattern of the abnormal data, providing key clues for the subsequent identification of pollution hotspots, and helping to more accurately locate and analyze areas with severe air pollution.

[0071] Spatial interpolation is performed using the anomaly clustering results and spatial correlation analysis results obtained in the previous period to determine the distribution of pollution hotspots. The anomaly clustering results show the spatial aggregation of anomaly data and clarify which areas have concentrated anomalies. These areas are likely to be key areas of pollution. The results of spatial correlation analysis reveal the intrinsic connection between spatial location and factors such as wind power, such as the difference in the impact of wind power in different regions on the diffusion of pollutants. When performing spatial interpolation, first determine the diffusion trend and speed of pollutants in different areas based on the results of spatial correlation analysis and combined with wind data. For example, if the spatial correlation analysis shows that the wind direction and wind speed in a certain area will cause pollutants to diffuse in a specific direction, this factor will be considered when performing spatial interpolation. Then, with the anomaly clustering area in the anomaly clustering results as the core, the spatial interpolation algorithm (such as inverse distance weighted interpolation, Kriging interpolation, etc.) is used to estimate the area in the monitoring area where data is not directly monitored based on the known monitoring point data and spatial correlation information. Reasonable pollution values ​​are assigned to the target area based on factors such as the distance between the surrounding monitoring points and the target area and the strength of the correlation. The closer the area is to the center of the abnormal cluster and the stronger the correlation with the surrounding monitoring points, the greater the impact of the abnormal clustering results, and the higher the estimated pollution value; vice versa. In this way, the discrete monitoring data and abnormal clustering information are extended to the entire monitoring area to form a continuous pollution distribution model, thereby clearly distributing the pollution hotspots. These pollution hotspots are areas where air pollution is more serious and need to be focused on and governed, providing a key basis for the subsequent tracing of air pollution diffusion paths and pollution prevention and control.

[0072] In a possible implementation, step S300 further includes:

[0073] Step S310: The high-precision sensor includes a FTIR spectrometer and a GC-MS gas analyzer.

[0074] Specifically, the high-precision sensors configured in pollution hotspots include FTIR spectrometers and GC-MS gas analyzers. FTIR spectrometer, or Fourier transform infrared spectrometer, analyzes the composition of substances by measuring the absorption of infrared rays of different wavelengths. In the atmospheric pollution monitoring scenario, qualitative and quantitative analysis of various organic and inorganic pollutants in the air of pollution hotspots can be performed, the types of pollutants can be identified based on the characteristic absorption peaks of different pollutants, and the concentration of pollutants can be determined by the intensity of the absorption peaks. GC-MS gas analyzer, or gas chromatography-mass spectrometry, first uses gas chromatography to separate the different components in the mixed gas, and then uses a mass spectrometer to accurately qualitatively and quantitatively detect the separated components. It has extremely high detection sensitivity for volatile organic pollutants and can effectively detect trace amounts of organic pollutants in the atmosphere. These two high-precision sensors play a key role in pollution hotspots, helping to obtain more accurate pollutant information.

[0075] In a possible implementation, step S300 further includes:

[0076] Step S310: Using pollution hotspots to establish independent collection attention.

[0077] Step S320: After the independent collection attention is compensated for the adjacent association of attention, attention contamination collection is performed.

[0078] Specifically, first, based on the pollution hotspots identified by the previous operations of fixed threshold screening of time-series point monitoring data set anomalies, spatial interpolation, and point anomaly identification, the independent collection attention is constructed. Since the pollution status of these pollution hotspots is crucial for accurate tracing paths, when establishing independent collection attention, the severity of pollution in different parts of the pollution hotspots, pollution change trends, and spatial locations will be comprehensively considered. For areas with high pollution concentrations and large fluctuations, a higher collection attention is given, which means that the monitoring frequency and detail of the area will be increased in the future, and more abundant data will be obtained to accurately analyze the pollution situation; while for areas with relatively light pollution and relatively stable changes, a relatively low collection attention is given, but a certain monitoring intensity is still maintained. In this way, an independent collection attention system is established for pollution hotspots, making subsequent monitoring and collection work more targeted and efficient, and laying a solid foundation for accurately obtaining pollution data and accurately establishing three-dimensional data of pollution sites.

[0079] Since pollution is diffuse in space, the pollution status of an area is closely related to the adjacent areas. Therefore, it is necessary to consider the relationship between each monitoring area and the surrounding adjacent areas to compensate for the adjacent correlation of independent collection attention. For example, if the independent collection attention of a certain area is originally high, and there are signs of pollution in its adjacent areas, then based on the continuity of pollutant diffusion, the attention of this area should be further improved; conversely, if the pollution level of the adjacent area is low and the possibility of pollution impact on it is small, then the adjustment range of the attention of this area should be appropriately reduced. After completing the adjacent correlation compensation of attention, the collection of attention pollution is started. At this time, monitoring resources are allocated in a targeted manner according to the adjusted attention. For areas with high attention, increase the frequency and accuracy of monitoring, use high-precision sensors for more detailed monitoring, such as using FTIR spectrometers and GC-MS gas analyzers to accurately analyze the composition of pollutants, and use monitoring drones to obtain data from different angles and heights; for areas with relatively low attention, maintain a certain monitoring intensity to ensure that potential pollution changes can be discovered in a timely manner. Through this method of collecting pollution data based on adjacent correlation compensation of attention degree, pollution data of pollution hotspots can be obtained more comprehensively and accurately, providing strong support for establishing accurate three-dimensional data of pollution sites, thereby facilitating the precise tracing of subsequent atmospheric pollution diffusion paths.

[0080] In a possible implementation, step S600 further includes:

[0081] Step S650: updating tracking and identifying candidate solutions through the particle fitness values ​​of the global particle search results, and establishing updated tracking and identifying results.

[0082] Step S660: Utilize the updated tracking identification result to optimize the position of the candidate solution to complete the double-layer optimization of the local inversion fitting.

[0083] Specifically, after completing the global particle search, a series of particle fitness values ​​will be obtained. Work is carried out based on these particle fitness values, aiming to update, track and identify candidate solutions and establish corresponding results. The particle fitness value reflects the degree of fit between the combination of pollution source parameters (such as pollution source location, emission intensity, and emission time) represented by each particle and the actual pollution situation. Based on these fitness values, focus on particles with higher fitness values, because the candidate solutions corresponding to them are more likely to be close to the actual pollution source parameters. By comparing the changing trends of the fitness values ​​of different particles, observe which candidate solutions show better adaptive changes during the search process, such as the fitness value gradually increases or remains at a high level within a certain range. These candidate solutions with good trends are recorded in detail, including their parameter changes at different search stages, their spatial locations and other information, so as to establish updated tracking and identification results.

[0084] A genetic algorithm is used to optimize the position of candidate solutions using the update tracking identification results to complete the double-layer optimization of local inversion fitting. First, the candidate solutions in the update tracking identification results are regarded as individuals in the genetic algorithm, and the parameters of each candidate solution such as the pollution source location, emission intensity, and emission time are the genes of the individual. Then, the population is initialized, and a certain number of candidate solutions are selected from the update tracking identification results to form the initial population. In the evolutionary process of each generation, the selection operator is used to select better individuals based on the fitness value of the individual (which can be determined by comparing the difference between the simulated pollutant distribution and the three-dimensional data of the pollution field. The smaller the difference, the higher the fitness value). For example, using the roulette selection method, individuals with high fitness are more likely to be selected. Then, the selected individuals are cross-operated, two individuals are randomly selected, and some of their genes are exchanged to generate new candidate solutions, thereby simulating genetic recombination in biological evolution and increasing the diversity of the population. At the same time, the individuals are mutated with a certain probability to randomly change some genes of the individuals to avoid the algorithm from falling into the local optimal solution. After multiple generations of evolution, the individuals in the population gradually develop in a better direction, that is, the position of the candidate solution is continuously optimized. When the preset termination conditions are met, such as the number of evolutionary generations reaches the set value or the fitness value of the population no longer increases significantly, the optimal individual obtained at this time is the optimized candidate solution, successfully completing the double-layer optimization of local inversion fitting, providing a more accurate basis for accurately tracing the diffusion path of atmospheric pollution.

[0085] Embodiment 2 is based on the same inventive concept as the method for tracing the diffusion path of atmospheric pollution based on meteorological data in the above embodiment. Figure 2 As shown, the present application provides an atmospheric pollution diffusion path tracing system based on meteorological data, and the system and method embodiments in the present application embodiments are based on the same inventive concept. The system includes:

[0086] The high-density monitoring network establishment module 10 is used to connect small air quality sensors using NB-IoT and LoRa to establish a high-density monitoring network.

[0087] The pollution hotspot area distribution module 20 is used to obtain the time-series point monitoring data set of the high-density monitoring network, filter the anomalies of the time-series point monitoring data set through a fixed threshold, and distribute the pollution hotspot areas based on spatial interpolation and point anomaly identification results.

[0088] The pollution field three-dimensional data establishment module 30 is used to configure high-precision sensors and monitoring drones in pollution hotspots, perform pollution collection, establish pollution field three-dimensional data, and simultaneously obtain wind data sets.

[0089] The state variable establishing module 40 is used to establish state variables and construct observation variables using the wind data set and the pollution field three-dimensional data.

[0090] The candidate solution establishment module 50 is used to use the atmospheric diffusion model to predict and fit a set of initial parameters based on the wind data set of observed variables to generate pollutant distribution, and to perform inversion fitting by minimizing the error function between the pollutant distribution and the three-dimensional data of the pollution field to establish candidate solutions, wherein the initial parameters are parameters set according to the state variables.

[0091] The pollution diffusion path tracing result establishing module 60 is used to perform a global particle search on the candidate solution, and perform a double-layer optimization of local inversion fitting based on the global particle search results to establish the pollution diffusion path tracing result.

[0092] Furthermore, the system is also used to implement the following functions:

[0093] Taking the candidate solution as the center, particles are evenly distributed to establish a candidate particle set, wherein each particle in the candidate particle set represents a pollution source parameter, and the pollution source parameter includes pollution source location, emission intensity, and emission time; whether the update stage of the particle meets the preset stage threshold is judged; if the update stage of the particle meets the preset stage threshold, the position of the candidate particle set is updated through distance constraints and recorded in the regional exploration window; when the update stage of the particle does not meet the preset stage threshold, the exploration result of the regional exploration window is called to perform regional elimination identification, and the corresponding particles of the candidate particle set are eliminated using the regional elimination identification result, and the search is continued to be updated according to the retained particles to establish a global particle search result.

[0094] Furthermore, the system is also used to implement the following functions:

[0095] The pollutant distribution of each particle in the candidate particle set is predicted to establish a prediction result; the weight of each particle is calculated according to the prediction result, as follows: ;in, Characterizing particles The weight of Characterize the actual observed pollutant concentration data obtained from the three-dimensional data of the pollution site, Characterizing particles The simulated pollutant concentration is Characterize the standard deviation of observation noise; use the particle weight as the particle fitness value to perform position update of the candidate particle set.

[0096] Furthermore, the system is also used to implement the following functions:

[0097] A particle shrinkage factor is configured according to the particle fitness value; an update interval is created based on the current particle position and the distance constraint; a random disturbance amplitude is configured, and the position of the candidate particle set is updated according to the random disturbance amplitude, the particle shrinkage factor, and the update interval.

[0098] Furthermore, the system is also used to implement the following functions:

[0099] The flight trajectory of the monitoring UAV is obtained, and the particle with the maximum fitness value among the retained particles is calibrated as the standard particle; a trajectory impact compensation is established according to the flight trajectory and the standard particle; and the update search constraint of the retained particles is performed through the trajectory impact compensation.

[0100] Furthermore, the system is also used to implement the following functions:

[0101] Acquire spatial data of the space covered by the high-density monitoring network, and read the retrospective wind data; perform spatial correlation analysis based on the spatial data and the retrospective wind data; perform anomaly clustering on the point anomaly identification results to establish anomaly clustering results; perform spatial interpolation using the anomaly clustering results and spatial correlation analysis results to distribute pollution hotspots.

[0102] Furthermore, the system is also used to implement the following functions:

[0103] The high-precision sensors include a FTIR spectrometer and a GC-MS gas analyzer.

[0104] Furthermore, the system is also used to implement the following functions:

[0105] An independent collection focus is established using the pollution hotspot area; after the independent collection focus is compensated for the adjacent association of the focus, the pollution collection is performed.

[0106] Furthermore, the system is also used to implement the following functions:

[0107] The candidate solutions are updated and tracked by the particle fitness values ​​of the global particle search results to establish an updated and tracked identification result; the positions of the candidate solutions are optimized by using the updated and tracked identification result to complete the double-layer optimization of the local inversion fitting.

[0108] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0109] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0110] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for tracing the diffusion path of atmospheric pollution based on meteorological data, characterized in that: The method comprises: Use NB-IoT and LoRa to connect small air quality sensors and establish a high-density monitoring network; Acquire the time-series point monitoring data set of the high-density monitoring network, filter the anomalies of the time-series point monitoring data set by a fixed threshold, and distribute the pollution hotspot areas based on the spatial interpolation and point anomaly identification results; High-precision sensors and monitoring drones are deployed in pollution hotspots to collect pollution data, establish three-dimensional data of pollution sites, and simultaneously obtain wind data sets; Establishing state variables, and constructing observation variables with the wind data set and the three-dimensional data of the pollution field; Using an atmospheric diffusion model, a set of initial parameters are used to predict and fit the wind data set based on the observed variables to generate pollutant distribution, and inverse fitting is performed by minimizing the error function between the pollutant distribution and the three-dimensional data of the pollution field to establish a candidate solution, wherein the initial parameters are parameters set according to the state variables; A global particle search is performed on the candidate solution, and a double-layer optimization of local inversion fitting is performed based on the global particle search results to establish a pollution diffusion path tracing result.

2. The method for tracing the diffusion path of atmospheric pollution based on meteorological data according to claim 1, characterized in that: The performing a global particle search on the candidate solution comprises: Taking the candidate solution as the center, uniformly distributing particles to establish a candidate particle set, wherein each particle in the candidate particle set represents a pollution source parameter, and the pollution source parameter includes a pollution source location, emission intensity, and emission time; Determine whether the particle's update phase meets a preset phase threshold; If the update stage of the particle meets the preset stage threshold, the position of the candidate particle set is updated through the distance constraint and recorded in the area exploration window; When the update stage of the particle does not meet the preset stage threshold, the exploration result of the regional exploration window is called to perform regional elimination identification, the corresponding particles of the candidate particle set are eliminated using the regional elimination identification result, and the update search is continued based on the retained particles to establish a global particle search result.

3. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 2, characterized in that: The updating of the position of the candidate particle set by the distance constraint also includes: Predict pollutant distribution for each particle in the candidate particle set and establish prediction results; The weight of each particle is calculated according to the prediction results, as follows: ； in, Characterizing particles The weight of Characterize the actual observed pollutant concentration data obtained from the three-dimensional data of the pollution site, Characterizing particles The simulated pollutant concentration is Characterizes the standard deviation of the observation noise; The weight of the particle is used as the particle fitness value to perform the position update of the candidate particle set.

4. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 3, characterized in that: The method of using the particle weight as the particle fitness value to perform position update of the candidate particle set includes: configuring a particle shrinkage factor according to the particle fitness value; Creating an update interval based on the current particle position and the distance constraint; The random disturbance amplitude is configured, and the position of the candidate particle set is updated according to the random disturbance amplitude, the particle shrinkage factor, and the update interval.

5. The method for tracing the diffusion path of atmospheric pollution based on meteorological data according to claim 2, characterized in that: The continuing updating search according to the retained particles includes: Obtain the flight trajectory of the monitoring drone, and calibrate the particle with the maximum fitness value among the retained particles as the standard particle; Establishing trajectory impact compensation according to the flight trajectory and the standard particles; The updated search constraints for the retained particles are performed by the trajectory impact compensation.

6. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 1, characterized in that: The distribution of pollution hotspot areas based on spatial interpolation and point anomaly identification results includes: Obtain spatial data of the space covered by the high-density monitoring network and read backtracking wind data; Performing spatial correlation analysis based on the spatial data and the retrospective wind data; Performing abnormality clustering on the point abnormality recognition results to establish abnormality clustering results; The abnormal clustering results and spatial correlation analysis results are used to perform spatial interpolation to distribute pollution hot spots.

7. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 1, characterized in that: The high-precision sensors include a FTIR spectrometer and a GC-MS gas analyzer.

8. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 1, characterized in that: The execution focuses on pollution collection and establishes three-dimensional data of the pollution field, including: Use pollution hotspots to establish independent collection attention; After the independent collection attention is compensated for the adjacent association of attention, attention contamination collection is performed.

9. The method for tracing the diffusion path of air pollution based on meteorological data according to claim 1, characterized in that: The double-layer optimization of performing local inversion fitting according to the global particle search results includes: Performing update tracking and identification of candidate solutions through the particle fitness values ​​of the global particle search results, and establishing an update tracking and identification result; The update tracking identification result is used to optimize the position of the candidate solution to complete the double-layer optimization of the local inversion fitting.

10. The atmospheric pollution diffusion path tracing system based on meteorological data is characterized by: The system is used to implement the atmospheric pollution diffusion path tracing method based on meteorological data according to any one of claims 1 to 9, and the system comprises: High-density monitoring network establishment module, used to connect small air quality sensors using NB-IoT and LoRa to establish a high-density monitoring network; A pollution hotspot area distribution module is used to obtain the time-series point monitoring data set of the high-density monitoring network, filter the anomalies of the time-series point monitoring data set through a fixed threshold, and distribute the pollution hotspot areas based on spatial interpolation and point anomaly identification results; The pollution site 3D data establishment module is used to configure high-precision sensors and monitoring drones in pollution hotspots, perform pollution collection, establish pollution site 3D data, and simultaneously obtain wind data sets; A state variable establishment module, used to establish state variables and construct observation variables with the wind data set and the pollution field three-dimensional data; A candidate solution establishment module is used to use an atmospheric diffusion model to perform a prediction fit of a set of initial parameters based on a wind data set of observed variables to generate a pollutant distribution, and to perform inversion fitting by minimizing an error function between the pollutant distribution and the three-dimensional data of the pollution field to establish a candidate solution, wherein the initial parameters are parameters set according to the state variables; The pollution diffusion path tracing result establishing module is used to perform a global particle search on the candidate solution, and perform a double-layer optimization of local inversion fitting based on the global particle search results to establish the pollution diffusion path tracing result.

Images