PM2.5 data analysis method, device, medium and product based on bidirectional clustering
Through the bidirectional clustering method, the time and space dimension analysis of PM2.5 data was solved, and the inaccuracy caused by single-dimensional analysis in the existing technology was achieved, more efficient and accurate data analysis was realized, and the air pollution was revealed. Complex modes and changes.
Patent Information
- Application Number
- CN202410571458.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-10
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-05-10
AI Technical Summary
When analyzing PM2.5 data, the prior art analyzes only from a single dimension (space or time), resulting in inaccurate analysis results and prone to dimensional disaster problems.
Using a bidirectional clustering method, the PM2.5 data was analyzed in depth in two dimensions in time and space, and the objective function was determined through the KL divergence quantization method, and the cluster configuration information was determined through iterative convergence processing to characterize PM2.5 distribution.
It effectively reveals the complex distribution patterns and changes in air pollution, improves the efficiency and accuracy of data analysis, and realizes cluster analysis on multiple dimensions.
Smart Images

Figure CN118134121B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data analysis, and in particular to a PM2.5 data analysis method, device, medium and product based on bidirectional clustering. Background Art
[0002] With the rapid development of global industrialization and urbanization, air pollution has become one of the major environmental problems facing the world. Among them, the pollution of fine particulate matter PM2.5 has attracted widespread attention. Analyzing the dynamic changes of PM2.5 concentration in time and space and deeply understanding the spatiotemporal distribution characteristics of air pollutants are of great significance for public health protection and environmental management.
[0003] As a method of spatiotemporal data mining, cluster analysis reveals pollution patterns by grouping similar elements, which is extremely important for understanding the spatial distribution and temporal changes of PM2.5 concentrations. This method not only helps to identify pollution hotspots and analyze possible pollution sources, but can also be combined with other methods to predict pollution trends.
[0004] At present, the research on PM2.5 analysis using clustering methods mainly focuses on analyzing and predicting PM2.5 concentration data using clustering methods. For example, Chinese patent CN113223392A discloses a hybrid integration model for PM2.5 hourly concentration prediction, in which the method includes clustering components with similar characteristics in PM2.5 data using FCM clustering. Although this method has high prediction accuracy, low calculation level and strong adaptability, it can only analyze data from a single dimension, which is prone to dimensional disaster problems.
[0005] Another example is a clustering-based integrated neural network PM2.5 prediction method published in Chinese patent CN111340292B. By establishing a clustering method based on wind direction, the influence of wind direction on PM2.5 is considered to improve the prediction accuracy. The method includes: dividing the wind direction data into 5 categories, namely, calm, northeast wind, southeast wind, southwest wind and northwest wind; using hot encoding to convert each wind direction into a 4-dimensional binary vector; analyzing the influence of PM2.5 and wind direction through the Pearson correlation coefficient; clustering analysis of PM2.5 based on wind direction, and improving the prediction accuracy by considering the influence of wind direction on PM2.5. Although this method has low cost and high prediction accuracy, it only considers the analysis of a single dimension of space or time, resulting in inaccurate analysis results. Summary of the invention
[0006] The purpose of the present invention is to provide a PM2.5 data analysis method, device, medium and product based on bidirectional clustering, which can realize accurate analysis of PM2.5 data.
[0007] To achieve the above object, the present invention provides the following solutions:
[0008] A PM2.5 data analysis method based on bidirectional clustering, the method comprising:
[0009] Acquire information data and determine a data matrix; the information data includes: PM2.5 concentration values corresponding to different time points in multiple sites; the data matrix is a matrix organized according to the information data;
[0010] Based on the bidirectional clustering method, bidirectional clustering initialization processing is performed according to the data matrix to obtain a clustering data matrix;
[0011] The objective function is determined by using a KL divergence quantization method; the objective function is determined based on the data matrix and the clustering data matrix;
[0012] According to the set clustering information, the objective function is iterated and converged until a set stop condition is met; the set stop condition is that the change of the objective function is less than a preset threshold;
[0013] According to the objective function after iterative convergence processing, cluster configuration information is determined; the cluster configuration information is used to characterize the distribution of PM2.5.
[0014] Optionally, based on a two-way clustering method, a two-way clustering initialization process is performed according to the data matrix to obtain a clustering data matrix, specifically including:
[0015] According to the data matrix, random allocation is performed according to the set number of site clusters to obtain site cluster data;
[0016] According to the data matrix, random allocation is performed according to the number of clusters at set time points to obtain time point cluster data;
[0017] A clustering data matrix is determined according to the site clustering data and the time point clustering data.
[0018] Optionally, the objective function is expressed as:
[0019] ;
[0020] in, is the objective function; is information divergence; is the data matrix; is the clustering data matrix; is the mutual information between sites and time points; is the mutual information between site clusters and time point clusters.
[0021] Optionally, the setting of clustering division information includes: location clustering division and time point clustering division.
[0022] Optionally, the expression for dividing the locations into clusters is:
[0023] ;
[0024] in, The ordinal number for clustering the locations; To set the number of site clusters; is the data matrix With the clustering data matrix The information divergence between .
[0025] Optionally, the expression for dividing the time points into clusters is:
[0026] ;
[0027] in, The sequence number for clustering the time points; The number of clusters at a set time point.
[0028] A computer device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the PM2.5 data analysis method based on bidirectional clustering as described above.
[0029] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the PM2.5 data analysis method based on bidirectional clustering as described above.
[0030] A computer program product includes a computer program, which, when executed by a processor, implements the PM2.5 data analysis method based on bidirectional clustering as described above.
[0031] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0032] The present invention discloses a PM2.5 data analysis method, device, medium and product based on bidirectional clustering, which obtain information data and determine a data matrix; based on the bidirectional clustering method, bidirectional clustering initialization processing is performed according to the data matrix to obtain a clustering data matrix; a KL divergence quantification method is used to determine an objective function; according to set clustering information, an iterative convergence processing is performed on the objective function until a set stop condition is met; according to the objective function after iterative convergence processing, clustering configuration information is determined to characterize the distribution of PM2.5; the present invention effectively reveals the complex distribution patterns and change laws in air pollution through in-depth analysis in two dimensions of time and space, thereby improving the efficiency and accuracy of data analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0034] Figure 1 A schematic flow chart of a PM2.5 data analysis method based on bidirectional clustering provided in an embodiment of the present invention.
[0035] Figure 2 This is a flow chart of the PM2.5 data analysis method based on the two-way clustering method.
[0036] Figure 3 Schematic diagram of the PM2.5 data analysis device based on the two-way clustering method.
[0037] Figure 4 Schematic diagram of the distribution of optimal location clusters and time point clusters. DETAILED DESCRIPTION
[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0039] The existing PM2.5 concentration clustering analysis method only considers a single dimension of space or time. There is an urgent need for a PM2.5 data analysis method based on a two-way clustering method to solve the problem of clustering analysis in multiple dimensions and reveal the complex patterns in PM2.5 data.
[0040] The purpose of the present invention is to provide a PM2.5 data analysis method, device, medium and product based on bidirectional clustering, aiming to achieve accurate analysis of PM2.5 data.
[0041] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.
[0042] Example 1
[0043] like Figure 1 As shown, a PM2.5 data analysis method based on bidirectional clustering in this embodiment includes:
[0044] Step 100: Obtain information data and determine a data matrix. The information data includes: PM2.5 concentration values corresponding to different time points in multiple sites; the data matrix is a matrix organized according to the information data.
[0045] Step 200: Based on the bidirectional clustering method, bidirectional clustering initialization processing is performed according to the data matrix to obtain a clustering data matrix.
[0046] Step 300: Determine the objective function using the KL divergence quantization method. The objective function is determined based on the data matrix and the clustering data matrix.
[0047] Specifically, the expression of the objective function is:
[0048] .
[0049] in, is the objective function; is information divergence; is the data matrix; is the clustering data matrix; is the mutual information between sites and time points; is the mutual information between site clusters and time point clusters.
[0050] Step 400: According to the set clustering information, the objective function is iterated and converged until the set stop condition is met. The set stop condition is that the change amount of the objective function is less than a preset threshold.
[0051] The setting of clustering information includes: location clustering and time point clustering.
[0052] Specifically, the expression for location partition clustering is:
[0053] .
[0054] in, The ordinal number for clustering the locations; To set the number of site clusters; is the data matrix With the clustering data matrix The information divergence between .
[0055] The expression for time point partitioning clusters is:
[0056] .
[0057] in, The sequence number for clustering the time points; The number of clusters at a set time point.
[0058] Step 500: Determine cluster configuration information according to the target function after iterative convergence processing. The cluster configuration information is used to characterize the distribution of PM2.5.
[0059] Among them, based on the two-way clustering method, two-way clustering initialization processing is performed according to the data matrix to obtain the clustering data matrix, which specifically includes:
[0060] According to the data matrix, random allocation is performed according to the set number of site clusters to obtain site cluster data.
[0061] According to the data matrix, random allocation is performed according to the set number of clusters at the time point to obtain the time point cluster data.
[0062] The clustering data matrix is determined based on the site clustering data and the time point clustering data.
[0063] like Figure 2 As shown, in practical application, the method provided by the present invention includes the following steps.
[0064] Step S1: Data organization. Organize PM2.5 in a data matrix middle.
[0065] Step S2: Bidirectional clustering initialization. Initialize the sites and time points in the data into site clusters and Clusters at each time point.
[0066] Step S3: constructing the objective function. The objective function is established according to the initialization result and the information loss before and after clustering.
[0067] (1).
[0068] Step S4: Divide the locations into new location clusters.
[0069] (2).
[0070] Step S5: Divide the time points into new time point clusters.
[0071] (3).
[0072] Step S6: Recalculate the objective function formula (1) using the new location clusters and time point clusters. Repeat steps S4 and S5, and the loop ends when the objective function converges to the minimum value.
[0073] The data matrix in step S1 , the rows in the matrix represent sites, the columns represent time points, and the elements represent PM2.5 concentration values. The expression is:
[0074] (4).
[0075] in For the Line and PM2.5 concentration values in the column. and Represents the site and time point respectively. and Represent the number of sites and time points, respectively.
[0076] In step S2, the data matrix The method for initializing the bidirectional clustering of the sites and time points in the is random initialization. First, the number of site clusters to be divided is determined according to relevant research. and time point clustering , randomly Sites are assigned to site clusters and Time points are allocated to Each site is marked as belonging to a specific site cluster (for example, site 1 belongs to site cluster 1), and each time point is also marked as belonging to a specific time point cluster (for example, January belongs to time point cluster 1), so that the data matrix after initialization of bidirectional clustering can be obtained. ,in Indicates site clustering, Represents time point clustering.
[0077] In step S3, the objective function (1) is calculated by using the Kullback-Leibler (KL) divergence to measure the information loss before and after bidirectional clustering. The probability distribution of each site and time point is compared with the probability distribution after clustering. The KL divergence constructs the objective function by quantifying the amount of information lost during the clustering process. After the objective function is constructed, the information loss before and after clustering is reduced by cyclically dividing the sites into new site clusters and the time points into new time point clusters, so that the objective function value is minimized.
[0078] In step S4, the locations are divided into new location clusters according to calculation (2) by adjusting the cluster allocation of the sites to minimize the overall KL divergence, that is, to reduce the information loss before and after clustering. By comparing the KL divergence values under different site cluster allocation schemes, the site cluster allocation that can provide the minimum information loss is identified.
[0079] In step S5, the time points are divided into new time point clusters according to calculation (3) to minimize the overall KL divergence by adjusting the cluster allocation of the time points, that is, to reduce the information loss before and after clustering. By comparing the KL divergence values under different time point cluster allocation schemes, the time point cluster allocation that can provide the minimum information loss is identified.
[0080] In step S6, the iterative optimization process aims to minimize the information loss between the original data and the clustered data. The clustering configuration that minimizes the objective function value is found by continuously iteratively adjusting the clustering allocation of sites and time points. The criterion for judging whether the iteration is over is the change in the objective function value: if this change is less than a preset threshold in consecutive iterations, it is considered that the objective function has converged, and the iterative process ends.
[0081] Specifically, the detailed design of the PM2.5 data analysis method based on the two-way clustering method includes the following steps:
[0082] Step S1: Data organization. Organize PM2.5 in a data matrix In , the behavioral sites are listed as time points.
[0083] The hourly PM2.5 concentration data set of 18 stations in a certain area from February 1, 2013 to January 31, 2014 was obtained. 299 days were selected as the study period, and the criterion was to remove the days with zero PM2.5 concentration at all stations within 24 hours. The daily average PM2.5 concentration data were obtained and organized in a data matrix of size 18×299.
[0084] Step S2: Two-way clustering initialization. Randomly divide the sites and 299 days in the data into site clusters and Clusters at each time point.
[0085] Furthermore, based on existing research, three site clusters and four day clusters are set. 2.5 The 18 sites and 299 days in the concentration dataset are randomly divided into 3 site clusters and 4 day clusters. The site cluster initialization results are shown in Table 1. Due to limited space, the day cluster initialization results are not shown here.
[0086]
[0087] Step S3: Constructing the objective function: According to the initialization result, the objective function is established according to the information loss before and after clustering.
[0088] Furthermore, the objective function is established based on the information loss of the data matrix before and after the bidirectional clustering according to the initialized site clusters and day clusters. The objective function is defined by quantifying the information divergence between the original data matrix and the data matrix after bidirectional clustering initialization, aiming to minimize the information loss before and after clustering.
[0089] BBAC_I uses the Kullback-Leibler (KL) divergence, also known as relative entropy, to calculate the information divergence between the original data matrix and the two-way clustered data matrix. The KL divergence quantifies the amount of information lost in the clustering process by comparing the probability distribution of each site and day with the probability distribution after clustering. The objective function seeks the best clustering result by minimizing this information loss (i.e., minimizing the KL divergence), thereby ensuring that the two-way clustered data matrix is as close as possible to the original data matrix in terms of information content.
[0090] Step S4: Divide the 18 sites into new site clusters.
[0091] Furthermore, site clusters are set for all sites according to the objective function, and new site clusters are selected to minimize the objective function.
[0092] The cluster assignment of sites is adjusted to minimize the overall KL divergence, i.e., to reduce the information loss before and after clustering. This process involves trying different cluster assignments for each site and calculating the impact of these potential assignments on the overall objective function. The objective function is defined here as a global information loss measure, i.e., the sum of the KL divergences for all sites and all site cluster configurations. The algorithm identifies the site cluster assignment that provides the minimum information loss by comparing the objective function values under different site cluster assignment schemes.
[0093] Step S5: Divide the 299 days into new day clusters.
[0094] Furthermore, day clusters are set for all days according to the objective function, and new day clusters are selected to minimize the objective function.
[0095] The cluster assignments of the days are adjusted to minimize the overall KL divergence, i.e., to reduce the information loss before and after clustering. This process involves trying different cluster assignments for each day and calculating the impact of these potential assignments on the overall objective function. The objective function is defined here as a global information loss measure, i.e., the sum of the KL divergences for all days and all day cluster configurations. The algorithm compares the objective function values under different day cluster assignments to identify the day cluster assignment that provides the minimum information loss.
[0096] Step S6: Recalculate the objective function formula (1) using the new site clusters and day clusters. Repeat steps S4 and S5, and the loop ends when the objective function converges to the minimum value.
[0097] Furthermore, the objective function is recalculated using the new site clusters and day clusters and an iterative cycle is started to minimize the objective function to obtain the optimal site clusters and day clusters.
[0098] Using the new site clusters and day clusters, recalculate the KL divergence between the original data matrix and the bidirectional clustered data matrix generated based on the new cluster assignments. This process involves converting the original data matrix and the bidirectional clustered data matrix generated based on the new cluster assignments into probability distribution forms, and then applying the KL divergence formula to quantify the difference between the two distributions, reflecting the degree of information preservation in the clustering process.
[0099] After recalculating the objective function based on the new site clusters and day clusters, the algorithm enters an iterative optimization loop, including step S4 (dividing the sites into new site clusters) and step S5 (dividing the 299 days into new day clusters). In this iterative process, the bidirectional clustering algorithm further reduces the value of the objective function by adjusting the clustering of sites and days. In each iteration, the impact of new site clusters and day clusters on the objective function is evaluated, and site clusters and day clusters that minimize information loss are selected.
[0100] When the value of the objective function converges to a minimum value, or the improvement is lower than a predetermined threshold, the iterative optimization process stops. The convergence judgment in BBAC_I is based on the change of the objective function value in continuous iterations. If the change is less than a certain set threshold, the objective function is considered to have converged, and the site clusters and day clusters at this time are considered to be the optimal two-way clustering results. The convergence judgment criterion ensures that the two-way clustering algorithm can stop after achieving full optimization, ensuring the stability and reliability of the two-way clustering results. The final output site clusters are shown in Table 2.
[0101]
[0102] The visualization results of the optimal location clustering and sky clustering are as follows: Figure 4 shown.
[0103] The method proposed in the present invention effectively integrates time and space dimension analysis, and through two-way clustering analysis of PM2.5 data, reveals the complex distribution patterns in air pollution and enhances the understanding of the changing laws of PM2.5.
[0104] In addition, in practical applications, a PM2.5 data analysis device based on a two-way clustering method is also provided, such as Figure 3 As shown, the PM2.5 data analysis device based on the two-way clustering method includes an acquisition unit, a first calculation unit, a second calculation unit, a third calculation unit, a fourth calculation unit, a fifth calculation unit and a sixth calculation unit.
[0105] The acquisition unit is used to acquire the existing data source in step S1; that is, the acquisition unit is used to acquire the existing data source, and the acquired existing data source is used in step S1; specifically, it is used to acquire the PM2.5 concentration data set to be processed. The first calculation unit is used for step S1 data to organize the existing data source. Specifically, the first calculation unit is used to organize the data acquired by the acquisition unit into a two-dimensional data matrix.
[0106] The second computing unit is used for step S2 bidirectional clustering initialization, which initializes the site clusters and time point clusters based on the data organized in step S1. That is, the data matrix organized in step S1 is divided into Location clusters and time point clusters; specifically, it is used to establish initial bidirectional clustering, randomly dividing the locations in the data into The time points in the data are randomly divided into Clusters at each time point.
[0107] The third calculation unit is used for step S3 objective function construction, and establishes the objective function; that is, based on the bidirectional clusters randomly divided in step S2, the objective function is established according to the information loss before and after the random division. The fourth calculation unit is used for step S4 to divide the locations into new location clusters; specifically, the fourth calculation unit maps the locations to the new location clusters based on the objective function constructed in step S3, so that the mutual information loss of the objective function is minimized.
[0108] The fifth computing unit is used to divide the time points into new time point clusters in step S5; specifically, based on the objective function constructed in step S3, the time points are mapped to new location clusters so that the mutual information loss of the objective function is minimized.
[0109] The sixth calculation unit is used to determine whether the step S6 is converged. Specifically, based on the location clusters and time point clusters updated in steps S4 and S5, the objective function is recalculated and judged. If the objective function converges, the algorithm ends. Otherwise, steps S4 and S5 are looped.
[0110] The advantages and effects of the present invention are as follows:
[0111] 1. The method provided by the present invention reveals the complex spatiotemporal patterns of PM2.5 data centers by clustering PM2.5 data from both spatial and temporal dimensions.
[0112] 2. The device provided by the present invention includes computing units such as data organization, bidirectional clustering initialization, target function construction, new location clustering, new time point clustering and convergence judgment.
[0113] Through bidirectional cluster analysis of PM2.5 data, the time and space dimension analysis are effectively integrated to reveal the complex distribution patterns in air pollution, enhance the understanding of the changing laws of PM2.5, optimize data processing, and improve analysis efficiency and accuracy.
[0114] Example 2
[0115] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the PM2.5 data analysis method based on bidirectional clustering in Example 1.
[0116] Example 3
[0117] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the PM2.5 data analysis method based on bidirectional clustering in Example 1.
[0118] Example 4
[0119] A computer program product includes a computer program, which, when executed by a processor, implements the PM2.5 data analysis method based on bidirectional clustering in Example 1.
[0120] Example 5
[0121] A computer device, which may be a database. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store pending transactions. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the PM2.5 data analysis method based on bidirectional clustering in Example 1 is implemented.
[0122] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0123] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided by the present invention may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided by the present invention may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.
[0124] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0125] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of the present invention. At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.
Claims
1. A PM2.5 data analysis method based on bidirectional clustering, characterized in that: The method comprises: Acquire information data and determine a data matrix; the information data includes: PM2.5 concentration values corresponding to different time points at multiple sites; the data matrix is a matrix organized according to the information data; organize PM2.5 in a data matrix O(S, T), where rows are sites and columns are time points; Based on the bidirectional clustering method, bidirectional clustering initialization processing is performed according to the data matrix to obtain a clustering data matrix; The objective function is determined by using a KL divergence quantization method; the objective function is determined based on the data matrix and the clustering data matrix; According to the set clustering information, the objective function is iterated and converged until a set stop condition is met; the set stop condition is that the change of the objective function is less than a preset threshold; Determine cluster configuration information according to the objective function after iterative convergence processing; the cluster configuration information is used to characterize the distribution of PM2.5; The expression of the objective function is: Where y is the objective function; D I (·||·) is the information divergence; O(S, T) is the data matrix; is the clustering data matrix; I(S, T) is the mutual information between sites and time points; is the mutual information between site clusters and time point clusters; Visualize based on cluster configuration information; The setting of clustering information includes: location clustering and time point clustering; The expression for dividing the locations into clusters is: Among them, i is the sequence number of the location cluster; k is the number of set site clusters; is the data matrix O(S, T) and the clustering data matrix The information divergence between The expression for dividing the clusters at the time point is: Among them, j is the sequence number of the time point clustering; l is the number of clusters at the set time point.
2. The PM2.5 data analysis method based on bidirectional clustering according to claim 1, characterized in that: Based on the bidirectional clustering method, bidirectional clustering initialization processing is performed according to the data matrix to obtain a clustering data matrix, which specifically includes: According to the data matrix, random allocation is performed according to the set number of site clusters to obtain site cluster data; According to the data matrix, random allocation is performed according to the number of clusters at set time points to obtain time point cluster data; A clustering data matrix is determined according to the site clustering data and the time point clustering data.
3. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the PM2.5 data analysis method based on bidirectional clustering as described in any one of claims 1 to 2.
4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the PM2.5 data analysis method based on bidirectional clustering described in any one of claims 1 to 2 is implemented.
5. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the PM2.5 data analysis method based on bidirectional clustering described in any one of claims 1 to 2 is implemented.
