A method for evaluating the hazard of microbial aerosol diffusion based on multidimensional data
Through multi-dimensional data processing, including outlier interval positioning, classification unit clustering and Fourier descriptor transformation, a multi-dimensional hazard heat map is generated, solving the single-dimensional and simulation accuracy of aerosol diffusion hazard assessment in the prior art. The problem has achieved more comprehensive and reliable evaluation results.
Patent Information
- Application Number
- CN202411292370.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-14
AI Technical Summary
The existing aerosol diffusion hazard assessment method is limited to single-dimensional data analysis, and it is impossible to fully understand and evaluate the diffusion hazards of microbial aerosols. The multi-scale diffusion simulation process is affected by model parameter selection and accuracy control, resulting in the accuracy of the evaluation results and Credibility is challenged.
Using a multi-dimensional data-based method, by obtaining meteorological environment data and aerosol sample data, outlier interval location, factor interpolation, classification unit clustering processing, Fourier descriptor transformation, diffusion direction and velocity estimation, multi-scale simulation, Space-time collaborative analysis and adaptive core density estimation generate multi-dimensional hazard heat maps and conduct comprehensive hazard assessment.
It improves the accuracy and systematicity of aerosol diffusion hazard assessment, provides scientific and reliable assessment results, helps identify potential hazard areas and intensity distribution, improves early warning and protection capabilities, and forms an effective response strategy for aerosol diffusion.
Smart Images

Figure CN118798665B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of environmental hazard assessment, and in particular to a microbial aerosol diffusion hazard assessment method based on multidimensional data. Background Art
[0002] In the current environment, the diffusion of microbial aerosols is of great significance to public health and environmental protection. However, the existing aerosol diffusion hazard assessment methods are often limited to single-dimensional data analysis and lack a systematic evaluation of comprehensive multidimensional data, which challenges the accuracy and comprehensiveness of the assessment results. In addition, the traditional meteorological environmental data and aerosol sample data processing methods have certain limitations, and cannot fully explore the correlation and spatiotemporal characteristics between data, which limits the comprehensive understanding and assessment of the hazards of microbial aerosol diffusion. Secondly, the multi-scale diffusion simulation process in the existing methods will be affected by the selection of model parameters and precision control, resulting in a certain degree of uncertainty in the reliability and accuracy of the simulation results. This will affect the scientificity and credibility of the final diffusion hazard assessment results. In addition, there are problems of model simplification or improper parameter selection in the estimation of aerosol diffusion direction and speed and the diffusion simulation process, which affects the accuracy of the assessment results. In addition, the current technology lacks the ability to comprehensively analyze and evaluate multidimensional data in the process of aerosol diffusion hazard assessment, and cannot achieve a comprehensive assessment of aerosol diffusion hazards, resulting in one-sidedness in the assessment results. Summary of the invention
[0003] Based on this, it is necessary to provide a microbial aerosol diffusion hazard assessment method based on multidimensional data to solve at least one of the above technical problems.
[0004] To achieve the above purpose, a microbial aerosol diffusion hazard assessment method based on multidimensional data includes the following steps:
[0005] Step S1: Acquire meteorological environment data; locate outlier intervals of the meteorological environment data, and interpolate outlier factors to generate filled meteorological environment data;
[0006] Step S2: Acquire aerosol sample data; perform classification unit clustering processing on the aerosol sample data, and perform Fourier descriptor transformation to obtain an aerosol sample feature vector;
[0007] Step S3: positioning the diffusion direction of the aerosol sample characteristic vector based on the filled meteorological environment data to obtain aerosol diffusion direction data; estimating the time-delay diffusion velocity of the aerosol sample characteristic vector based on the filled meteorological environment data to generate aerosol diffusion velocity data;
[0008] Step S4: performing multi-scale diffusion simulation on the aerosol sample data according to the aerosol diffusion direction data and the aerosol diffusion velocity data to obtain simulated diffusion aerosol data;
[0009] Step S5: Perform spatiotemporal collaborative analysis on simulated diffuse aerosol data and meteorological environment data, and perform adaptive kernel density estimation to generate a multi-dimensional hazard heat map;
[0010] Step S6: Perform diffusion hazard assessment on aerosol sample data based on the multi-dimensional hazard heat map and simulated diffusion aerosol data, and generate a hazard assessment report to perform microbial aerosol diffusion hazard assessment based on multi-dimensional data.
[0011] The present invention can identify outliers or data points that do not conform to the norm by locating the outlier interval of meteorological environmental data; through the factor interpolation technology, the data missing caused by these outliers can be effectively filled to generate a more complete meteorological data set. Ensure the integrity and accuracy of meteorological environmental data, reduce the analysis errors caused by missing data or outliers, and provide a reliable data basis for subsequent diffusion evaluation. Through the classification unit clustering processing technology, the aerosol sample data can be divided into groups with similar characteristics; the Fourier descriptor transform mathematically describes these characteristics and generates a feature vector representing the aerosol sample. Through feature extraction and cluster analysis, the data dimension can be effectively simplified, the key features of the aerosol sample can be extracted, and the subsequent diffusion behavior analysis and prediction can be facilitated. Using the filled meteorological environmental data, combined with the aerosol sample feature vector to locate the diffusion direction and estimate the time-delay diffusion speed, the diffusion path and speed of the aerosol under different meteorological conditions can be inferred. Provide accurate initial conditions for diffusion simulation, so that the simulation results are closer to the actual situation, which helps to accurately predict the diffusion range and impact area of the aerosol. Based on the diffusion direction data and diffusion velocity data, multi-scale diffusion simulation is carried out, which can reconstruct the diffusion process of aerosols from different time and space scales. It provides multi-dimensional diffusion simulation results, which can more comprehensively evaluate the diffusion behavior of aerosols under different conditions and facilitate the identification of potential hazard areas. By conducting spatiotemporal collaborative analysis of simulated diffusion aerosol data and meteorological environment data, and combining adaptive kernel density estimation technology, a multi-dimensional hazard heat map is generated. It can intuitively display the dangerous areas and intensity distribution of aerosol diffusion, help relevant departments quickly identify and respond to potential hazard areas, and improve early warning and protection capabilities. Based on the multi-dimensional hazard heat map and simulated diffusion data, a comprehensive diffusion hazard assessment is carried out, and a detailed assessment report is generated. It provides scientific and systematic diffusion hazard assessment results, forms response strategies and measures for aerosol diffusion, and provides reliable support for decision makers. A comprehensive and accurate assessment of the diffusion behavior and potential hazards of aerosols is carried out from multiple dimensions and scales. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 A schematic diagram of the steps of a microbial aerosol diffusion hazard assessment method based on multidimensional data;
[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 microbial aerosol diffusion hazard assessment method based on multidimensional data, comprising the following steps:
[0020] Step S1: Acquire meteorological environment data; locate outlier intervals of the meteorological environment data, and interpolate outlier factors to generate filled meteorological environment data;
[0021] Step S2: Acquire aerosol sample data; perform classification unit clustering processing on the aerosol sample data, and perform Fourier descriptor transformation to obtain an aerosol sample feature vector;
[0022] Step S3: positioning the diffusion direction of the aerosol sample characteristic vector based on the filled meteorological environment data to obtain aerosol diffusion direction data; estimating the time-delay diffusion velocity of the aerosol sample characteristic vector based on the filled meteorological environment data to generate aerosol diffusion velocity data;
[0023] Step S4: performing multi-scale diffusion simulation on the aerosol sample data according to the aerosol diffusion direction data and the aerosol diffusion velocity data to obtain simulated diffusion aerosol data;
[0024] Step S5: Perform spatiotemporal collaborative analysis on simulated diffuse aerosol data and meteorological environment data, and perform adaptive kernel density estimation to generate a multi-dimensional hazard heat map;
[0025] Step S6: Perform diffusion hazard assessment on aerosol sample data based on the multi-dimensional hazard heat map and simulated diffusion aerosol data, and generate a hazard assessment report to perform microbial aerosol diffusion hazard assessment based on multi-dimensional data.
[0026] The present invention can identify outliers or data points that do not conform to the norm by locating the outlier interval of meteorological environmental data; through the factor interpolation technology, the data missing caused by these outliers can be effectively filled to generate a more complete meteorological data set. Ensure the integrity and accuracy of meteorological environmental data, reduce the analysis errors caused by missing data or outliers, and provide a reliable data basis for subsequent diffusion evaluation. Through the classification unit clustering processing technology, the aerosol sample data can be divided into groups with similar characteristics; the Fourier descriptor transform mathematically describes these characteristics and generates a feature vector representing the aerosol sample. Through feature extraction and cluster analysis, the data dimension can be effectively simplified, the key features of the aerosol sample can be extracted, and the subsequent diffusion behavior analysis and prediction can be facilitated. Using the filled meteorological environmental data, combined with the aerosol sample feature vector to locate the diffusion direction and estimate the time-delay diffusion speed, the diffusion path and speed of the aerosol under different meteorological conditions can be inferred. Provide accurate initial conditions for diffusion simulation, so that the simulation results are closer to the actual situation, which helps to accurately predict the diffusion range and impact area of the aerosol. Based on the diffusion direction data and diffusion velocity data, multi-scale diffusion simulation is carried out, which can reconstruct the diffusion process of aerosols from different time and space scales. It provides multi-dimensional diffusion simulation results, which can more comprehensively evaluate the diffusion behavior of aerosols under different conditions and facilitate the identification of potential hazard areas. By conducting spatiotemporal collaborative analysis of simulated diffusion aerosol data and meteorological environment data, and combining adaptive kernel density estimation technology, a multi-dimensional hazard heat map is generated. It can intuitively display the dangerous areas and intensity distribution of aerosol diffusion, help relevant departments quickly identify and respond to potential hazard areas, and improve early warning and protection capabilities. Based on the multi-dimensional hazard heat map and simulated diffusion data, a comprehensive diffusion hazard assessment is carried out and a detailed assessment report is generated. It provides scientific and systematic diffusion hazard assessment results, forms response strategies and measures for aerosol diffusion, and provides reliable support for decision makers. A comprehensive and accurate assessment of the diffusion behavior and potential hazards of aerosols is carried out from multiple dimensions and scales, and the resulting assessment report has high reference value and guiding significance.
[0027] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for evaluating the hazard of microbial aerosol diffusion based on multidimensional data of the present invention. In this example, the method for evaluating the hazard of microbial aerosol diffusion based on multidimensional data includes the following steps:
[0028] Step S1: Acquire meteorological environment data; locate outlier intervals of the meteorological environment data, and interpolate outlier factors to generate filled meteorological environment data;
[0029] In an embodiment of the present invention, the process of obtaining meteorological environment data includes using a high-precision meteorological sensor network to collect temperature, humidity, wind speed, and air pressure environmental parameters in real time. In order to ensure the integrity and accuracy of the data, the collected data is first located in the outlier interval, and the outliers or outliers in the data set are identified through statistical methods such as box plots or standard deviation analysis. Next, the outlier data is filled using an interpolation algorithm, such as Kriging interpolation or linear regression interpolation, to generate a complete and continuous set of filled meteorological environment data. These filled data will be used in subsequent analysis steps to ensure the consistency and integrity of the meteorological environment data in space and time.
[0030] Step S2: Acquire aerosol sample data; perform classification unit clustering processing on the aerosol sample data, and perform Fourier descriptor transformation to obtain an aerosol sample feature vector;
[0031] In an embodiment of the present invention, first, aerosol sample data is collected, samples are obtained through efficient aerosol sampling equipment, and preliminary processing is performed on them to extract sample features. The aerosol sample data is clustered by classification units, and the sample data is divided into several categories using the K-means clustering algorithm or the hierarchical clustering method to ensure that each category contains aerosol samples with similar features. Subsequently, a Fourier descriptor transform is performed on each cluster category to convert the spatial distribution characteristics of the sample into Fourier coefficients to form a feature vector of the aerosol sample. These feature vectors are used to describe the shape and distribution characteristics of the sample, providing basic data for subsequent diffusion analysis.
[0032] Step S3: positioning the diffusion direction of the aerosol sample characteristic vector based on the filled meteorological environment data to obtain aerosol diffusion direction data; estimating the time-delay diffusion velocity of the aerosol sample characteristic vector based on the filled meteorological environment data to generate aerosol diffusion velocity data;
[0033] In an embodiment of the present invention, based on the filled meteorological environment data and the aerosol sample characteristic vector, the wind speed and wind direction information in the meteorological data is first used to locate the diffusion direction of the aerosol sample characteristic vector. A fluid dynamics model or a wind field analysis model is used to calculate the main diffusion direction of the aerosol under meteorological conditions. The diffusion direction data of the aerosol is determined by combining it with the meteorological environment data. Next, based on the filled meteorological environment data, the time-lagged diffusion velocity of the aerosol sample characteristic vector is estimated. Time series analysis methods such as the autoregressive integrated moving average model (ARIMA) are applied to estimate the diffusion velocity of the aerosol in combination with the diffusion characteristics of the sample at different time points. The generated diffusion velocity data will be used to simulate the propagation behavior of the aerosol in the environment.
[0034] Step S4: performing multi-scale diffusion simulation on the aerosol sample data according to the aerosol diffusion direction data and the aerosol diffusion velocity data to obtain simulated diffusion aerosol data;
[0035] In an embodiment of the present invention, aerosol diffusion direction data and diffusion velocity data are used to perform multi-scale diffusion simulation. First, by establishing a multi-scale model (e.g., a composite model of macro-scale and micro-scale), the diffusion process of aerosol is simulated at different spatial and temporal scales. Numerical simulation techniques, such as finite element method (FEM) or finite difference method (FDM), are applied to input aerosol diffusion data into the model to simulate the propagation process of aerosol in the environment. Simulated diffusion aerosol data are generated through the simulation results, which can reflect the actual diffusion pattern and range of aerosol under different environmental conditions.
[0036] Step S5: Perform spatiotemporal collaborative analysis on simulated diffuse aerosol data and meteorological environment data, and perform adaptive kernel density estimation to generate a multi-dimensional hazard heat map;
[0037] In the embodiment of the present invention, the simulated diffusion aerosol data and meteorological environment data are subjected to spatiotemporal collaborative analysis. First, the two types of data are aligned in time series to ensure that the time windows of the data are consistent, and then spatial interpolation and alignment are performed to ensure the consistency of the data in space. The adaptive kernel density estimation (AKDE) method is applied to smooth the analysis results and generate a density distribution map of aerosol diffusion. This method is used to generate a multi-dimensional hazard heat map, which intuitively displays the hazard intensity and regional distribution of aerosol diffusion, providing visual support for the evaluation.
[0038] Step S6: Perform diffusion hazard assessment on aerosol sample data based on the multi-dimensional hazard heat map and simulated diffusion aerosol data, and generate a hazard assessment report to perform microbial aerosol diffusion hazard assessment based on multi-dimensional data.
[0039] In an embodiment of the present invention, a diffusion hazard assessment is performed based on a multi-dimensional hazard heat map and simulated diffusion aerosol data. First, the aerosol sample data is analyzed in detail in combination with the hazardous area information in the heat map to identify high-risk areas and diffusion trends. Then, a risk assessment model (such as a risk matrix model or Monte Carlo simulation) is used to calculate the hazard level of each area. Finally, a hazard assessment report is generated based on the assessment results, which describes in detail the potential hazards and impacts of aerosol diffusion, and provides targeted suggestions and prevention and control measures to formulate effective response strategies.
[0040] Preferably, step S1 comprises the following steps:
[0041] Step S11: Acquire meteorological environment data; perform spatiotemporal interpolation processing on the meteorological environment data to generate a unified grid meteorological environment data set;
[0042] Step S12: performing singular value analysis on the unified grid meteorological environment data set to obtain a principal component meteorological environment matrix;
[0043] Step S13: locating the outlier interval of the main component meteorological environment matrix and generating an outlier interval index;
[0044] Step S14: performing outlier data analysis on the principal component meteorological environment matrix based on the outlier interval index to obtain an outlier data subset;
[0045] Step S15: performing tensor interpolation completion processing on the outlier data subset to generate completed outlier data;
[0046] Step S16: interpolating outlier factors of the principal component meteorological environment matrix according to the completed outlier data to generate filled meteorological environment data.
[0047] The benefit of the present invention is that by performing spatiotemporal interpolation processing on meteorological environmental data, a unified grid data set is generated, which effectively solves the inconsistency problem of the original data in space and time. Subsequent singular value analysis and outlier interval positioning can accurately identify abnormal data points, and repair them by tensor interpolation completion, thereby improving the integrity and reliability of meteorological environmental data. This lays a solid foundation for subsequent data analysis. The outlier factor interpolation is performed on the completed principal component meteorological environmental matrix to generate filled meteorological environmental data. This method can not only solve the problem of missing data, but also extract the most critical meteorological environmental features that affect aerosol diffusion, and provide high-quality input data for subsequent diffusion simulation and hazard assessment. Through the above-mentioned preprocessing steps, high-quality meteorological environmental data has been obtained. This provides favorable conditions for subsequent fusion with aerosol sample data, can better realize the coupling analysis between meteorological environmental factors and aerosol characteristics, and improve the effectiveness of multi-source data fusion. Based on the filled meteorological environmental data and aerosol sample characteristics, the diffusion direction and speed of aerosols can be more accurately estimated, thereby providing more reliable input parameters for multi-scale diffusion simulation. This will greatly improve the accuracy of diffusion simulation results and lay a solid foundation for subsequent hazard assessment. It also lays a solid foundation for the multi-dimensional data microbial aerosol diffusion hazard assessment method, which is conducive to improving the reliability and effectiveness of the overall analysis process.
[0048] In an embodiment of the present invention, the process of obtaining meteorological environment data includes collecting data on temperature, humidity, wind speed, and air pressure environmental parameters in real time through meteorological sensor systems deployed in different geographical locations. After data collection, spatiotemporal interpolation processing is performed to fill in the gaps and missing values in time and space. In a specific implementation, a three-dimensional spatial interpolation method (such as Kriging interpolation) and a time series interpolation technique (such as spline interpolation) are used to integrate data at different times and locations to generate a meteorological environment data set of a unified grid. By calculating the covariance matrix of the data set, its eigenvalues and eigenvectors are obtained. Using the singular value decomposition (SVD) technique, the meteorological environment data matrix is decomposed into the product of three matrices, including a singular value matrix and a principal component matrix. The results of the singular value analysis are used to extract the principal components and generate a principal component meteorological environment matrix. The matrix retains the main variation information in the data set and reduces the dimension of the data, so that the main features of the meteorological environment data are clearly represented. Statistical methods, such as Z-score analysis or box plot analysis, are used to identify outlier intervals in the principal component matrix. These outlier intervals are usually manifested as values that deviate greatly from the normal data set. In the process of locating outlier intervals, the range of outliers is determined and marked as outlier interval indexes. The generated outlier interval index will indicate which data points belong to the outlier range in the principal component meteorological environment matrix, providing a basis for further outlier data analysis. The outlier data points are located by indexing, and these abnormal data points are extracted to form an outlier data subset. In this step, outlier detection algorithms such as DBSCAN (density-based spatial clustering algorithm) or LOF (local outlier factor) are used to further analyze the characteristics and distribution of outlier data. The outlier data subset is processed by tensor interpolation. First, the outlier data subset is regarded as the missing part in the tensor data. Using tensor completion techniques, such as tensor singular value decomposition (Tucker decomposition) or tensor completion algorithms, the missing values are predicted and completed by using known data points. Tensor interpolation completion processing will generate completed outlier data by combining adjacent data in space and time through low-rank tensor approximation technology. The completed outlier data is used as the interpolation factor to interpolate the abnormal data points in the principal component meteorological environment matrix. Through the interpolation algorithm (such as linear regression interpolation or multiple interpolation), the value of the completed outlier data is filled into the outlier data position in the principal component matrix to generate the filled meteorological environment data. This step ensures the improvement of the data integrity and accuracy of the principal component meteorological environment matrix by repairing the outlier data points, providing a high-quality data foundation for subsequent analysis.
[0049] Preferably, step S2 comprises the following steps:
[0050] Step S21: Acquire aerosol sample data; perform multi-scale wavelet decomposition processing on the aerosol sample data to obtain a multi-resolution characteristic coefficient set;
[0051] Step S22: performing basic non-negative matrix decomposition processing on the multi-resolution feature coefficient set to generate a basic component matrix;
[0052] Step S23: clustering the basic component matrix into taxonomic units to obtain a taxonomic unit clustering component matrix;
[0053] Step S24: extracting shape contour features from the classification unit cluster component matrix to generate cluster unit contour point feature data;
[0054] Step S25: performing discrete Fourier transform processing on the cluster unit contour point feature data to obtain a Fourier descriptor;
[0055] Step S26: Perform collaborative vector concatenation on the cluster unit contour point feature data and the Fourier descriptor to generate an aerosol sample feature vector.
[0056] The benefit of the present invention is that by performing multi-scale wavelet decomposition on aerosol sample data, characteristic coefficient sets at different resolutions can be obtained, which fully captures the multi-scale characteristic information of aerosol particles. This provides a richer data basis for subsequent feature extraction and pattern recognition. Basic non-negative matrix decomposition can extract basic components from aerosol data, reflecting the characteristics of different aerosol components. Further clustering of classification units can classify these basic components to obtain clearer aerosol component characteristics. This is conducive to identifying and distinguishing aerosols of different sources and components. By performing discrete Fourier transform on the clustering unit contour point feature data, the Fourier descriptor of aerosol particles can be obtained. This description method based on shape features can more finely characterize the geometric features of aerosol particles and provide richer information for subsequent morphological analysis. Finally, the clustering unit contour point feature data and the Fourier descriptor are collaboratively vectored to generate a comprehensive feature vector of the aerosol sample. This method of integrating multi-dimensional features can fully reflect the geometric structure, component characteristics and multi-scale information of aerosols, providing more accurate input parameters for subsequent diffusion simulation and hazard assessment. It lays a solid foundation for the multi-dimensional data microbial aerosol diffusion hazard assessment method. Through multi-scale expression, component analysis, shape characterization and feature fusion technology, the information richness and description accuracy of aerosol sample data are effectively improved, providing high-quality input data for subsequent diffusion simulation and hazard assessment.
[0057] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:
[0058] Step S21: Acquire aerosol sample data; perform multi-scale wavelet decomposition processing on the aerosol sample data to obtain a multi-resolution characteristic coefficient set;
[0059] In an embodiment of the present invention, aerosol sample data is obtained by a high-precision aerosol sampling device, and these data contain the concentration information of aerosol particles at different time points and spatial positions. The acquired data is subjected to multi-scale wavelet decomposition processing to extract features at different scales. The original aerosol data is processed using discrete wavelet transform (DWT), and an appropriate wavelet basis function (such as Daubechies wavelet or Symlet wavelet) is selected to decompose the data into approximation coefficients and detail coefficients at multiple scales. The coefficients of each scale capture the changes in aerosol data at different resolutions, thereby generating a multi-resolution feature coefficient set that reflects the characteristic distribution of aerosol sample data at different scales.
[0060] Step S22: performing basic non-negative matrix decomposition processing on the multi-resolution feature coefficient set to generate a basic component matrix;
[0061] In an embodiment of the present invention, a basic non-negative matrix decomposition (NMF) technique is applied to the generated multi-resolution feature coefficient set. First, a non-negative matrix decomposition model is constructed, in which the original feature coefficient set is decomposed into the product of two non-negative matrices, namely, a basic component matrix and a coefficient matrix. The multi-resolution feature coefficient set is decomposed using an NMF algorithm (such as an ALS algorithm or a multiplication update rule) to extract basic components that represent the main patterns and structures of the aerosol sample data. The basic component matrix finally generated contains the core feature information of the aerosol sample, and the data representation and subsequent processing process are simplified through matrix decomposition technology.
[0062] Step S23: clustering the basic component matrix into taxonomic units to obtain a taxonomic unit clustering component matrix;
[0063] In an embodiment of the present invention, a classification unit clustering algorithm is applied to the basic component matrix for processing. K-means (k-means) clustering or spectral clustering algorithm is used to classify the data in the basic component matrix. A reasonable number of clusters (K value) is set to group the data points in the basic component matrix into different clusters, each cluster representing a classification unit of the aerosol sample. The clustering results generate a classification unit clustering component matrix, which reflects the distribution pattern of different aerosol samples in the feature space and provides structured information for further feature extraction and analysis.
[0064] Step S24: extracting shape contour features from the classification unit cluster component matrix to generate cluster unit contour point feature data;
[0065] In an embodiment of the present invention, shape contour feature extraction is performed on the classification unit cluster component matrix. First, the sample data of each classification unit is mapped to a two-dimensional space to form a shape contour map. The shape contour points of each cluster unit are extracted by applying an edge detection algorithm (such as Canny edge detection) and a contour extraction technique. These contour points represent the shape characteristics of the aerosol sample in space. The generated cluster unit contour point feature data includes the contour coordinates and shape description information of each unit, providing a detailed characterization of the geometric characteristics of the aerosol sample.
[0066] Step S25: performing discrete Fourier transform processing on the cluster unit contour point feature data to obtain a Fourier descriptor;
[0067] In an embodiment of the present invention, discrete Fourier transform (DFT) processing is performed on the feature data of the contour points of the clustering unit. First, the coordinate data of the contour points are converted into complex form to construct Fourier transform input data. The fast Fourier transform (FFT) algorithm is applied to perform frequency domain analysis on the data to obtain Fourier descriptors. These Fourier descriptors represent the frequency characteristics of the contour shape and capture the periodicity and texture information of the shape. The Fourier descriptor provides a mathematical shape feature representation for the aerosol sample, which facilitates subsequent feature fusion and analysis.
[0068] Step S26: Perform collaborative vector concatenation on the cluster unit contour point feature data and the Fourier descriptor to generate an aerosol sample feature vector.
[0069] In an embodiment of the present invention, the cluster unit contour point feature data and the Fourier descriptor are collaboratively vector spliced. First, the geometric shape information extracted from the contour point feature data is aligned with the frequency domain information in the Fourier descriptor. Through vector splicing, these two parts of feature information are merged into a unified feature vector. This vector combines the shape characteristics and frequency domain characteristics of the aerosol sample to form an aerosol sample feature vector, which is used to describe the all-round characteristics of the aerosol sample. This step provides a rich data basis for further analysis and evaluation of aerosol samples by fusing multiple feature information.
[0070] Preferably, step S3 comprises the following steps:
[0071] Step S31: performing wind field vector decomposition on the filled meteorological environment data to generate wind speed component and wind direction component;
[0072] Step S32: performing a rotation coordinate transformation on the aerosol sample characteristic vector based on the wind speed component and the wind direction component to obtain a coordinate system aerosol characteristic vector;
[0073] Step S33: performing main direction projection processing on the aerosol sample characteristic vector according to the coordinate system aerosol characteristic vector and the wind direction component to obtain aerosol diffusion direction data;
[0074] Step S34: performing time-delay diffusion velocity estimation on the aerosol sample characteristic vector based on the aerosol diffusion direction data and the wind speed component to generate aerosol diffusion velocity data.
[0075] The benefit of the present invention lies in that the wind speed component and wind direction component parameters are obtained by decomposing the filled meteorological environmental data into wind field vectors. This provides necessary environmental information support for the subsequent coordinate transformation and main direction projection of the aerosol sample characteristic vector, so that the estimation of the aerosol diffusion direction and diffusion speed is closer to the actual environmental conditions. Based on the wind speed component and wind direction component, the aerosol sample characteristic vector is subjected to a rotation coordinate transformation and converted into a coordinate system consistent with the environmental wind field. This step ensures that the subsequent diffusion direction and diffusion speed estimation are more accurate and avoids the deviation caused by inconsistent coordinate systems. Using the coordinate system aerosol characteristic vector and wind direction component, the main direction of aerosol diffusion can be effectively extracted by means of main direction projection. This diffusion direction estimation method based on characteristic vector and wind direction information is more accurate and reliable than relying solely on wind direction data. Combining aerosol diffusion direction data and wind speed component, the time-delay diffusion speed estimation of the aerosol sample characteristic vector is performed. This method of combining diffusion direction and wind speed information can more accurately reflect the diffusion process and propagation speed of aerosols in the actual environment, providing more reliable input parameters for subsequent hazard assessment. The effective fusion of environmental wind field information, the coordinate system transformation of aerosol characteristic vectors, the accurate estimation of diffusion direction, and the time-lag estimation of diffusion speed provide more accurate and reliable data support for the multi-dimensional data microbial aerosol diffusion hazard assessment method. These technical means help to better simulate and predict the diffusion process and hazard characteristics of aerosols in the actual environment, and provide more refined decision-making basis for subsequent risk assessment and emergency management.
[0076] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:
[0077] Step S31: performing wind field vector decomposition on the filled meteorological environment data to generate wind speed component and wind direction component;
[0078] In an embodiment of the present invention, the filled meteorological environment data contains the spatial distribution information of meteorological parameters, including wind speed and wind direction. The wind field information in the filled meteorological environment data is decomposed into wind speed components and wind direction components using a wind field vector decomposition algorithm. First, a vector field analysis is performed on the meteorological environment data, and the principal component analysis (PCA) or singular value decomposition (SVD) method is applied to convert the wind field data from a complex multidimensional representation into wind speed components and wind direction components. These wind speed components represent the wind speed at each location, while the wind direction component provides information on the wind speed direction. The generated wind speed component and wind direction component will provide basic data for subsequent aerosol sample feature vector processing.
[0079] Step S32: performing a rotation coordinate transformation on the aerosol sample characteristic vector based on the wind speed component and the wind direction component to obtain a coordinate system aerosol characteristic vector;
[0080] In the embodiment of the present invention, a wind field coordinate system is established, the wind direction component defines the direction of the coordinate axis, and the wind speed component is used to adjust the scale of the coordinate axis. The wind field coordinate system is established based on the wind speed component V and the wind direction component θ, where the wind speed component V represents the magnitude of the wind speed, and the wind direction component θ represents the angle of the wind direction (the deflection angle relative to a reference direction, such as the north direction). The wind direction component θ is used to construct the rotation matrix R(θ), ;
[0081] Where θ is the wind direction component, and the angle unit is usually radians. This matrix rotates the aerosol sample feature vector in the original coordinate system to the wind field coordinate system so that it is aligned with the wind field. Apply the rotation matrix to transform the aerosol sample feature vector from the original coordinate system to the wind field coordinate system. In this process, the wind direction component is used to construct a rotation matrix, the aerosol feature vector is adjusted according to the wind direction, and the wind speed component is used to adjust the scale to generate a coordinate system aerosol feature vector. This transformation aligns the feature vector of the aerosol sample to a coordinate system consistent with the wind field, providing a standardized feature representation for subsequent data analysis.
[0082] Step S33: performing main direction projection processing on the aerosol sample characteristic vector according to the coordinate system aerosol characteristic vector and the wind direction component to obtain aerosol diffusion direction data;
[0083] In an embodiment of the present invention, a high-order tensor decomposition is performed using a coordinate system aerosol feature vector. A high-dimensional tensor is constructed to represent the aerosol feature data in the form of a multidimensional matrix. Then, a tensor decomposition technique, such as CP decomposition or Tucker decomposition, is applied to decompose the tensor into multiple factor matrices. The obtained principal component matrix describes the distribution of aerosol features in different principal component directions, and these principal component matrices provide the main feature representations of aerosol features in various directions. The wind direction component is projected using the principal component matrix. The wind direction component is first converted into a unit vector, and then mapped into the principal component matrix to determine the performance of the wind direction in the principal component space. This process includes calculating the inner product of the wind direction component and the principal component matrix to obtain a projection vector representing the wind direction in the principal component space. The projection vector represents the position and influence of the wind direction in the principal component space of the aerosol sample. The wind direction projection vector is subjected to multi-scale orthogonal decomposition, usually using wavelet transform technology. By decomposing the wind direction projection vector, characteristic components at different scales are obtained. These components represent the information of aerosol diffusion in different directions and form multi-directional data of aerosol diffusion. These data provide wind direction information at different scales, which helps to understand the multi-directional diffusion characteristics of aerosols. The principal component matrix is weighted reconstructed using the multi-directional data of aerosol diffusion. The weighted reconstructed data is obtained by applying each multi-directional data component to the corresponding direction of the principal component matrix. This reconstruction process combines the wind direction component and the aerosol diffusion characteristics. The generated aerosol diffusion direction data provides detailed diffusion information of aerosols in different directions for subsequent diffusion analysis and evaluation.
[0084] Step S34: performing time-delay diffusion velocity estimation on the aerosol sample characteristic vector based on the aerosol diffusion direction data and the wind speed component to generate aerosol diffusion velocity data.
[0085] In an embodiment of the present invention, a delayed embedding method is applied to aerosol diffusion direction data to create a time-delay embedding matrix. The specific operation includes selecting an appropriate embedding dimension and time lag step size, and reorganizing the aerosol diffusion direction data according to these parameters to form a high-dimensional matrix. In this matrix, the time lag points are used as the columns of the matrix, and the diffusion direction data at each time point forms the rows of the matrix. This processing method converts the time series data into a high-dimensional state space, which is helpful for the subsequent analysis of the dynamic characteristics of aerosol diffusion. The wind speed component is decomposed into a multi-scale time series using the data in the time-delay embedding matrix. Select a suitable wavelet transform or other time series decomposition method to decompose the wind speed component into components on different time scales. Each scale represents the behavior of the wind speed data at different frequencies. Through this decomposition, multiple wind speed component sequences are obtained, which contain the dynamic characteristics of the wind speed data at various time scales and provide basic data for the next similarity matching. Similarity matching is performed on the obtained multi-scale wind speed component sequence to generate a time-delay diffusion speed association matrix. First, each wind speed component sequence is compared with the corresponding data in the time-delay embedding matrix, and the similarity measure between them is calculated. Commonly used measures include Euclidean distance or dynamic time warping distance. The calculated similarity scores are filled into the association matrix, and each element of the matrix represents the degree of match between the wind speed and aerosol diffusion direction at the corresponding time points. This association matrix reveals the relationship between wind speed and aerosol diffusion at different time lags. The time-delay diffusion velocity association matrix is used to perform main direction weighted averaging of the aerosol sample feature vector. First, each value in the time-delay diffusion velocity association matrix is used as a weighting coefficient and applied to each component of the aerosol sample feature vector. The product of each feature vector component and its corresponding weight is calculated, and then the weighted results of all components are summed to obtain the final aerosol diffusion velocity data. This process integrates the information of different time lags and provides a comprehensive diffusion velocity estimate, which is used to describe the diffusion behavior of aerosols under different time and space conditions.
[0086] Preferably, performing main direction projection processing on the aerosol sample characteristic vector according to the coordinate system aerosol characteristic vector and the wind direction component comprises the following steps:
[0087] Perform high-order tensor decomposition on the coordinate system aerosol eigenvector to obtain the coordinate system tensor principal component matrix;
[0088] Perform principal component projection on the wind direction component based on the principal component matrix of the coordinate system tensor to obtain the wind direction projection vector in the principal component space;
[0089] Perform multi-scale orthogonal decomposition on the principal component space wind direction projection vector to generate multi-directional data of aerosol diffusion;
[0090] According to the multi-directional data of aerosol diffusion, the principal component matrix of the coordinate system tensor is reconstructed by weighting the components in each direction to generate the aerosol diffusion direction data.
[0091] The benefit of the present invention is that by performing high-order tensor decomposition processing on the coordinate system aerosol feature vector, the principal component matrix of the tensor can be extracted. This helps to capture the main change patterns in the characteristics of the aerosol sample and provide support for the subsequent principal component projection and direction reconstruction. Based on the coordinate system tensor principal component matrix, the wind direction component is projected on the principal component to obtain the wind direction projection vector in the principal component space. This method of integrating wind direction information into the principal component space ensures that the subsequent aerosol diffusion direction estimation can be better combined with the environmental wind field. Multi-scale orthogonal decomposition of the wind direction projection vector in the principal component space can generate multi-directional data of aerosol diffusion. This multi-scale decomposition method can more comprehensively describe the diffusion characteristics of aerosols at different time and space scales and provide richer data support for hazard assessment. According to the generated multi-directional data of aerosol diffusion, the coordinate system tensor principal component matrix is weightedly reconstructed for each direction component, and finally the aerosol diffusion direction data is obtained. This weighted reconstruction method based on the principal component matrix can better reflect the dominant diffusion direction of aerosols in the actual environment and provide accurate input for subsequent hazard assessment. By making full use of aerosol characteristic vectors, wind direction information and multi-scale analysis techniques, accurate modeling and estimation of aerosol diffusion direction are achieved. This provides a more reliable and detailed data basis for the multi-dimensional data microbial aerosol diffusion hazard assessment method, helps to better predict and evaluate the aerosol diffusion process and hazard characteristics in the actual environment, and provides a more accurate basis for emergency management and decision support.
[0092] In an embodiment of the present invention, first, a high-order tensor decomposition is performed using a coordinate system aerosol feature vector. A high-dimensional tensor is constructed to represent the aerosol feature data in the form of a multidimensional matrix. Then, a tensor decomposition technique, such as CP decomposition or Tucker decomposition, is applied to decompose the tensor into multiple factor matrices. The obtained principal component matrix describes the distribution of aerosol features in different principal component directions, and these principal component matrices provide the main feature representations of aerosol features in various directions. The wind direction component is projected using the principal component matrix. The wind direction component is first converted into a unit vector, and then mapped into the principal component matrix to determine the performance of the wind direction in the principal component space. This process includes calculating the inner product of the wind direction component and the principal component matrix to obtain a projection vector representing the wind direction in the principal component space. The projection vector represents the position and influence of the wind direction in the principal component space of the aerosol sample. The wind direction projection vector is subjected to multi-scale orthogonal decomposition, usually using wavelet transform technology. By decomposing the wind direction projection vector, characteristic components at different scales are obtained. These components represent the information of aerosol diffusion in different directions and form multi-directional data of aerosol diffusion. These data provide wind direction information at different scales, which helps to understand the multi-directional diffusion characteristics of aerosols. The principal component matrix is weighted reconstructed using the multi-directional data of aerosol diffusion. The weighted reconstructed data is obtained by applying each multi-directional data component to the corresponding direction of the principal component matrix. This reconstruction process combines the wind direction component and the aerosol diffusion characteristics, and the generated aerosol diffusion direction data provides detailed diffusion information of aerosols in different directions for subsequent diffusion analysis and evaluation.
[0093] Preferably, estimating the time-delayed diffusion velocity of the aerosol sample characteristic vector based on the aerosol diffusion direction data and the wind speed component comprises the following steps:
[0094] Perform delay embedding processing on aerosol diffusion direction data to obtain a time-delay embedding matrix;
[0095] Based on the time-delay embedding matrix, the wind speed component is decomposed into a multi-scale time series to obtain a multi-scale wind speed component series;
[0096] Similarity matching is performed on multi-scale wind speed component sequences to generate a time-delay diffusion velocity correlation matrix;
[0097] The aerosol diffusion velocity data were obtained by performing weighted averaging of the aerosol sample characteristic vectors in the main direction according to the time-delay diffusion velocity correlation matrix.
[0098] The benefit of the present invention is that the aerosol diffusion direction data is subjected to delayed embedding processing to obtain a time-delay embedding matrix. This time-delay embedding method can capture the time correlation in the aerosol diffusion process and provide a basis for the subsequent multi-scale time series decomposition. Based on the time-delay embedding matrix, the wind speed component is subjected to multi-scale time series decomposition to obtain a wind speed component sequence at different time scales. This multi-scale decomposition method can more comprehensively reflect the influence of the wind field on aerosol diffusion and provide support for the estimation of the time-delay diffusion speed. By performing similarity matching on the multi-scale wind speed component sequence, a time-delay diffusion speed correlation matrix can be generated. This similarity-based matching method can better capture the correlation characteristics of aerosols with wind fields at different time and space scales and provide a basis for subsequent weighted averaging. According to the time-delay diffusion speed correlation matrix, the aerosol sample feature vector is weighted averaged in the main direction to obtain aerosol diffusion speed data. This weighted average method based on the correlation matrix can comprehensively consider the aerosol diffusion direction and wind speed component factors to obtain a more accurate diffusion speed estimation result. By making full use of aerosol diffusion direction data and wind speed component information, accurate estimation of aerosol diffusion rate is achieved through time-delay embedding, multi-scale time series analysis and similarity matching technology. This provides more reliable and detailed data support for the multi-dimensional data microbial aerosol diffusion hazard assessment method, helps to better predict and evaluate the aerosol diffusion process and hazard characteristics in the actual environment, and provides a more reliable basis for emergency management and decision support.
[0099] In an embodiment of the present invention, first, a delayed embedding method is applied to the aerosol diffusion direction data to create a time-delay embedding matrix. The specific operation includes selecting an appropriate embedding dimension and a time lag step size, and reorganizing the aerosol diffusion direction data according to these parameters to form a high-dimensional matrix. In this matrix, the time lag points are used as the columns of the matrix, and the diffusion direction data at each time point forms the rows of the matrix. This processing method converts the time series data into a high-dimensional state space, which is helpful for the subsequent analysis of the dynamic characteristics of aerosol diffusion. The wind speed component is decomposed into a multi-scale time series using the data in the time-delay embedding matrix. Select a suitable wavelet transform or other time series decomposition method to decompose the wind speed component into components on different time scales. Each scale represents the behavior of the wind speed data at different frequencies. Through this decomposition, multiple wind speed component sequences are obtained, which contain the dynamic characteristics of the wind speed data at various time scales and provide basic data for the next similarity matching. Similarity matching is performed on the obtained multi-scale wind speed component sequence to generate a time-delay diffusion speed association matrix. First, each wind speed component sequence is compared with the corresponding data in the time-delay embedding matrix, and the similarity measure between them is calculated. Commonly used measures include Euclidean distance or dynamic time warping distance. The calculated similarity scores are filled into the association matrix, and each element of the matrix represents the degree of match between the wind speed and aerosol diffusion direction at the corresponding time points. This association matrix reveals the relationship between wind speed and aerosol diffusion at different time lags. The time-delay diffusion velocity association matrix is used to perform main direction weighted averaging of the aerosol sample feature vector. First, each value in the time-delay diffusion velocity association matrix is used as a weighting coefficient and applied to each component of the aerosol sample feature vector. The product of each feature vector component and its corresponding weight is calculated, and then the weighted results of all components are summed to obtain the final aerosol diffusion velocity data. This process integrates the information of different time lags and provides a comprehensive diffusion velocity estimate, which is used to describe the diffusion behavior of aerosols under different time and space conditions.
[0100] Preferably, step S4 comprises the following steps:
[0101] Step S41: performing multi-dimensional Gaussian kernel density estimation on aerosol sample data according to aerosol diffusion direction data to obtain preliminary diffusion probability distribution data;
[0102] Step S42: performing multi-scale decomposition on the preliminary diffusion probability distribution data to generate a multi-scale diffusion probability matrix;
[0103] Step S43: performing time series convolution processing on the multi-scale diffusion probability matrix according to the aerosol diffusion velocity data to obtain a multi-scale diffusion simulation matrix;
[0104] Step S44: performing reverse mapping processing on the multi-scale diffusion simulation matrix to obtain simulated diffusion aerosol data.
[0105] The benefit of the present invention is that by performing multi-dimensional Gaussian kernel density estimation on aerosol diffusion direction data, preliminary diffusion probability distribution data can be obtained. This probability density-based method can better reflect the probability of aerosol diffusion in different directions, providing a basis for subsequent multi-scale decomposition. Multi-scale decomposition of the preliminary diffusion probability distribution data can generate a multi-scale diffusion probability matrix. This multi-scale decomposition method can capture the probability characteristics of aerosol diffusion at different time scales, and provide rich data support for subsequent time series convolution processing. According to the aerosol diffusion velocity data, the multi-scale diffusion probability matrix is subjected to time series convolution processing to obtain a multi-scale diffusion simulation matrix. This method based on time series convolution can organically combine the two key factors of diffusion probability and diffusion velocity to generate more accurate and detailed diffusion simulation data. The multi-scale diffusion simulation matrix is subjected to reverse mapping processing to obtain simulated diffusion aerosol data. This reverse mapping method can convert the multi-scale simulation results into specific aerosol diffusion data, providing intuitive and usable input data for subsequent hazard analysis and assessment.
[0106] In an embodiment of the present invention, preliminary diffusion probability distribution data is generated by a Gaussian kernel density estimation method according to aerosol diffusion direction data. First, an appropriate Gaussian kernel function is selected for each aerosol sample data point, and a two-dimensional Gaussian kernel function is usually used to describe the local characteristics of the aerosol diffusion direction. The width (bandwidth parameter) of the Gaussian kernel function determines the smoothness of the kernel density estimation. Then, the Gaussian kernel functions of all aerosol sample data points are superimposed to form a smooth density estimation map. The map provides the diffusion probability distribution of aerosol samples at different locations, and these distributions are used to reflect the diffusion trend of aerosols in space. Multiscale decomposition technology is applied to the preliminary diffusion probability distribution data to generate a multiscale diffusion probability matrix. Wavelet transform or scale recursive decomposition is selected as a decomposition tool to decompose the preliminary diffusion probability distribution data into data at different scales. In this process, the original data is decomposed into components of different frequencies, each of which represents the diffusion characteristics of the data at a specific scale. The generated multiscale diffusion probability matrix contains the diffusion probability information at each scale. These matrices describe the diffusion characteristics of aerosols at multiple spatial scales, which are helpful for subsequent simulation and analysis. According to the aerosol diffusion velocity data, the multiscale diffusion probability matrix is subjected to time series convolution processing to generate a multiscale diffusion simulation matrix. First, appropriate convolution kernels are selected, whose shapes and parameters are determined by the aerosol diffusion velocity data. The convolution kernels are used to capture the diffusion behavior of aerosols at different time scales. These convolution kernels are convolved element by element with the multiscale diffusion probability matrix to calculate the diffusion simulation results at each scale. This processing process combines the timeliness of the aerosol diffusion velocity data to generate a matrix containing multiscale diffusion simulation data, which is used to describe the aerosol diffusion simulation in the time dimension. The multiscale diffusion simulation matrix is subjected to reverse mapping processing to obtain simulated diffuse aerosol data. First, the mapping relationship between the original data space and the diffusion simulation matrix is determined, which usually involves mapping the data points of the simulation matrix back to the original data space. In the reverse mapping process, the inverse transformation or interpolation method is applied to restore the data of the multiscale diffusion simulation matrix to the original spatial coordinate system. This process ensures that the simulation results can accurately reflect the diffusion of aerosols in the actual environment, and the generated simulated diffuse aerosol data are used for further analysis and evaluation.
[0107] Preferably, step S5 comprises the following steps:
[0108] Step S51: Perform spatiotemporal gridding on the simulated diffuse aerosol data to generate a global spatiotemporal grid data set:
[0109] Step S52: performing a bidirectional multivariate regression analysis on the global spatiotemporal grid data set and the meteorological environment data to obtain a spatiotemporal regression matrix;
[0110] Step S53: performing unit local kernel density estimation on the spatiotemporal regression matrix to generate local hazard density data;
[0111] Step S54: Perform hazard density interpolation on the global spatiotemporal grid data set based on the local hazard density data to generate a multi-dimensional hazard heat map.
[0112] The benefit of the present invention is that by performing spatiotemporal gridding processing on the simulated diffusion aerosol data, a global spatiotemporal grid data set covering the entire study area can be generated. This gridding processing method can systematize and standardize the original diffusion data, providing a unified data basis for subsequent spatial analysis and hazard assessment. A two-way multivariate regression analysis can be performed on the global spatiotemporal grid data set and the meteorological environment data to obtain a spatiotemporal regression matrix. This method based on regression analysis can dig out the complex relationship between the aerosol diffusion process and environmental factors, providing an important basis for the subsequent local hazard density estimation. The unit local kernel density estimation of the spatiotemporal regression matrix can generate local hazard density data. This method based on kernel density estimation can better reflect the distribution characteristics of aerosol diffusion hazards within different grid units, and provide more refined data input for subsequent global hazard interpolation. According to the local hazard density data, the hazard density interpolation of the global spatiotemporal grid data set can generate a multi-dimensional hazard heat map. This interpolation-based method can organically integrate local hazard density information into the spatial pattern of the entire study area, forming an intuitive hazard risk distribution map, and providing a more visual reference for decision support and emergency management. The simulation of diffuse aerosol data and meteorological environment data information was fully utilized, and the multi-dimensional simulation and visualization of aerosol diffusion hazards were achieved through spatiotemporal gridding, regression analysis, local kernel density estimation, and spatial interpolation techniques. This provides more comprehensive and sophisticated technical support for the multi-dimensional data microbial aerosol diffusion hazard assessment method, helps to better analyze and predict the hazard distribution characteristics of aerosols in the actual environment, and provides a more intuitive and reliable reference for emergency management and decision support.
[0113] In an embodiment of the present invention, appropriate time and space resolutions are selected to map the simulated diffuse aerosol data to a unified space-time grid system. The time resolution is set to hours or days, while the space resolution is determined according to specific application requirements, such as per square kilometer or per ten kilometers. The original simulated data points are mapped to predetermined grid nodes by interpolation methods, and each grid node is assigned a corresponding aerosol concentration value and diffusion information. The global space-time grid data set generated by this process provides uniformly distributed aerosol diffusion information, laying the foundation for subsequent data analysis. A bidirectional regression model is selected, with the aerosol diffusion characteristics of the global space-time grid data set and the relevant variables of the meteorological environment data as independent variables. The meteorological environment data includes temperature, humidity, and wind speed, while the aerosol data includes concentration and diffusion speed. A regression equation is established, and multivariate regression analysis is performed on the independent variables and the dependent variables to calculate the regression coefficient. The space-time regression matrix generated by the regression analysis results describes the influence of meteorological environmental factors on aerosol diffusion behavior, as well as the prediction of aerosol concentration under different time and space conditions. The data points in the space-time regression matrix are divided into several small spatial units, and local kernel density estimation is performed for each unit. Select a suitable kernel function (such as Gaussian kernel function) and determine the bandwidth parameter, and smooth the aerosol concentration of each small unit through kernel density estimation. In the process of local kernel density estimation, the data points in the unit are combined with the weights of the surrounding points to generate the local hazard density value of each unit. These local density values reflect the aerosol hazard level in different regions and time periods, providing a data basis for further heat map generation. Select a suitable interpolation method, such as Kriging interpolation or inverse distance weighted interpolation, to expand the local hazard density data to the global spatiotemporal grid dataset. The interpolation method estimates the aerosol hazard density value of the grid node by combining the local density data and the spatial position of the grid node. The generated multi-dimensional hazard heat map will show the aerosol hazard situation in different regions and time periods, and the depth of the color in the figure indicates the intensity of the hazard density. This heat map provides decision makers with a detailed view of the hazard distribution, which helps to assess the potential risk of aerosol diffusion.
[0114] Preferably, step S52 includes the following steps:
[0115] Step S521: performing feature space decomposition processing on the global spatiotemporal grid data set to obtain a feature variable matrix;
[0116] Step S522: performing linear interpolation processing on the characteristic variable matrix according to the meteorological environment data to generate a time series interpolation matrix;
[0117] Step S523: performing a bidirectional cross regression process on the time series interpolation matrix and the characteristic variable matrix to obtain a regression coefficient matrix;
[0118] Step S524: Perform spatial redistribution processing on the characteristic variables according to the regression coefficient matrix to obtain a spatiotemporal regression matrix.
[0119] The invention is beneficial in that the feature space decomposition process is performed on the global spatiotemporal grid data set, and a feature variable matrix can be obtained. This method can effectively extract the key feature variables contained in the data set, laying the foundation for subsequent time series interpolation and two-way cross regression analysis. According to the meteorological environment data, the feature variable matrix is subjected to linear interpolation processing, and a time series interpolation matrix can be generated. This interpolation method can make up for the problem of missing or discontinuous time series in the original data, so that the input data of the subsequent regression analysis is more complete and continuous. The time series interpolation matrix and the feature variable matrix are subjected to two-way cross regression processing to obtain a regression coefficient matrix. This two-way regression analysis method can capture the mutual influence relationship between the feature variables and the time series, and provide a more scientific and reliable basis for the generation of the final spatiotemporal regression matrix. According to the regression coefficient matrix, the feature variables are subjected to spatial redistribution processing to obtain a spatiotemporal regression matrix. This spatial redistribution method based on the regression coefficient can organically combine the time series information with the spatial distribution characteristics to form a more complete and rich spatiotemporal regression matrix, and provide more accurate data input for the subsequent local hazard density estimation and global hazard heat map construction. By making full use of the spatiotemporal characteristics of multi-source data, and through a series of technical means such as feature space decomposition, time series interpolation, two-way cross regression, and spatial redistribution, we have achieved in-depth exploration and comprehensive modeling of various influencing factors in the aerosol diffusion process. This provides more sophisticated and scientific technical support for the multidimensional data microbial aerosol diffusion hazard assessment method, helps to better characterize and predict the spatiotemporal dynamic characteristics of aerosol diffusion, and provides a more reliable and comprehensive reference for emergency management and decision support.
[0120] In an embodiment of the present invention, a principal component analysis (PCA) method is used to perform feature space decomposition on a spatiotemporal grid data set. The process includes calculating the covariance matrix of the global spatiotemporal grid data set and performing eigenvalue decomposition on the matrix to extract the main characteristic components. The first few main characteristic components are selected as characteristic variables, which can represent the main variability in the data set. The generated characteristic variable matrix includes the coefficients and weights of these characteristic components for further processing and analysis. This processing step reduces the complexity of the data through dimensionality reduction technology and retains the most important information in the data. Time series data points in the meteorological environment data are determined and aligned with the time axis of the characteristic variable matrix. The time series data is interpolated using a linear interpolation method to fill in missing values or discontinuous data in the characteristic variable matrix. The interpolation process smoothes the data points in the characteristic variable matrix according to the trend and change of the time series data. The generated time series interpolation matrix contains characteristic variable data that is continuous and uniform in the time series, providing complete data support for subsequent analysis. A cross regression model of the time series interpolation matrix and the characteristic variable matrix is set. The two-way cross-regression processing includes taking the time series interpolation matrix as the dependent variable and the characteristic variable matrix as the independent variable for regression analysis. Two regression models are established: one model predicts the time series interpolation matrix as a function of the characteristic variable matrix, and the other model predicts the characteristic variable matrix as a function of the time series interpolation matrix. The regression coefficient matrix calculated by regression analysis describes the relationship between the time series data and the characteristic variables, providing a coefficient basis for subsequent spatiotemporal data processing. The characteristic variable matrix is spatially redistributed using the regression coefficient matrix. The specific steps include applying the coefficients of the regression coefficient matrix to each spatial position of the characteristic variable matrix to adjust the distribution of the characteristic variables in space. The weighted values calculated by the regression coefficient matrix are applied to redistribute the data points in the characteristic variable matrix. The resulting spatiotemporal regression matrix integrates the information of the time series interpolation matrix and the characteristic variable matrix, providing comprehensive regression results in space and time. This matrix reflects the aerosol diffusion characteristics under different temporal and spatial conditions, providing detailed spatiotemporal data support for further analysis and evaluation.
[0121] Preferably, step S6 comprises the following steps:
[0122] Step S61: performing sample diffusion mapping on the aerosol sample data based on the simulated diffusion aerosol data to obtain a regional diffusion trend map;
[0123] Step S62: Perform diffusion point hazard assessment on the multi-dimensional hazard heat map and the regional diffusion trend map to obtain a diffusion point hazard assessment data set;
[0124] Step S63: Perform hazard evaluation mapping on the aerosol sample data according to the diffusion point hazard assessment data set, and generate a hazard assessment report to perform microbial aerosol diffusion hazard assessment based on multidimensional data.
[0125] The benefit of the present invention lies in that based on the simulated diffusion aerosol data, the aerosol sample data is subjected to sample diffusion mapping, and a regional diffusion trend map can be obtained. This diffusion mapping method based on sample data can more intuitively reflect the dynamic characteristics of aerosol diffusion in the study area. The diffusion point hazard assessment is performed on the multi-dimensional hazard heat map and the regional diffusion trend map, and a diffusion point hazard assessment data set can be obtained. This comprehensive assessment method based on heat maps and diffusion trends can more comprehensively consider the spatiotemporal dynamic characteristics and hazard density distribution in the aerosol diffusion process. According to the diffusion point hazard assessment data set, the aerosol sample data is subjected to hazard assessment mapping, and a hazard assessment report can be generated. This mapping method based on assessment data can ultimately convert the results of the previous simulation analysis, spatial interpolation and diffusion trend assessment into hazard assessment information for specific sampling points, making full use of the simulated diffusion data, spatial heat maps and diffusion point assessment information, and through a series of technical means such as sample diffusion mapping, comprehensive hazard assessment and result mapping, a comprehensive quantification and accurate assessment of aerosol diffusion hazards are achieved. This provides more complete and reliable technical support for the multidimensional data-based microbial aerosol diffusion hazard assessment method, and helps to better analyze and predict the risk distribution characteristics of aerosols in actual environments.
[0126] In an embodiment of the present invention, the simulated diffuse aerosol data is spatially registered with the aerosol sample data to ensure that the two are in the same geographic coordinate system. Spatial interpolation techniques, such as Kriging interpolation, are applied to map the sample data on the basis of the simulated diffuse aerosol data. The spatial distribution of the aerosol sample data is mapped to the simulated diffuse aerosol data using an interpolation algorithm to generate an image containing a diffusion trend. The regional diffusion trend map shows the diffusion trend of aerosol samples in different regions, providing a visual display of the aerosol diffusion pattern to support further analysis. The multi-dimensional hazard heat map and the regional diffusion trend map are spatially aligned to ensure that the data of the two are consistent in the same geographic reference system. A hazard assessment model is used to perform a hazard assessment on each diffusion point in the heat map and the trend map. The model includes calculating a hazard index for each point, combining the data characteristics of the aerosol sample and the regional diffusion information displayed in the diffusion trend map. The generated diffusion point hazard assessment data set includes the hazard values of each diffusion point and its spatial distribution. The aerosol sample data is subjected to hazard evaluation mapping according to the diffusion point hazard assessment data set to generate a hazard assessment report. First, the diffusion point hazard assessment dataset is integrated with the aerosol sample data, and the hazard assessment model is applied to perform hazard mapping for each aerosol point in the sample data. The model is used to map the hazard values in the assessment dataset to the corresponding aerosol sample data points, generating aerosol sample data with hazard level labels. Finally, a detailed hazard assessment report is generated based on the hazard assessment results. The report includes the hazard level, diffusion pattern, and recommended protective measures of aerosol samples in different areas.
[0127] 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.
[0128] 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 microbial aerosol diffusion hazard assessment method based on multidimensional data, characterized in that: The following steps are involved: Step S1: Acquire meteorological environment data; locate outlier intervals of the meteorological environment data, and interpolate outlier factors to generate filled meteorological environment data, wherein step S1 includes the following steps: Step S11: Acquire meteorological environment data; perform spatiotemporal interpolation processing on the meteorological environment data to generate a unified grid meteorological environment data set; Step S12: performing singular value analysis on the unified grid meteorological environment data set to obtain a principal component meteorological environment matrix; Step S13: locating the outlier interval of the main component meteorological environment matrix and generating an outlier interval index; Step S14: performing outlier data analysis on the principal component meteorological environment matrix based on the outlier interval index to obtain an outlier data subset; Step S15: performing tensor interpolation completion processing on the outlier data subset to generate completed outlier data; Step S16: interpolating outlier factors of the main component meteorological environment matrix according to the completed outlier data to generate filled meteorological environment data; Step S2: Acquire aerosol sample data; perform classification unit clustering processing on the aerosol sample data, and perform Fourier descriptor transformation to obtain an aerosol sample feature vector, wherein step S2 includes the following steps: Step S21: Acquire aerosol sample data; perform multi-scale wavelet decomposition processing on the aerosol sample data to obtain a multi-resolution characteristic coefficient set; Step S22: performing basic non-negative matrix decomposition processing on the multi-resolution feature coefficient set to generate a basic component matrix; Step S23: clustering the basic component matrix into taxonomic units to obtain a taxonomic unit clustering component matrix; Step S24: extracting shape contour features from the classification unit cluster component matrix to generate cluster unit contour point feature data; Step S25: performing discrete Fourier transform processing on the cluster unit contour point feature data to obtain a Fourier descriptor; Step S26: performing collaborative vector concatenation on the cluster unit contour point feature data and the Fourier descriptor to generate an aerosol sample feature vector; Step S3: positioning the diffusion direction of the aerosol sample characteristic vector based on the filled meteorological environment data to obtain aerosol diffusion direction data; estimating the time-delay diffusion velocity of the aerosol sample characteristic vector based on the filled meteorological environment data to generate aerosol diffusion velocity data, wherein step S3 includes the following steps: Step S31: performing wind field vector decomposition on the filled meteorological environment data to generate wind speed component and wind direction component; Step S32: performing a rotation coordinate transformation on the aerosol sample characteristic vector based on the wind speed component and the wind direction component to obtain a coordinate system aerosol characteristic vector; Step S33: performing main direction projection processing on the aerosol sample characteristic vector according to the coordinate system aerosol characteristic vector and the wind direction component to obtain aerosol diffusion direction data, wherein step S33 includes the following steps: Perform high-order tensor decomposition on the coordinate system aerosol eigenvector to obtain the coordinate system tensor principal component matrix; Perform principal component projection on the wind direction component based on the principal component matrix of the coordinate system tensor to obtain the wind direction projection vector in the principal component space; Perform multi-scale orthogonal decomposition on the principal component space wind direction projection vector to generate multi-directional data of aerosol diffusion; According to the multi-directional data of aerosol diffusion, the principal component matrix of the coordinate system tensor is weightedly reconstructed in each direction to generate the aerosol diffusion direction data; Step S34: based on the aerosol diffusion direction data and the wind speed component, the time-delayed diffusion velocity of the aerosol sample characteristic vector is estimated to generate aerosol diffusion velocity data, wherein step S34 includes the following steps: Perform delay embedding processing on aerosol diffusion direction data to obtain a time-delay embedding matrix; Based on the time-delay embedding matrix, the wind speed component is decomposed into a multi-scale time series to obtain a multi-scale wind speed component series; Similarity matching is performed on multi-scale wind speed component sequences to generate a time-delay diffusion velocity correlation matrix; According to the time-delay diffusion velocity correlation matrix, the aerosol sample characteristic vectors are weighted averaged in the main direction to obtain the aerosol diffusion velocity data; Step S4: performing multi-scale diffusion simulation on the aerosol sample data according to the aerosol diffusion direction data and the aerosol diffusion velocity data to obtain simulated diffusion aerosol data, wherein step S4 includes the following steps: Step S41: performing multi-dimensional Gaussian kernel density estimation on aerosol sample data according to aerosol diffusion direction data to obtain preliminary diffusion probability distribution data; Step S42: performing multi-scale decomposition on the preliminary diffusion probability distribution data to generate a multi-scale diffusion probability matrix; Step S43: performing time series convolution processing on the multi-scale diffusion probability matrix according to the aerosol diffusion velocity data to obtain a multi-scale diffusion simulation matrix; Step S44: performing reverse mapping processing on the multi-scale diffusion simulation matrix to obtain simulated diffusion aerosol data; Step S5: Perform spatiotemporal collaborative analysis on the simulated diffuse aerosol data and the meteorological environment data, and perform adaptive kernel density estimation to generate a multi-dimensional hazard heat map, wherein step S5 includes the following steps: Step S51: Perform spatiotemporal gridding on the simulated diffuse aerosol data to generate a global spatiotemporal grid data set: Step S52: Perform a bidirectional multivariate regression analysis on the global spatiotemporal grid data set and the meteorological environment data to obtain a spatiotemporal regression matrix, wherein step S52 includes the following steps: Step S521: performing feature space decomposition processing on the global spatiotemporal grid data set to obtain a feature variable matrix; Step S522: performing linear interpolation processing on the characteristic variable matrix according to the meteorological environment data to generate a time series interpolation matrix; Step S523: performing a bidirectional cross regression process on the time series interpolation matrix and the characteristic variable matrix to obtain a regression coefficient matrix; Step S524: performing spatial redistribution processing on the characteristic variables according to the regression coefficient matrix to obtain a spatiotemporal regression matrix; Step S53: performing unit local kernel density estimation on the spatiotemporal regression matrix to generate local hazard density data; Step S54: performing hazard density interpolation on the global spatiotemporal grid data set according to the local hazard density data to generate a multi-dimensional hazard heat map; Step S6: Based on the multi-dimensional hazard heat map and the simulated diffusion aerosol data, a diffusion hazard assessment is performed on the aerosol sample data to generate a hazard assessment report to perform a microbial aerosol diffusion hazard assessment based on multi-dimensional data, wherein step S6 includes the following steps: Step S61: performing sample diffusion mapping on the aerosol sample data based on the simulated diffusion aerosol data to obtain a regional diffusion trend map; Step S62: Perform diffusion point hazard assessment on the multi-dimensional hazard heat map and the regional diffusion trend map to obtain a diffusion point hazard assessment data set; Step S63: Perform hazard evaluation mapping on the aerosol sample data according to the diffusion point hazard assessment data set, and generate a hazard assessment report to perform microbial aerosol diffusion hazard assessment based on multidimensional data.
