A bioaerosol sensor network monitoring method based on multi-source data analysis
Through multi-source data analysis and dynamic aerosol sensing network monitoring methods, the problem that a single data source is difficult to reflect the aerosol diffusion characteristics is solved, and more accurate and efficient aerosol monitoring is achieved, and the real-time response capability of the monitoring network is improved.
Patent Information
- Application Number
- CN202411199202.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-08-29
AI Technical Summary
The existing bioaerosol monitoring methods rely on a single data source and cannot fully reflect the aerosol diffusion characteristics, especially in dynamic environments, which is difficult to reflect aerosol changes in real time, resulting in insufficient monitoring accuracy and reliability.
Using a bioaerosol sensing network monitoring method based on multi-source data analysis, the target meteorological station and its adjacent station data are obtained for spatial and temporal registration, a unified meteorological time series is generated, and the meteorological conditions standardization process and environmental field simulation are carried out to build multiple The hierarchical diffusion network performs perturbation diffusion simulation and prediction to form a dynamic aerosol sensing network.
Improve the accuracy and reliability of aerosol monitoring data, can reflect the aerosol distribution more realistically, enhance the accuracy and applicability of diffusion simulation, capture the aerosol changes in real time, and improve the real-time response and efficiency of the monitoring network. sex.
Smart Images

Figure CN118707053B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of aerosol monitoring, and in particular to a bioaerosol sensor network monitoring method based on multi-source data analysis. Background Art
[0002] The importance of bioaerosols in environmental monitoring is becoming increasingly prominent, especially in the fields of public health and biosafety. Traditional monitoring methods mainly rely on a single data source, such as meteorological data or aerosol concentration data. However, these methods often cannot fully reflect the complex aerosol diffusion characteristics, especially in dynamic environments. Therefore, it is particularly important to propose a monitoring method that can integrate multi-source data. In existing monitoring technologies, the isolated processing and analysis of data leads to insufficient understanding of bioaerosol behavior. Specifically, due to differences in spatial and temporal resolution of different data sources and the lack of effective fusion methods between these data, the monitoring results are limited. In addition, existing methods are difficult to reflect the dynamic changes of aerosols in the environment in real time, especially under complex meteorological conditions, and the accuracy and reliability of monitoring are still major problems. Summary of the invention
[0003] Based on this, it is necessary to provide a bioaerosol sensor network monitoring method based on multi-source data analysis to solve at least one of the above technical problems.
[0004] To achieve the above purpose, a bioaerosol sensor network monitoring method based on multi-source data analysis includes the following steps:
[0005] Step S1: acquiring target meteorological station data and target adjacent meteorological station data; performing spatiotemporal registration processing on the target meteorological station data based on the target adjacent meteorological station data to obtain a unified meteorological time series;
[0006] Step S2: Acquire regional coverage aerosol monitoring data; perform meteorological condition standardization processing on the regional coverage aerosol monitoring data based on a unified meteorological time series to generate aerosol monitoring data;
[0007] Step S3: simulating the environment field of the unified meteorological time series to obtain a simulated meteorological environment field; reconstructing the aerosol distribution of the aerosol monitoring data based on the simulated meteorological environment field to obtain aerosol distribution data;
[0008] Step S4: construct a multi-level diffusion network for aerosol distribution data according to the unified meteorological time series, and perform disturbance diffusion simulation to generate aerosol diffusion simulation data;
[0009] Step S5: performing diffusion prediction on the aerosol diffusion simulation data and performing a propagation impact assessment to obtain aerosol propagation impact data;
[0010] Step S6: Dynamically monitor and track aerosol distribution data based on aerosol propagation impact data to generate a dynamic aerosol sensor network.
[0011] The present invention obtains the data of the target meteorological station and its adjacent meteorological stations, and performs spatiotemporal registration processing to generate a unified meteorological time series. This processing method significantly enhances the spatiotemporal consistency of meteorological data, lays a reliable foundation for the subsequent processing of aerosol monitoring data, and ensures the accuracy and reliability of the data. By performing meteorological condition standardization processing on regional coverage aerosol monitoring data based on a unified meteorological time series, the interference of meteorological factors on aerosol monitoring data can be effectively eliminated. This process improves the accuracy of monitoring data, so that it can more objectively reflect the actual distribution of aerosols. The unified meteorological time series is used for environmental field simulation to generate a simulated meteorological environmental field. This high-precision environmental field simulation can more realistically reproduce the aerosol distribution under actual meteorological conditions, and provides a reliable reference basis for aerosol distribution reconstruction, thereby improving the accuracy of aerosol distribution data. A multi-level diffusion network is constructed based on the simulated meteorological environmental field and aerosol distribution data, and a disturbance diffusion simulation is performed. This method can comprehensively consider the diffusion paths and patterns at different levels, simulate more complex and realistic aerosol diffusion conditions, and improve the accuracy and applicability of diffusion simulation. By predicting aerosol diffusion simulation data and evaluating the impact of propagation, the diffusion trend of aerosols under different conditions and their impact range can be predicted in advance. This prediction and evaluation function helps to formulate more effective monitoring and response strategies and reduce the potential hazards caused by aerosol propagation. Based on the aerosol propagation impact data, the aerosol distribution data is dynamically monitored and tracked to form a dynamic aerosol sensor network. This dynamic monitoring method can capture the changing trend of aerosols in real time, update the data of the sensor network in a timely manner, improve the real-time response capability and monitoring accuracy of the system, and ensure the efficiency and reliability of the monitoring network. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 A schematic diagram of the steps of a bioaerosol sensor network monitoring method based on multi-source data analysis;
[0013] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;
[0014] Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG.
[0015] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0016] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.
[0017] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.
[0018] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.
[0019] To achieve this, please refer to Figures 1 to 3 , a bioaerosol sensor network monitoring method based on multi-source data analysis, comprising the following steps:
[0020] Step S1: acquiring target meteorological station data and target adjacent meteorological station data; performing spatiotemporal registration processing on the target meteorological station data based on the target adjacent meteorological station data to obtain a unified meteorological time series;
[0021] Step S2: Acquire regional coverage aerosol monitoring data; perform meteorological condition standardization processing on the regional coverage aerosol monitoring data based on a unified meteorological time series to generate aerosol monitoring data;
[0022] Step S3: simulating the environment field of the unified meteorological time series to obtain a simulated meteorological environment field; reconstructing the aerosol distribution of the aerosol monitoring data based on the simulated meteorological environment field to obtain aerosol distribution data;
[0023] Step S4: construct a multi-level diffusion network for aerosol distribution data according to the unified meteorological time series, and perform disturbance diffusion simulation to generate aerosol diffusion simulation data;
[0024] Step S5: performing diffusion prediction on the aerosol diffusion simulation data and performing a propagation impact assessment to obtain aerosol propagation impact data;
[0025] Step S6: Dynamically monitor and track aerosol distribution data based on aerosol propagation impact data to generate a dynamic aerosol sensor network.
[0026] The present invention obtains the data of the target meteorological station and its adjacent meteorological stations, and performs spatiotemporal registration processing to generate a unified meteorological time series. This processing method significantly enhances the spatiotemporal consistency of meteorological data, lays a reliable foundation for the subsequent processing of aerosol monitoring data, and ensures the accuracy and reliability of the data. By performing meteorological condition standardization processing on regional coverage aerosol monitoring data based on a unified meteorological time series, the interference of meteorological factors on aerosol monitoring data can be effectively eliminated. This process improves the accuracy of monitoring data, so that it can more objectively reflect the actual distribution of aerosols. The unified meteorological time series is used for environmental field simulation to generate a simulated meteorological environmental field. This high-precision environmental field simulation can more realistically reproduce the aerosol distribution under actual meteorological conditions, and provides a reliable reference basis for aerosol distribution reconstruction, thereby improving the accuracy of aerosol distribution data. A multi-level diffusion network is constructed based on the simulated meteorological environmental field and aerosol distribution data, and a disturbance diffusion simulation is performed. This method can comprehensively consider the diffusion paths and patterns at different levels, simulate more complex and realistic aerosol diffusion conditions, and improve the accuracy and applicability of diffusion simulation. By predicting aerosol diffusion simulation data and evaluating the impact of propagation, the diffusion trend of aerosols under different conditions and their impact range can be predicted in advance. This prediction and evaluation function helps to formulate more effective monitoring and response strategies and reduce the potential hazards caused by aerosol propagation. Based on the aerosol propagation impact data, the aerosol distribution data is dynamically monitored and tracked to form a dynamic aerosol sensor network. This dynamic monitoring method can capture the changing trend of aerosols in real time, update the data of the sensor network in a timely manner, improve the real-time response capability and monitoring accuracy of the system, and ensure the efficiency and reliability of the monitoring network.
[0027] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a bioaerosol sensor network monitoring method based on multi-source data analysis of the present invention. In this example, the bioaerosol sensor network monitoring method based on multi-source data analysis includes the following steps:
[0028] Step S1: acquiring target meteorological station data and target adjacent meteorological station data; performing spatiotemporal registration processing on the target meteorological station data based on the target adjacent meteorological station data to obtain a unified meteorological time series;
[0029] In an embodiment of the present invention, high-precision meteorological data acquisition equipment is used to obtain data of the target meteorological station, including temperature, humidity, wind speed, wind direction and air pressure meteorological parameters. At the same time, the same type of meteorological data is obtained from the target adjacent meteorological station to ensure the spatiotemporal continuity of the data. Next, a spatiotemporal registration algorithm is used, such as a spatiotemporal interpolation method based on weighted average, to perform spatial and temporal corrections on the target meteorological station data. In this process, the data of the adjacent meteorological stations need to be used as a reference, and the original data of the target meteorological station needs to be corrected point by point to reduce spatial errors and time offsets, and finally generate a unified meteorological time series.
[0030] Step S2: Acquire regional coverage aerosol monitoring data; perform meteorological condition standardization processing on the regional coverage aerosol monitoring data based on a unified meteorological time series to generate aerosol monitoring data;
[0031] In an embodiment of the present invention, aerosol concentration monitoring data covering the target area is obtained through a regional aerosol monitoring network system. The meteorological conditions of the aerosol monitoring data are standardized using a unified meteorological time series. Specifically, linear regression analysis or a neural network correction model is used to correct meteorological influencing factors in the aerosol concentration data, such as temperature, humidity, and wind speed. The processed aerosol data eliminates non-essential fluctuations introduced by changes in meteorological conditions, and obtains standardized aerosol monitoring data that reflects the actual changes in aerosol concentration.
[0032] Step S3: simulating the environment field of the unified meteorological time series to obtain a simulated meteorological environment field; reconstructing the aerosol distribution of the aerosol monitoring data based on the simulated meteorological environment field to obtain aerosol distribution data;
[0033] In an embodiment of the present invention, computational fluid dynamics (CFD) tools are used to simulate the environmental field of a unified meteorological time series. First, the meteorological data is input into the environmental field simulation software to construct a three-dimensional meteorological field model. Then, numerical simulation methods, such as finite element method (FEM) or large eddy simulation (LES), are used to calculate the wind field, temperature field, and humidity field in the target area to obtain a simulated meteorological environmental field. Next, based on the simulated meteorological environmental field, the inverse modeling technology is used to reconstruct the aerosol distribution of the aerosol monitoring data. The specific operation includes using Bayesian inference or least squares algorithm to reconstruct the spatial distribution of aerosols in the target area and generate aerosol distribution data.
[0034] Step S4: construct a multi-level diffusion network for aerosol distribution data according to the unified meteorological time series, and perform disturbance diffusion simulation to generate aerosol diffusion simulation data;
[0035] In the embodiment of the present invention, the shortest path algorithm in graph theory is used to construct the aerosol propagation path between nodes. Then, the perturbation diffusion model is used to simulate the diffusion behavior of aerosols under different meteorological conditions. The perturbation diffusion simulation includes applying random perturbations to network nodes to simulate uncertain factors in the actual environment. The simulation results are statistically analyzed through Monte Carlo simulation to generate aerosol diffusion simulation data containing multiple diffusion paths.
[0036] Step S5: performing diffusion prediction on the aerosol diffusion simulation data and performing a propagation impact assessment to obtain aerosol propagation impact data;
[0037] In the embodiment of the present invention, the autoregressive integrated moving average model (ARIMA) or long short-term memory network (LSTM) time series prediction method is used to perform multi-step prediction of aerosol diffusion data. Then, the propagation impact is evaluated based on the prediction results. During the evaluation process, it is necessary to perform statistical analysis on the diffusion path and concentration changes at different time points to quantify the impact of aerosols on the target area, and finally generate aerosol propagation impact data.
[0038] Step S6: Dynamically monitor and track aerosol distribution data based on aerosol propagation impact data to generate a dynamic aerosol sensor network.
[0039] In the embodiment of the present invention, a dynamic sensor network is constructed, aerosol propagation impact data is input into the sensor network, and the layout of the monitoring nodes is adjusted in real time. The spatial distribution of aerosols is dynamically updated by online analysis of the sensor's monitoring data. This process relies on real-time data stream processing technology, combined with the Bayesian update algorithm, to ensure the accuracy and timeliness of aerosol monitoring, and ultimately generate a dynamic aerosol sensor network.
[0040] Preferably, step S1 comprises the following steps:
[0041] Step S11: Acquire target weather station data and target adjacent weather station data;
[0042] Step S12: performing outlier identification on the target weather station data based on the target adjacent weather station data to obtain abnormal data of the target weather station;
[0043] Step S13: performing Kalman filtering on the target weather station data according to the abnormal data of the target weather station to generate corrected target weather station data;
[0044] Step S14: Performing spatiotemporal registration processing on the modified target meteorological station data to obtain a unified meteorological time series.
[0045] The present invention effectively filters out the existing errors or abnormal data by identifying abnormal values of the target weather station data based on the data of adjacent weather stations. This process ensures the integrity and accuracy of the data and provides a more reliable basis for subsequent data processing. The abnormal data of the target weather station is corrected by using Kalman filtering. This technology can dynamically adjust the data, reduce the influence of noise, and generate more accurate target weather station data, thereby improving the credibility and availability of the data. The corrected target weather station data is subjected to spatiotemporal registration processing to generate a unified meteorological time series. The execution of this step is based on high-quality data, ensuring the accuracy of the registration processing, so that the generated time series can better reflect the actual meteorological conditions, and provide accurate data support for the standardization of meteorological conditions and environmental field simulation in the subsequent steps. Not only the accuracy of the single weather station data is improved, but also the reliability of multi-source data fusion is improved. The data of each weather station is subjected to accurate abnormal value identification and correction processing, which ensures the consistency and coordination of the data during fusion, and lays a solid foundation for building a high-precision meteorological time series.
[0046] In an embodiment of the present invention, by configuring an automated meteorological data acquisition system, high-precision sensors are installed at the target meteorological station and its adjacent meteorological stations to collect temperature, humidity, wind speed, wind direction, and atmospheric pressure meteorological parameters respectively. The acquisition frequency is set to once every 5 minutes, and the data is transmitted to the central data server in real time through wireless transmission technology. During the transmission process, a secure encryption protocol is used to ensure the integrity and security of the data. The server stores the meteorological data of different sites in a distributed database for subsequent processing and analysis. This process provides basic data support for subsequent outlier identification and data processing. The target meteorological station data and the target adjacent meteorological station data stored in the distributed database are first preliminarily screened by a statistical analysis method. The specific operation includes using the 3σ criterion to preliminarily identify the extreme values in the target meteorological station data, and marking the extreme values exceeding the 3σ range as outliers. Subsequently, the outliers of the target meteorological station are further confirmed by using an anomaly detection algorithm based on neighboring points, combined with the synchronous data of adjacent meteorological stations. This algorithm identifies outliers and generates an abnormal data record of the target meteorological station by calculating the difference between the target meteorological station and the adjacent meteorological station data. Based on the identified abnormal data, the Kalman filter algorithm is used to correct the target weather station data. The Kalman filter process is divided into two steps: prediction and update. First, based on the historical data and current data trends of the target weather station, the future meteorological data values are predicted. Then, the Kalman gain coefficient is used to weight the predicted value and the observed value to correct the abnormal data. The corrected data is updated through multiple iterations to gradually eliminate the influence of the abnormal values, and finally generate stable and reliable corrected target weather station data. The corrected target weather station data is processed for spatiotemporal registration. This process first uses a bilinear interpolation algorithm to register the corrected target weather station data with the adjacent weather station data in the time and space dimensions. Temporal registration uses an interpolation algorithm to align data at different time points to ensure the temporal consistency of the data; spatial registration uses geographic coordinate conversion to spatially align data at different stations to ensure the spatial consistency of the data. After processing, a unified meteorological time series is obtained.
[0047] Preferably, step S2 comprises the following steps:
[0048] Step S21: Acquire regional coverage aerosol monitoring data; perform missing value pattern recognition on the regional coverage aerosol monitoring data to obtain missing value distribution characteristics;
[0049] Step S22: performing multiple interpolation on the regional coverage aerosol monitoring data based on the missing value distribution characteristics to generate interpolated aerosol monitoring data;
[0050] Step S23: performing time series registration on the unified meteorological time series and the interpolated aerosol monitoring data to obtain a time-meteorological joint feature matrix;
[0051] Step S24: Performing spatiotemporal projection processing on the unified meteorological time series according to the time-meteorological joint feature matrix to generate aerosol monitoring data.
[0052] The present invention can accurately identify the distribution characteristics of missing values in regional coverage aerosol monitoring data through missing value pattern recognition technology. This process helps to fully understand the data missing situation, provides a scientific basis for subsequent interpolation processing, and ensures the accuracy and rationality of interpolation data. Multiple interpolation is performed based on the distribution characteristics of missing values to generate interpolated aerosol monitoring data. The multiple interpolation method effectively fills the missing values, restores the integrity of the data, improves the continuity and reliability of the data, and thus optimizes the quality of aerosol monitoring data. Time series alignment is performed on the unified meteorological time series and the interpolated aerosol monitoring data to obtain a time-meteorological joint feature matrix. This process ensures the accurate alignment of the time series with the meteorological data, enhances the accuracy of data integration, and provides a high-quality basic data set for subsequent data analysis. The unified meteorological time series is subjected to spatiotemporal projection processing through the time-meteorological joint feature matrix to generate aerosol monitoring data. This processing method can effectively integrate meteorological and aerosol data, generate high-resolution aerosol monitoring data, improve the consistency of the spatial distribution and time series of the monitoring data, and provide comprehensive data support for subsequent analysis and application.
[0053] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:
[0054] Step S21: Acquire regional coverage aerosol monitoring data; perform missing value pattern recognition on the regional coverage aerosol monitoring data to obtain missing value distribution characteristics;
[0055] In an embodiment of the present invention, aerosol monitoring data with regional coverage is obtained, and a high-precision aerosol sensor network is used to collect real-time aerosol concentration data from multiple monitoring points. These data include aerosol concentration values in different regions and time points. Next, the missing values in the regional coverage aerosol monitoring data are identified by using a missing data pattern recognition algorithm (such as the expectation maximization algorithm) through a statistical learning method. The algorithm identifies the distribution pattern of missing values based on the spatial and temporal distribution characteristics of the aerosol concentration data, generates a missing value distribution feature map, and identifies the location and missing pattern of missing values in the data.
[0056] Step S22: performing multiple interpolation on the regional coverage aerosol monitoring data based on the missing value distribution characteristics to generate interpolated aerosol monitoring data;
[0057] In an embodiment of the present invention, multiple predictions are made on missing values through a statistical model (such as multiple interpolation chain equations (MICE)) to generate interpolation data. Each interpolation is performed by combining the spatial correlation and time trend of aerosol concentration data with a random forest regression model to fill the missing values. The generated interpolation data set is used to replace the missing values in the original data to form complete aerosol monitoring data. This method improves the accuracy and stability of the interpolation data through multiple interpolations and result fusion.
[0058] Step S23: performing time series registration on the unified meteorological time series and the interpolated aerosol monitoring data to obtain a time-meteorological joint feature matrix;
[0059] In an embodiment of the present invention, the interpolated aerosol monitoring data is combined with a unified meteorological time series to perform time series registration processing. A time series alignment algorithm (such as dynamic time warping (DTW)) is used to synchronously align the aerosol monitoring data and the meteorological time series data. The algorithm adjusts the timestamps of the aerosol data to match the time series of the meteorological data by calculating the minimum distance between different time series. The resulting time-meteorological joint feature matrix aligns the aerosol concentration data with meteorological parameters (such as temperature, humidity, and wind speed) in the time dimension to generate a standardized data matrix for subsequent analysis.
[0060] Step S24: Performing spatiotemporal projection processing on the unified meteorological time series according to the time-meteorological joint feature matrix to generate aerosol monitoring data.
[0061] In an embodiment of the present invention, spatiotemporal projection processing is performed according to the time-meteorological joint feature matrix. Principal component analysis (PCA) is applied to reduce the dimension of the joint feature matrix to extract the main spatiotemporal features. Then, spatial interpolation technology (such as Kriging interpolation) is used to map the reduced-dimensional data to the aerosol monitoring points in the target area to generate a complete aerosol data map. This figure shows the distribution and change trend of aerosol concentration through spatiotemporal projection. The processed aerosol monitoring data covers the aerosol concentration information in the entire area, providing detailed data support for subsequent aerosol analysis and monitoring.
[0062] Preferably, step S3 comprises the following steps:
[0063] Step S31: constructing a meteorological simulation model for the unified meteorological time series to obtain a simulated meteorological environment field;
[0064] Step S32: performing meteorological simulation calculation on the simulated meteorological environment field to generate a simulated airflow turbulence field;
[0065] Step S33: Tracking the aerosol monitoring data based on the simulated airflow turbulence field to generate an aerosol trajectory set;
[0066] Step S34: performing kernel density estimation on the aerosol trajectory set to obtain a continuous aerosol distribution field;
[0067] Step S35: performing intrinsic mode decomposition on the continuous aerosol distribution field to generate an aerosol distribution characteristic mode;
[0068] Step S36: performing high-dimensional state estimation on the continuous aerosol distribution field according to the aerosol distribution characteristic pattern to obtain aerosol distribution data.
[0069] The present invention obtains a simulated meteorological environment field by constructing a meteorological simulation model. This model can accurately simulate the meteorological environment of the target area, provide a reliable meteorological data basis for subsequent airflow turbulence field calculation and aerosol trajectory tracking, and improve the authenticity and accuracy of the simulation. Meteorological simulation calculation is performed on the simulated meteorological environment field to generate a simulated airflow turbulence field. This process can describe the turbulent characteristics of the airflow in detail, improve the understanding of airflow movement and its impact on aerosol propagation, and enhance the accuracy and effectiveness of the simulation. Based on the simulated airflow turbulence field, the trajectory of the aerosol monitoring data is tracked to generate an aerosol trajectory set. This step can accurately track the movement path of the aerosol in the airflow, provide detailed trajectory data for subsequent distribution field analysis, and improve the accuracy of the analysis. By performing kernel density estimation on the aerosol trajectory set, a continuous aerosol distribution field is obtained. This analysis method can generate smooth aerosol distribution data, carefully depict the distribution of aerosols in space, and improve the spatial resolution and accuracy of the monitoring data. The continuous aerosol distribution field is subjected to intrinsic mode decomposition to generate an aerosol distribution characteristic pattern. This technology can extract the main characteristic patterns of aerosol distribution, reveal the underlying laws and patterns of aerosol distribution, and help to better understand the behavioral characteristics of aerosols. According to the characteristic patterns of aerosol distribution, the high-dimensional state of the continuous aerosol distribution field is estimated to obtain aerosol distribution data. This step can perform comprehensive aerosol state estimation in high-dimensional space, improve the accuracy and comprehensiveness of aerosol distribution data, and provide accurate reference data for scientific research and practical applications.
[0070] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:
[0071] Step S31: constructing a meteorological simulation model for the unified meteorological time series to obtain a simulated meteorological environment field;
[0072] In an embodiment of the present invention, unified meteorological time series data are collected and organized, including temperature, humidity, wind speed and wind direction meteorological parameters. These data usually come from observation records of different meteorological stations. A numerical weather prediction (NWP) model, such as a WRF (Weather Research and Forecasting) model, is used to input these meteorological parameters. A meteorological simulation model is established by configuring the initial conditions, boundary conditions and parameter settings of the model. After the model is run, a simulation result of the meteorological environment field is generated, which characterizes the changes in the meteorological environment within a specified time period. The simulation result provides a spatial distribution map of meteorological elements such as airflow, temperature distribution and air pressure, i.e., a simulated meteorological environment field.
[0073] Step S32: performing meteorological simulation calculation on the simulated meteorological environment field to generate a simulated airflow turbulence field;
[0074] In an embodiment of the present invention, simulated meteorological environment field data is used and input into an airflow simulation tool, such as the LES (Large Eddy Simulation) model. This model focuses on simulating the turbulent characteristics of airflow and processes the turbulence of airflow through computational fluid dynamics equations (Navier-Stokes equations). The specific process includes calculating the airflow velocity, pressure and its turbulent characteristics on a three-dimensional grid. During the simulation calculation process, a turbulence model (such as the Smagorinsky model) is used to process small-scale turbulence. The final output result is a simulated airflow turbulence field, which is manifested as the velocity and turbulence intensity distribution of the airflow within a specified time and space range.
[0075] Step S33: Tracking the aerosol monitoring data based on the simulated airflow turbulence field to generate an aerosol trajectory set;
[0076] In an embodiment of the present invention, in a simulated airflow turbulence field, aerosol monitoring data is used as a particle source, and a particle tracking algorithm, such as the Lagrangian particle tracking method, is used. Aerosol particles are emitted from a monitoring point, and the motion trajectory of the particles is tracked according to the speed and direction of the airflow turbulence field. In a specific operation, the position change of the particle within the time step is calculated, and the speed and direction of the particle are updated. This process is repeated until the particle leaves the monitoring area or reaches a specified time length. The final result is an aerosol trajectory set, which is represented by the motion path of the aerosol particles in the simulated environment.
[0077] Step S34: performing kernel density estimation on the aerosol trajectory set to obtain a continuous aerosol distribution field;
[0078] In the embodiment of the present invention, the kernel density estimation method is applied to calculate the spatial distribution of aerosols based on the aerosol trajectory set. First, an appropriate kernel function (such as a Gaussian kernel) and bandwidth parameters are selected, and then the kernel function is calculated for each trajectory point to generate a local density estimate. The density estimation results of all trajectory points are weighted and averaged to obtain a continuous distribution field of aerosols in space. The result generated by the Gaussian kernel density estimation in this step is a continuous aerosol distribution field, which reflects the concentration change of aerosols in the specified area.
[0079] Step S35: performing intrinsic mode decomposition on the continuous aerosol distribution field to generate an aerosol distribution characteristic mode;
[0080] In an embodiment of the present invention, the Empirical Mode Decomposition (EMD) method is applied to analyze the continuous aerosol distribution field data. The specific operation includes decomposing the continuous aerosol distribution field data into a number of intrinsic mode functions (IMFs) and a trend term. Each IMF represents a different frequency component of the aerosol distribution field, which can effectively extract the local features and periodic patterns in the data. By analyzing these IMFs, the characteristic pattern of aerosol distribution is obtained, that is, the distribution pattern of aerosol at different time scales and spatial scales.
[0081] Step S36: performing high-dimensional state estimation on the continuous aerosol distribution field according to the aerosol distribution characteristic pattern to obtain aerosol distribution data.
[0082] In an embodiment of the present invention, a high-dimensional state estimation method is used to perform fine modeling of the aerosol distribution field based on the aerosol distribution characteristic pattern. High-dimensional data fusion technology, such as principal component analysis (PCA) or independent component analysis (ICA), is used to reduce the dimension of the aerosol distribution characteristic pattern. The reduced-dimensional features are combined with the actual observed data of the aerosol distribution field, and the final state estimation is performed using a linear or nonlinear state estimation algorithm, such as a Kalman filter or a particle filter. The obtained aerosol distribution data is a high-dimensional state estimation result, which is expressed as the distribution of aerosols in different time and space dimensions.
[0083] Preferably, step S32 includes the following steps:
[0084] Step S321: recursively divide the simulated meteorological environment field into octrees to obtain octree hierarchical coding;
[0085] Step S322: performing hierarchical grid mapping on the octree hierarchical coding to generate a multi-scale grid of the meteorological field;
[0086] Step S323: performing gas turbulence potential mapping on the multi-scale grid of the meteorological field to obtain a turbulence potential matrix;
[0087] Step S324: performing coherent structure identification on the turbulence potential matrix to generate a coherent vortex matrix;
[0088] Step S325: mapping the energy transfer relationship of the coherent vortex matrix to obtain energy exchange network data;
[0089] Step S326: Iteratively adjust the simulated meteorological environment field point by point based on the energy exchange network data to generate a simulated airflow turbulence field.
[0090] The present invention generates octree hierarchical coding by recursively dividing the simulated meteorological environment field by octree. This division method can subdivide the meteorological environment into multiple levels, improve the flexibility and efficiency of data processing, and help to accurately manage and analyze meteorological data. Hierarchical grid mapping is performed through octree hierarchical coding to generate multi-scale grids of the meteorological field. This multi-scale grid representation method can effectively capture meteorological characteristics at different scales, so that the simulation results can more comprehensively reflect the complexity of the meteorological environment. The multi-scale grid of the meteorological field is mapped with gas turbulence potential to obtain a turbulence potential matrix. This analysis step can accurately evaluate the turbulence potential of the airflow and provide detailed basic data for subsequent vortex identification and energy transfer analysis. The coherent vortex matrix is generated by performing coherent structure identification on the turbulence potential matrix. This identification technology can reveal the coherent vortex structure in the airflow, help understand the vortex behavior and influence of the airflow, and improve the accuracy of airflow simulation. The coherent vortex matrix is mapped with energy transfer relationship to obtain energy exchange network data. This analysis can reveal the energy transfer relationship between different vortices and provide detailed information for the energy distribution and dynamic changes of the airflow. By iteratively adjusting the simulated meteorological environment field point by point based on the energy exchange network data, a more accurate simulated airflow turbulence field is generated. This process can improve the simulation accuracy of the airflow turbulence field, making the final generated turbulence field more consistent with actual meteorological conditions, thereby improving the accuracy of the overall meteorological environment simulation.
[0091] In an embodiment of the present invention, when processing the simulated meteorological environment field data, the three-dimensional space is first recursively divided using an octree algorithm. At the beginning of this process, the space cube is divided into eight sub-cubes, and then each sub-cube is further recursively divided until the set resolution is reached or a specific condition is met. Each sub-cube is assigned a unique hierarchical code in the octree structure, which identifies its position and level in the tree structure. This division method allows the meteorological environment field to be analyzed in detail at different resolutions, and the generated hierarchical code is used for subsequent grid mapping and data processing. The octree hierarchical code is input into a grid mapping tool for hierarchical grid mapping. This process includes converting the hierarchical code of the octree into multi-scale grids, which represent spatial resolutions at different levels. Specifically, a grid structure with different levels of detail is generated by matching the octree code with preset grid parameters. The multi-scale grid reflects the spatial variation of the meteorological environment field and provides grid data of different scales for subsequent gas turbulence potential calculations. After obtaining the multi-scale grid of the meteorological field, a gas turbulence model is applied for turbulence potential mapping. The turbulence potential of each grid cell is calculated using a numerical model (such as Large Eddy Simulation, LES) to generate a turbulence potential matrix. This matrix contains the turbulence intensity and gas turbulence potential distribution of each grid cell. The calculation process takes into account the factors of airflow velocity, temperature and pressure, and the turbulent kinetic energy and its dissipation rate are solved by numerical solutions. Finally, the turbulence potential matrix shows the potential energy distribution of turbulence in the meteorological field. The turbulence potential matrix is subjected to coherent structure analysis to identify the coherent vortices in the airflow. The vortex regions in the turbulence potential matrix are calculated using coherent structure detection algorithms, such as the coherent vortex identification algorithm. The algorithm identifies regions with significant vortex characteristics by analyzing the local extreme values and gradient changes of the turbulence potential. These regions represent the main vortex structures in the airflow. The resulting coherent vortex matrix shows the spatial distribution and intensity of the vortices, providing basic data for energy transfer analysis. Based on the coherent vortex matrix, the energy transfer analysis tool is applied to construct the energy exchange network. First, the energy transfer and exchange relationship of each vortex area is calculated, and the energy transfer path and intensity between vortices are analyzed. An energy transfer model, such as a coherent vortex model, is used to generate energy exchange network data, which includes the energy transfer matrix and its intensity between each vortex. This network data describes the energy flow and exchange pattern in the vortex system, and provides important information for the adjustment of the airflow turbulence field. According to the energy exchange network data, the simulated meteorological environment field is iteratively adjusted point by point. In each iteration, the energy exchange network data is applied to the corresponding grid cells of the meteorological environment field to adjust its turbulent potential and airflow characteristics. The iterative process includes modifying the velocity, pressure, and turbulence parameters of each grid cell to match the energy transfer and exchange pattern.After multiple iterations, the adjustment results will gradually approach the characteristics of the actual airflow turbulence field. The final generated simulated airflow turbulence field shows the adjusted airflow turbulence characteristics and can more accurately reflect the airflow changes in the actual environment.
[0092] Preferably, step S4 comprises the following steps:
[0093] Step S41: performing multifractal analysis on aerosol distribution data according to a unified meteorological time series to obtain a multifractal spectrum;
[0094] Step S42: performing principal component analysis on the multifractal spectrum to generate main variation patterns;
[0095] Step S43: constructing a characteristic field of aerosol distribution data based on the main change mode to obtain a multi-level characteristic field, wherein the multi-level characteristic field includes a macro-level characteristic field, a meso-level characteristic field and a micro-level characteristic field;
[0096] Step S44: performing topological data analysis on the macro-level feature field to obtain continuous homology features;
[0097] Step S45: performing manifold learning on the meso-level feature field to obtain a low-dimensional embedding representation;
[0098] Step S46: performing fractal analysis on the microscopic level characteristic field to obtain multiple fractal features;
[0099] Step S47: Perform multi-scale coupling simulation on the continuous homology features, low-dimensional embedding representation and multi-fractal features to generate aerosol diffusion simulation data.
[0100] The present invention obtains a multifractal spectrum by performing multifractal analysis on aerosol distribution data. This analysis method can reveal the complexity and self-similarity of aerosol distribution, and provides profound insights for understanding the multi-scale characteristics of aerosol distribution. The multifractal spectrum is subjected to principal component analysis to generate the main change pattern. The principal component analysis can extract the main change trend in aerosol distribution data, simplify the complexity of the data, and help identify and understand the key patterns in the data. A multi-level feature field is constructed based on the main change pattern, including macro-level, meso-level and micro-level feature fields. This hierarchical construction method can comprehensively describe the different scale characteristics of aerosol distribution and provide hierarchical feature data for subsequent analysis. Topological data analysis is performed on the macro-level feature field to obtain continuous homology features. This analysis can reveal the topological structure characteristics in aerosol distribution and provide an in-depth understanding of the spatial structure of aerosol distribution. Manifold learning is performed on the meso-level feature field to obtain a low-dimensional embedded representation. Manifold learning technology can map complex feature field data to a low-dimensional space, simplify data representation and processing, and improve the efficiency of data analysis. Fractal analysis of the microscopic characteristic field is performed to obtain multiple fractal features. This process can accurately extract the microscopic characteristics in the aerosol distribution and reveal the subtle changes and complexity of the aerosol distribution. Aerosol diffusion simulation data is generated by multi-scale coupling simulation of continuous homology features, low-dimensional embedding representation and multiple fractal features. This comprehensive simulation method can combine data features at different scales to provide a comprehensive prediction and analysis of aerosol diffusion, improving the accuracy and practicality of the simulation data.
[0101] In an embodiment of the present invention, the unified meteorological time series is aligned with the aerosol distribution data to ensure the consistency and integrity of the data. Subsequently, the multifractal analysis technology is used to calculate the local and global fractal characteristics of the aerosol distribution data. The specific method includes using the "wave analysis" technology to decompose the data into components of different scales, and obtaining the multifractal spectrum of the aerosol distribution by calculating the local fractal dimension of each scale. This process involves continuous decomposition of the data, evaluating its complexity and distribution characteristics at different scales, and thus generating a detailed multifractal spectrum. The multifractal spectrum is subjected to principal component analysis (PCA). The data matrix in the multifractal spectrum is input into the principal component analysis algorithm, which extracts the main change mode by calculating the eigenvalues and eigenvectors of the covariance matrix. The results of the principal component analysis show the main variation directions and characteristics in the aerosol distribution data. The specific operation includes standardizing the multifractal spectrum data, performing singular value decomposition (SVD), and then determining the first few principal components, which represent the most significant change patterns in the data for subsequent feature field construction. The main change patterns are used to reduce the dimension of aerosol distribution data and extract representative features. These features are used to construct a multi-level feature field of aerosol, including macro-level, meso-level and micro-level feature fields. The macro-level feature field reflects the overall trend and large-scale distribution of aerosol distribution, the meso-level feature field shows the change characteristics of the medium scale, and the micro-level feature field shows subtle local changes. This process is completed through multi-level data modeling technology to ensure the hierarchy and precision of the feature field. Topological data analysis is performed on the macro-level feature field, and persistent homology analysis technology is used. By constructing a topological data representation of the feature field, its homology groups of different dimensions are calculated, and the topological characteristics of the data at different scales are analyzed. The specific method includes using topological data analysis tools to calculate the persistence map in persistent homology and identify persistent structures and topological features in the feature field. These features reveal the structural stability and topological characteristics of aerosol distribution at the macro level, providing basic data for aerosol diffusion simulation. Manifold learning algorithms are applied to the meso-level feature field to obtain low-dimensional embedding representations. First, the meso-level feature field data is analyzed using the manifold hypothesis, assuming that the data is embedded in a low-dimensional manifold in the high-dimensional space. The high-dimensional data of the meso-level feature field is mapped to the low-dimensional space using manifold learning techniques, such as local linear embedding (LLE) or t-SNE. Manifold learning algorithms generate low-dimensional embedding representations that show the main data features and distribution patterns at the meso level by minimizing reconstruction errors or maintaining the local structure of the data. Fractal analysis is performed on the micro-level feature field. Fractal dimension calculation methods, such as box counting or Wiener-Mandelbrot analysis, are used to evaluate the complexity and details of the micro-level data. The multi-fractal features of the micro-level feature field are identified by calculating the fractal dimension of the micro-level feature field at different scales.During the fractal analysis process, the data is decomposed into parts of several scales, and the fractal characteristics of each scale are calculated to generate multiple fractal features that describe the aerosol distribution at the microscopic level. The continuous homology features, low-dimensional embedding representations, and multiple fractal features are simulated by multi-scale coupling. The multi-scale coupling simulation technology is used to combine the characteristic data at different levels to simulate the diffusion behavior of aerosols in the actual environment. The specific operations include constructing a multi-scale model, integrating the characteristics of different levels into the model, and calculating the diffusion process of aerosols through numerical simulation. The simulation results generate aerosol diffusion simulation data, showing the spatial distribution and dynamic changes of aerosols at different scales.
[0102] Preferably, performing multi-scale coupling simulation on the continuous homology feature, the low-dimensional embedding representation and the multi-fractal feature comprises the following steps:
[0103] The inter-layer information channel is constructed for the continuous homology features, low-dimensional embedding representation and multi-fractal features to obtain the inter-layer information transmission channel;
[0104] Based on the inter-layer information transmission channel, the continuous homology features, low-dimensional embedding representation and multi-fractal features are multi-scale coupled to generate hierarchical diffusion dynamics data;
[0105] Perform random perturbation simulation on hierarchical diffusion dynamics data to obtain perturbation simulation field;
[0106] Deterministic equation calculations are performed on hierarchical diffusion dynamics data to generate continuous solution fields of equations;
[0107] Perform multi-scale coupling calculations on hierarchical diffusion dynamics data to obtain a multi-scale coupling solution process;
[0108] The disturbance simulation field, the continuous solution field of the equation and the multi-scale coupling solution process are integrated to generate aerosol diffusion simulation data.
[0109] The present invention can effectively realize information transmission between different feature levels through the construction of inter-layer information channels. This process ensures the integration and fusion of continuous homology features, low-dimensional embedding representations and multi-fractal features, and improves the comprehensive utilization efficiency and information integrity of feature data. The features are multi-scale coupled through the inter-layer information transmission channel to generate hierarchical diffusion dynamics data. This method can comprehensively analyze the diffusion behavior of aerosols at different scales, provide a more comprehensive dynamic description, and enhance the understanding of the aerosol diffusion process. The hierarchical diffusion dynamics data is subjected to random perturbation simulation to generate a perturbation simulation field. This process can simulate random interference in the actual environment, provide a detailed understanding of the behavior of aerosol diffusion under the influence of uncertain factors, and enhance the practical adaptability of the simulation results. The equation continuous solution field is generated by deterministic equation calculation. This calculation method can provide a theoretical solution to aerosol diffusion, ensure the mathematical continuity and theoretical basis of the model, and provide rigorous verification for data analysis. Through multi-scale coupling calculation, the influence of different scales and levels can be comprehensively considered to generate a multi-scale coupling solution process. This process can provide a multi-level perspective on aerosol diffusion and enhance the comprehensive understanding and prediction ability of the diffusion process. Aerosol diffusion simulation data is generated by integrating the disturbance simulation field, the continuous solution field of the equation, and the multi-scale coupled solution process. This integration method can comprehensively consider data from different sources and processing methods, improve the accuracy and reliability of the simulation results, and ensure the comprehensiveness and practicality of aerosol diffusion simulation.
[0110] In the embodiment of the present invention, the continuous homology feature, low-dimensional embedding representation and multi-fractal feature are structured, and the data is standardized and normalized to ensure that different feature data are processed under the same dimension. Subsequently, the inter-layer information transmission channel is constructed using the graph theory method. The specific operation includes defining channel nodes for the feature data of each level, establishing edge connections between the nodes, and these edges represent the information transmission path between the features. The graph convolution network algorithm is used to extract the correlation between different feature levels, calculate the weight of the information transmission channel, and integrate the feature data of each level to generate the inter-layer information transmission channel. This process ensures the effective transmission of feature information between different levels. After obtaining the inter-layer information transmission channel, multi-scale coupling processing is performed. The inter-layer information transmission channel is applied to the continuous homology feature, low-dimensional embedding representation and multi-fractal feature, and these feature data are integrated into a unified multi-scale coupling model by setting weight coefficients of different scales. A multi-scale data fusion algorithm, such as wavelet transform or multi-scale decomposition technology, is used to fuse the feature data at multiple scales to generate hierarchical diffusion dynamics data. These data reflect the aerosol diffusion dynamics and their interactions at different feature levels. Perform random perturbation simulation on the hierarchical diffusion dynamics data. The specific operation includes applying a random noise generation algorithm to superimpose noise on the hierarchical diffusion dynamics data to simulate the random perturbation effect in the actual environment. Select an appropriate noise model, such as Gaussian noise or Poisson noise, and adjust the intensity and distribution of the noise according to the simulation requirements. After the perturbation is applied to the hierarchical diffusion dynamics data, a perturbation simulation field is obtained, which shows the aerosol diffusion pattern and changes under random perturbation conditions. The hierarchical diffusion dynamics data is converted into an appropriate partial differential equation (PDE) model to describe the deterministic behavior of aerosol diffusion. The equations are discretized using numerical methods, such as the finite difference method or the finite element method. The calculated continuous solution field of the equation shows the diffusion behavior of aerosols under deterministic conditions. Ensure that the initial conditions, boundary conditions and parameter settings of the model are accurate, and the calculation results represent the stable diffusion field of aerosols under different conditions. Using the previously obtained perturbation simulation field and the continuous solution field of the equation, the aerosol diffusion process is comprehensively solved through a multi-scale coupling algorithm combined with characteristic data at different scales. Multiscale coupling calculations include combining disturbance effects with deterministic models, and applying hybrid model techniques, such as weighted averaging or model fusion algorithms, to process data at different scales. The generated multiscale coupling solution process provides the dynamic behavior of aerosols at different scales, reflecting the contribution of each scale to the overall diffusion process. Integrate the disturbance simulation field, the continuous solution field of the equation, and the multiscale coupling solution process. Through data fusion technology, the three data are matched in spatial and temporal dimensions to ensure the consistency and integrity of the data. Use data fusion algorithms, such as weighted fusion or principal component fusion, to integrate data from different sources to generate comprehensive aerosol diffusion simulation data.These data demonstrate the comprehensive diffusion behavior of aerosols under different conditions and scales, and provide detailed simulation results for subsequent analysis and monitoring.
[0111] Preferably, step S5 comprises the following steps:
[0112] Step S51: performing variational mode decomposition processing on the aerosol diffusion simulation data to obtain instantaneous frequency distribution characteristics;
[0113] Step S52: performing modal function grouping on the aerosol diffusion simulation data based on the instantaneous frequency distribution characteristics to generate time scale component data;
[0114] Step S53: construct a multivariate time series model for the time scale component data to obtain an integrated prediction model;
[0115] Step S54: Predicting the diffusion results of the aerosol diffusion simulation data based on the integrated prediction model to obtain aerosol diffusion result data;
[0116] Step S55: Perform a propagation impact assessment on the aerosol diffusion result data according to a preset propagation impact assessment level to obtain aerosol propagation impact data.
[0117] The present invention can extract the instantaneous frequency distribution characteristics by performing variational modal decomposition on aerosol diffusion simulation data. This process reveals the frequency characteristics in the aerosol diffusion process, helps identify important frequency components and change patterns in the data, and provides a deeper understanding of the diffusion process. Modal functions are grouped based on the instantaneous frequency distribution characteristics to generate time scale component data. This method can classify aerosol diffusion simulation data by time scale, so that data features on different time scales can be separated and analyzed, thereby improving the precision and accuracy of data processing. An integrated prediction model is obtained by constructing a multivariate time series model for the time scale component data. The integrated model can comprehensively consider the influence of multiple variables and time series, improve the prediction ability of the aerosol diffusion process, and enhance the prediction accuracy and reliability of the model. Based on the integrated prediction model, the diffusion results of the aerosol diffusion simulation data are predicted to obtain aerosol diffusion result data. This process can provide prediction data for aerosol diffusion, support analysis and decision-making on future diffusion conditions, and improve the forward-looking prediction ability of aerosol behavior. The aerosol diffusion result data is evaluated according to the preset transmission impact assessment level to obtain the aerosol transmission impact data. This assessment can quantify the actual impact of aerosol diffusion, help assess its potential threats to the environment and health, and provide a scientific basis for responding to and managing the risks of aerosol transmission.
[0118] In an embodiment of the present invention, a variational mode decomposition (VMD) algorithm is applied to aerosol diffusion simulation data for processing. First, the aerosol diffusion simulation data is preprocessed, including noise removal and standardization, to ensure the quality of the data. Then, the aerosol diffusion simulation data is decomposed into several modal components using a variational mode decomposition algorithm. Each modal component corresponds to a different frequency component. By analyzing these modal components, the instantaneous frequency distribution characteristics of each mode are calculated. The process involves using the iterative update step and optimization process in the VMD algorithm to obtain the instantaneous frequency of each frequency band and its corresponding amplitude spectrum, providing an accurate data basis for the subsequent grouping of modal functions. According to the instantaneous frequency distribution characteristics, the modal functions of the aerosol diffusion simulation data are grouped. First, according to the distribution characteristics of the instantaneous frequency, the frequency range and its corresponding time scale are defined, and the modal components are grouped according to these frequency ranges. Using a clustering algorithm such as K-means or hierarchical clustering, modal components with similar frequency characteristics are grouped into the same group. Each group represents a specific time scale and generates time scale component data. These data reflect the characteristics of aerosol diffusion at different time scales, which is helpful for subsequent multivariate time series modeling. The time scale component data are divided into training sets and test sets. Select a suitable multivariate time series model, such as a vector autoregression model (VAR) or a long short-term memory network (LSTM), for training. During the model training process, the dynamic characteristics and laws of aerosol diffusion are learned through historical time series data. After the training is completed, an integrated prediction model is obtained, which can comprehensively consider the impacts at each time scale and predict the aerosol diffusion behavior. The integrated prediction model is used to predict the aerosol diffusion simulation data. The aerosol diffusion simulation data to be predicted is input into the model, and the prediction results are generated through the forward propagation calculation of the model. This process includes the real-time processing and prediction of new data by the model, and the output of aerosol diffusion result data. These data show the diffusion trend and distribution of aerosols in the future time period, providing a basis for further analysis and decision-making. Establish a propagation impact assessment model, including assessment criteria and impact level definition. Using the aerosol diffusion result data, the assessment is carried out according to the set standards, and the impact level is assigned to different regions and time periods. The assessment model can conduct a comprehensive analysis based on diffusion concentration and exposure time factors. The generated aerosol transmission impact data reflects the potential impact of aerosol diffusion on the environment and health, and provides a scientific basis for the formulation of relevant protective measures and policies.
[0119] Preferably, step S53 includes the following steps:
[0120] Step S531: Perform component dimension time embedding on the time scale component data to obtain a time scale embedding matrix;
[0121] Step S532: performing cross-fusion processing on the time scale embedding matrix to obtain cross-time scale fusion features;
[0122] Step S533: performing hierarchical prediction modeling on the time scale component data based on the cross-time scale fusion features to generate a hierarchical prediction model;
[0123] Step S534: Perform a preliminary prediction test on the hierarchical prediction model to obtain a hierarchical prediction matrix;
[0124] Step S535: Perform multivariate prediction calibration processing on the hierarchical prediction model based on the hierarchical prediction matrix to obtain an integrated prediction model.
[0125] The present invention generates a time scale embedding matrix by embedding the component dimension of the time scale component data. This process can convert the time scale characteristics of the data into a high-dimensional matrix representation, thereby effectively capturing the complex dynamic characteristics in the time series and enhancing the accuracy and meticulousness of data processing. By cross-fusion processing of the time scale embedding matrix, a cross-time scale fusion feature is obtained. This processing method can integrate data features at different time scales, improve the comprehensive analysis ability of time series data, and enhance the comprehensiveness of feature extraction. Hierarchical prediction modeling is performed based on the cross-time scale fusion feature to generate a hierarchical prediction model. This hierarchical model can process time features at different levels, improve the accuracy and reliability of prediction, and adapt to the multi-level characteristics of complex time series data. A preliminary prediction test is performed on the hierarchical prediction model to obtain a hierarchical prediction matrix. The matrix can provide preliminary prediction results for the aerosol diffusion process and provide basic data for subsequent calibration and optimization. Based on the hierarchical prediction matrix, a multivariate prediction calibration process is performed on the hierarchical prediction model to generate an integrated prediction model. This process optimizes the performance of the prediction model by calibrating the multivariate information, improves the prediction ability and accuracy of the model for aerosol diffusion, and ensures the reliability and practicality of the prediction results.
[0126] In an embodiment of the present invention, a time series embedding process is performed on each time scale component. Using the time embedding technology, the component data on the time dimension is mapped to a high-dimensional space to form a time scale embedding matrix. In specific implementation, the delay embedding method can be used to combine the past observations of each time series into a high-dimensional vector. These high-dimensional vectors are arranged in chronological order to construct a time scale embedding matrix. This matrix contains a high-dimensional representation of the time series and is used to capture data features at different time scales. Define feature fusion rules across time scales and select embedding matrices of different time scales for cross-fusion. By calculating the similarity between the embedding matrices, the features of different time scales are fused using the weighted average method or collaborative filtering technology. This can be achieved by constructing a fusion function to combine the features at different time scales into a new feature vector. For example, principal component analysis (PCA) can be used to reduce the dimension of the cross-fused data and extract the most representative fusion features. Finally, cross-time scale fusion features are obtained for subsequent hierarchical prediction modeling. The cross-time scale fusion features are used as input to divide the data set into a training set and a test set. Select an appropriate prediction algorithm, such as a multi-layer perceptron (MLP) or a convolutional neural network (CNN), to train the model. During the model training process, use cross-validation techniques to optimize the model parameters and evaluate the model's prediction performance. The resulting hierarchical prediction model can handle data features at different time scales and perform time series predictions. The model is able to capture the impact of time scale fusion features on predictions and provide the ability to predict aerosol diffusion behavior. Perform preliminary prediction tests on the hierarchical prediction model to obtain a hierarchical prediction matrix. Input the test set data into the trained hierarchical prediction model for prediction calculations. This process involves using the model to forward propagate new data to generate prediction results. The predicted values for each time step form a prediction matrix, which includes the model's prediction results for different time periods. Evaluate the prediction results to check the model's performance at different time scales and adjust the model parameters to optimize the prediction performance. Finally, the resulting hierarchical prediction matrix provides a basis for further calibration and optimization. Perform error analysis on the prediction results in the hierarchical prediction matrix to identify systematic errors and biases in the predictions. Apply error correction methods, such as weighted regression or error correction models, to calibrate the model. The accuracy of the model is improved by adjusting model parameters and integrating prediction results at different time scales. The resulting integrated prediction model integrates prediction information at multiple time scales and provides more accurate aerosol diffusion prediction data.
[0127] Preferably, step S6 comprises the following steps:
[0128] Step S61: performing dynamic cluster analysis on aerosol distribution data to obtain distribution cluster data;
[0129] Step S62: dividing the distribution cluster data into monitoring areas to generate divided monitoring areas;
[0130] Step S63: marking the divided monitoring areas as key monitoring areas based on the aerosol propagation impact data to obtain important node monitoring areas;
[0131] Step S64: Based on the aerosol propagation impact data, the sensor network of the important node monitoring area is updated to generate a dynamic aerosol sensor network.
[0132] The present invention obtains distribution cluster data by performing dynamic cluster analysis on aerosol distribution data. This analysis method can identify natural aggregation areas in aerosol distribution, provides a clear understanding of aerosol distribution patterns, and is helpful for subsequent regional division and optimization of monitoring strategies. The distribution cluster data is divided into monitoring areas to generate divided monitoring areas. This division method can reasonably divide the monitoring areas according to the actual situation of aerosol distribution, improve the systematicness and pertinence of monitoring work, and help to effectively allocate resources and formulate corresponding monitoring strategies. Based on the aerosol propagation impact data, the divided monitoring areas are marked as key monitoring areas to obtain important node monitoring areas. This process can highlight key areas and optimize the layout of the monitoring network, so that areas with greater impact on aerosol propagation can be monitored first, thereby improving the efficiency and accuracy of monitoring. Based on the aerosol propagation impact data, the sensor network of important node monitoring areas is updated to generate a dynamic aerosol sensor network. This dynamic update process can timely adjust the sensor layout according to actual monitoring needs and aerosol propagation conditions, ensure the adaptability and real-time performance of the sensor network, and improve the flexibility and responsiveness of the overall monitoring system.
[0133] In an embodiment of the present invention, aerosol distribution data is processed by using methods such as K-means clustering (K-means) or density-based spatial clustering (DBSCAN). In specific implementation, a suitable distance metric, such as Euclidean distance or Manhattan distance, is selected to group the aerosol concentration distribution. Aerosol concentration, distribution area and time series data are used as input, and the data is updated and analyzed in real time using a dynamic clustering algorithm to form a clustering result of aerosol distribution. These clustering results are distribution clustering data, which represent different aerosol concentration areas and their changing trends. The boundaries of different areas are determined according to the distribution of data points in the clustering results. A spatial partitioning algorithm, such as a Voronoi diagram or a grid partitioning method, is used to map the clustering results of aerosol distribution data to the actual monitoring area. During the partitioning process, it is necessary to consider the changes in aerosol concentration in each area and its impact on the environment to ensure that each monitoring area can effectively cover the range of changes in aerosol concentration. Finally, the generated divided monitoring area will provide a basis for the layout of the monitoring network. The divided monitoring area is analyzed using aerosol propagation impact data to determine the key monitoring area. Taking the aerosol propagation impact data as input, the impact assessment algorithm, such as weighted average or spatial regression analysis, is applied to calculate the impact degree of each divided area. Areas with high impact are selected as key monitoring areas. During this process, data from the sensor network is used for comparison and verification to ensure that the selection of key areas is based on real propagation impact data. The marked important node monitoring areas will be the key monitoring objects. Based on the aerosol propagation impact data, the existing sensor network is evaluated to determine the number and location of sensors that need to be added or adjusted. According to the update requirements, new aerosol sensors are installed, or the configuration of existing sensors is adjusted to ensure that the key monitoring areas can be efficiently covered. The sensor configuration tool is used to optimize the layout to ensure that the sensor distribution can monitor the aerosol concentration and its changes in real time. Ultimately, the generated dynamic aerosol sensor network will have higher monitoring accuracy and response capabilities, and effectively track the propagation dynamics of aerosols.
[0134] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.
[0135] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.
Claims
1. A bioaerosol sensor network monitoring method based on multi-source data analysis, characterized in that: The following steps are involved: Step S1: Acquire the target weather station data and the target adjacent weather station data; The target meteorological station data is subjected to spatiotemporal registration processing based on the target adjacent meteorological station data to obtain a unified meteorological time series, wherein step S1 includes the following steps: Step S11: Acquire target weather station data and target adjacent weather station data; Step S12: performing outlier identification on the target weather station data based on the target adjacent weather station data to obtain abnormal data of the target weather station; Step S13: performing Kalman filtering on the target weather station data according to the abnormal data of the target weather station to generate corrected target weather station data; Step S14: performing spatiotemporal registration processing on the modified target meteorological station data to obtain a unified meteorological time series; Step S2: Acquire regional coverage aerosol monitoring data; perform meteorological condition standardization processing on the regional coverage aerosol monitoring data based on a unified meteorological time series to generate aerosol monitoring data, wherein step S2 includes the following steps: Step S21: Acquire regional coverage aerosol monitoring data; perform missing value pattern recognition on the regional coverage aerosol monitoring data to obtain missing value distribution characteristics; Step S22: performing multiple interpolation on the regional coverage aerosol monitoring data based on the missing value distribution characteristics to generate interpolated aerosol monitoring data; Step S23: performing time series registration on the unified meteorological time series and the interpolated aerosol monitoring data to obtain a time-meteorological joint feature matrix; Step S24: performing spatiotemporal projection processing on the unified meteorological time series according to the time-meteorological joint feature matrix to generate aerosol monitoring data; Step S3: simulating the environment field of the unified meteorological time series to obtain a simulated meteorological environment field; reconstructing the aerosol distribution of the aerosol monitoring data based on the simulated meteorological environment field to obtain aerosol distribution data, wherein step S3 includes the following steps: Step S31: constructing a meteorological simulation model for the unified meteorological time series to obtain a simulated meteorological environment field; Step S32: Perform meteorological simulation calculation on the simulated meteorological environment field to generate a simulated airflow turbulence field, wherein step S32 includes the following steps: Step S321: recursively divide the simulated meteorological environment field into octrees to obtain octree hierarchical coding; Step S322: performing hierarchical grid mapping on the octree hierarchical coding to generate a multi-scale grid of the meteorological field; Step S323: performing gas turbulence potential mapping on the multi-scale grid of the meteorological field to obtain a turbulence potential matrix; Step S324: performing coherent structure identification on the turbulence potential matrix to generate a coherent vortex matrix; Step S325: mapping the energy transfer relationship of the coherent vortex matrix to obtain energy exchange network data; Step S326: iteratively adjusting the simulated meteorological environment field point by point based on the energy exchange network data to generate a simulated airflow turbulence field; Step S33: Tracking the aerosol monitoring data based on the simulated airflow turbulence field to generate an aerosol trajectory set; Step S34: performing kernel density estimation on the aerosol trajectory set to obtain a continuous aerosol distribution field; Step S35: performing intrinsic mode decomposition on the continuous aerosol distribution field to generate an aerosol distribution characteristic mode; Step S36: performing high-dimensional state estimation on the continuous aerosol distribution field according to the aerosol distribution characteristic pattern to obtain aerosol distribution data; Step S4: construct a multi-level diffusion network for aerosol distribution data according to the unified meteorological time series, and perform disturbance diffusion simulation to generate aerosol diffusion simulation data, wherein step S4 includes the following steps: Step S41: performing multifractal analysis on aerosol distribution data according to a unified meteorological time series to obtain a multifractal spectrum; Step S42: performing principal component analysis on the multifractal spectrum to generate main variation patterns; Step S43: constructing a characteristic field of aerosol distribution data based on the main change mode to obtain a multi-level characteristic field, wherein the multi-level characteristic field includes a macro-level characteristic field, a meso-level characteristic field and a micro-level characteristic field; Step S44: performing topological data analysis on the macro-level feature field to obtain continuous homology features; Step S45: performing manifold learning on the meso-level feature field to obtain a low-dimensional embedding representation; Step S46: performing fractal analysis on the microscopic level characteristic field to obtain multiple fractal features; Step S47: performing multi-scale coupling simulation on the continuous homology feature, the low-dimensional embedding representation and the multi-fractal feature to generate aerosol diffusion simulation data, wherein step S47 includes the following steps: The inter-layer information channel is constructed for the continuous homology features, low-dimensional embedding representation and multi-fractal features to obtain the inter-layer information transmission channel; Based on the inter-layer information transmission channel, the continuous homology features, low-dimensional embedding representation and multi-fractal features are multi-scale coupled to generate hierarchical diffusion dynamics data; Perform random perturbation simulation on hierarchical diffusion dynamics data to obtain perturbation simulation field; Deterministic equation calculations are performed on hierarchical diffusion dynamics data to generate continuous solution fields of equations; Perform multi-scale coupling calculations on hierarchical diffusion dynamics data to obtain a multi-scale coupling solution process; Integrate the disturbance simulation field, equation continuous solution field and multi-scale coupling solution process to generate aerosol diffusion simulation data; Step S5: performing diffusion prediction on the aerosol diffusion simulation data and performing a propagation impact assessment to obtain aerosol propagation impact data, wherein step S5 includes the following steps: Step S51: performing variational mode decomposition processing on the aerosol diffusion simulation data to obtain instantaneous frequency distribution characteristics; Step S52: performing modal function grouping on the aerosol diffusion simulation data based on the instantaneous frequency distribution characteristics to generate time scale component data; Step S53: construct a multivariate time series model for the time scale component data to obtain an integrated prediction model, wherein step S53 includes the following steps: Step S531: Perform component dimension time embedding on the time scale component data to obtain a time scale embedding matrix; Step S532: performing cross-fusion processing on the time scale embedding matrix to obtain cross-time scale fusion features; Step S533: performing hierarchical prediction modeling on the time scale component data based on the cross-time scale fusion features to generate a hierarchical prediction model; Step S534: Perform a preliminary prediction test on the hierarchical prediction model to obtain a hierarchical prediction matrix; Step S535: performing multivariate prediction calibration processing on the hierarchical prediction model based on the hierarchical prediction matrix to obtain an integrated prediction model; Step S54: Predicting the diffusion results of the aerosol diffusion simulation data based on the integrated prediction model to obtain aerosol diffusion result data; Step S55: performing a propagation impact assessment on the aerosol diffusion result data according to a preset propagation impact assessment level to obtain aerosol propagation impact data; Step S6: Dynamically monitor and track the aerosol distribution data based on the aerosol propagation impact data to generate a dynamic aerosol sensor network, wherein step S6 includes the following steps: Step S61: performing dynamic cluster analysis on aerosol distribution data to obtain distribution cluster data; Step S62: dividing the distribution cluster data into monitoring areas to generate divided monitoring areas; Step S63: marking the divided monitoring areas as key monitoring areas based on the aerosol propagation impact data to obtain important node monitoring areas; Step S64: Based on the aerosol propagation impact data, the sensor network of the important node monitoring area is updated to generate a dynamic aerosol sensor network.
