PM2.5 concentration prediction method and system based on long time-space sequence
By combining time and spatial characteristics, the long-time spatiotemporal sequence prediction method is used to utilize the integrated learning of Informer-GCN, LSTM-GCN and XGBoost models to solve the problem that the impact of surrounding sites is not considered in the existing PM2.5 concentration prediction, achieving higher accuracy and efficiency prediction.
Patent Information
- Application Number
- CN202510555752.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-29
- Publication Date
- 2025-08-08
AI Technical Summary
The existing PM2.5 concentration prediction method fails to effectively consider the impact of the sites around the target site, resulting in low accuracy and efficiency of prediction results.
The prediction method based on long-time and space sequences is adopted, and the temporal and spatial characteristics of the target site are obtained, combined with Informer-GCN, LSTM-GCN and XGBoost models for integrated learning, and the concentration prediction model is trained using Stacking strategy to consider the fusion of time and spatial characteristics.
It improves the accuracy and efficiency of PM2.5 concentration prediction, and can provide timely and accurately support for heavy pollution weather warning data.
Smart Images
Figure CN120450136A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of atmospheric pollutant concentration prediction, and specifically relates to a PM2.5 concentration prediction method and system based on long-term space-time sequences. Background Art
[0002] With the rapid development of industrialization and urbanization in my country, air pollution has become a critical issue that cannot be ignored. As one of the most destructive particulate matter (PM) in the atmosphere, PM2.5 (PM2.5) is a major contributor to smog and reduced visibility, negatively impacting the climate system and human health. PM2.5 refers to inhalable particles in the atmosphere with an aerodynamic equivalent of 2.5 μm or less. They easily absorb harmful chemicals such as polycyclic aromatic hydrocarbons and mercury, reducing visibility and attaching to the respiratory tract and lungs, leading to serious illnesses such as respiratory problems, lung cancer, and cardiovascular disease. Therefore, timely and accurate prediction and early warning of PM2.5 concentrations are crucial for air pollution control and for people's daily lives and travel.
[0003] Because the formation and diffusion of PM2.5 are influenced by multiple complex factors, including climate, geography, and urban development, accurately predicting PM2.5 concentrations has become a challenging, nonlinear task. Currently, there are three broad approaches to PM2.5 concentration prediction, both domestically and internationally: numerical prediction methods based on atmospheric kinematics, traditional statistical models, and machine learning models. Numerical prediction methods rely on atmospheric diffusion theory and physical and chemical processes to develop mathematical and physical models to predict PM2.5 concentrations. Chuang et al. used the Weather Research and Forecasting Model (WRF-Chem-MADRID) with chemical-integrated methods to predict air quality in the southeastern United States. The results showed that the model had high accuracy for ozone and inhalable particulate matter. Although numerical prediction methods take into account the physical and chemical processes underlying the diffusion of atmospheric pollutant concentrations, their accuracy relies heavily on emission inventories of pollutant sources and is computationally intensive.
[0004] To address the lack of precision associated with long-term forecasts, the application of time series algorithms to PM2.5 concentration prediction has been considered, including BP neural networks, recurrent neural networks, deep belief networks, and long-short-term memory networks. Li et al. constructed a hybrid PM2.5 prediction model combining a convolutional neural network (CNN) with an attention mechanism and a long-short-term memory network (LSTM), addressing the issue of nonlinear time series prediction and demonstrating good performance. To address the low prediction accuracy of existing PM2.5 concentration time series models, Dong Hao et al. established a Seq2Seq, single-site, multi-step PM2.5 time series prediction model based on Informer. Using historical pollutant data and meteorological data as input, this model predicts PM2.5 concentrations for a period of time in the future.
[0005] However, the above methods only regard the PM2.5 concentration prediction problem as a single time series prediction, rarely considering the impact of PM2.5 concentrations at surrounding sites on the PM2.5 concentration level at the target site, and fail to consider the impact of spatiotemporal prediction in PM2.5 concentration prediction, resulting in low precision and low efficiency in the prediction results. Summary of the Invention
[0006] In order to solve the problem of low PM2.5 concentration prediction accuracy and efficiency of existing methods, the present invention provides a PM2.5 concentration prediction method and system based on long-term space-time series.
[0007] In order to achieve the above object, the present invention provides the following technical solutions:
[0008] A PM2.5 concentration prediction method based on a long spatiotemporal sequence comprises the following steps:
[0009] Acquire atmospheric pollutant concentration data and meteorological data for a continuous time period at a target site, and obtain a time feature sequence based on the atmospheric pollutant concentration data and meteorological data for the continuous time period at the target site; filter out feature columns related to PM2.5 concentration in the time feature sequence; obtain location distribution data and historical PM2.5 concentration data for the target site and its surrounding sites, input the site distribution data and site feature data into a graph convolutional network (GCN) model to obtain spatial features; fuse the filtered feature columns with the spatial features to output spatiotemporal series data;
[0010] The Informer-GCN model and the LSTM-GCN model are used as base learners, and the XGBoost model including the attention mechanism is used as the second-layer learner. The base learners and the second-layer learners are integrated through the Stacking integration strategy. The integrated model is trained according to the spatiotemporal series data to obtain a concentration prediction model;
[0011] The site distribution data and site characteristic data of the site to be predicted and its surrounding sites are used to obtain the spatial characteristics to be predicted, and the spatial characteristics to be predicted and the historical PM2.5 concentration data of the site to be predicted are fused to obtain the spatiotemporal characteristics to be predicted. The spatiotemporal characteristics to be predicted are input into the concentration prediction model to output the predicted PM2.5 concentration value within a preset time period in the future.
[0012] Preferably, before screening out feature columns that are highly correlated with PM2.5 concentration, the method also includes using the Z-score method to detect outliers, replacing the detected outliers with linear interpolation, and performing minimum-maximum normalization on each feature column and label column.
[0013] Preferably, feature columns related to PM2.5 concentration are screened out from the time feature sequence, specifically by using random forest and Pearson correlation coefficient for screening.
[0014] Preferably, it also includes using the root mean square error RMSE, mean absolute error MAE and correlation coefficient R 2 The accuracy of the concentration prediction model was evaluated.
[0015] Preferably, the time feature sequence specifically includes a timestamp column, a feature column, and a label column.
[0016] The present invention also provides a PM2.5 concentration prediction system based on long-term spatiotemporal sequences, specifically comprising:
[0017] The feature acquisition module is used to obtain the atmospheric pollutant concentration data and meteorological data of the target site for a continuous time period, and obtain a time feature sequence based on the atmospheric pollutant concentration data and meteorological data of the target site for a continuous time period; filter out feature columns related to PM2.5 concentration in the time feature sequence; obtain the location distribution data and historical PM2.5 concentration data of the target site and its surrounding sites, input the site distribution data and site feature data into the graph convolutional network (GCN) model to obtain spatial features; fuse the filtered feature columns with the spatial features to output spatiotemporal series data.
[0018] The model construction module is used to use the Informer-GCN model and the LSTM-GCN model as base learners, the XGBoost model containing the attention mechanism as the second-layer learner, integrate the base learners and the second-layer learner through the Stacking integration strategy, and train the integrated model according to the spatiotemporal series data to obtain a concentration prediction model.
[0019] The prediction module is used to use the site distribution data and site feature data of the site to be predicted and its surrounding sites to obtain the spatial features to be predicted, fuse the spatial features to be predicted with the historical PM2.5 concentration data of the site to be predicted to obtain the spatiotemporal features to be predicted, input the spatiotemporal features to be predicted into the concentration prediction model, and output the predicted PM2.5 concentration value within a preset time period in the future.
[0020] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps described in the PM2.5 concentration prediction method based on a long spatiotemporal sequence.
[0021] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is loaded by a processor, it can execute the steps described in the PM2.5 concentration prediction method based on a long spatiotemporal sequence.
[0022] The PM2.5 concentration prediction method based on long-term spatial and temporal sequences provided by the present invention has the following beneficial effects:
[0023] The present invention obtains a time feature column by acquiring atmospheric pollutant concentration data and meteorological data arranged in sequence within a continuous time period; obtains spatial features based on the distribution data and feature data of each monitoring station in the target area, and fuses the time feature column with the spatial feature to obtain a long spatiotemporal sequence, taking both time and space features into account. The Stacking integration strategy is used to combine different types of learners, and the integrated model is trained using long spatiotemporal sequence data to obtain a concentration prediction model. Each learner complements its own strengths and weaknesses, thereby improving the accuracy of the model's prediction and its problem-solving capabilities. The concentration prediction model can obtain a predicted value of PM2.5 concentration within a preset time, with higher prediction accuracy, which facilitates timely and accurate data support for forecasting heavy pollution weather. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] To more clearly illustrate the embodiments of the present invention and its design, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.
[0025] Figure 1 This is a flow chart of a PM2.5 concentration prediction method based on long-term space-time series in the present invention.
[0026] Figure 2 This is a distribution diagram of the areas where each monitoring station is located and the PM2.5 concentration in an embodiment of the present invention.
[0027] Figure 3 This is a flow chart of model predictor training in an embodiment of the present invention.
[0028] Figure 4 This is a comparison chart of the predicted value and the actual value of PM2.5 concentration in the first hour in an embodiment of the present invention.
[0029] Figure 5 This is a comparison chart of the predicted value and the actual value of PM2.5 concentration in the second hour in an embodiment of the present invention.
[0030] Figure 6 This is a comparison chart of the predicted value and the actual value of PM2.5 concentration in the third hour in an embodiment of the present invention. DETAILED DESCRIPTION
[0031] In order to enable those skilled in the art to better understand the technical solution of the present invention and to be able to implement it, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention.
[0032] Example
[0033] The present invention provides a PM2.5 concentration prediction method based on long-term spatial and temporal sequences, taking City A as the main research area, such as Figure 1 As shown, the specific steps include:
[0034] S1. Obtain atmospheric pollutant concentration data and meteorological data arranged in sequence within a continuous time period. Collect air pollution data and meteorological data from 12 national air quality stations in 8 districts of City A from January 2021 to December 2023, as well as PM2.5, PM10, NO2, SO2, CO, O3 and other air pollutant concentration data and meteorological data such as temperature, humidity, wind direction, wind force and precipitation. Assuming that the PM2.5 concentration data for 12 consecutive hours before time t is used to predict the PM2.5 concentration in the next 3 hours, the construction process of the time input data is divided into the following steps.
[0035] (1) The atmospheric pollutant concentration data and meteorological data are arranged in chronological order. The data has 12 columns, corresponding to the timestamp column, feature column, and label column.
[0036] (2) Use linear interpolation to fill missing values, adopt the Z-score method to detect outliers, replace the detected outliers with linear interpolation, and perform minimum-maximum normalization on each column, that is, normalize each feature column and label column.
[0037] (3) The feature importance score and Pearson correlation coefficient of random forest are comprehensively used to screen out feature columns with high correlation with PM2.5 concentration.
[0038] (4) The PM2.5 concentration data column is set as the label column as the variable for the Informer-GCN model and LSTM-GCN model to predict data.
[0039] (5) Establish a sliding window to process PM2.5 concentration and selected feature sequences. The window size is 12 and the step size is 3, and it is placed on top of the data.
[0040] (6) Extract the data in the sliding window and append the first 12 items to the variable Y column.
[0041] (7) If the window slides to the bottom of the data, the program is terminated and the overall information of the time series obtained based on multiple time steps is output; otherwise, the sliding window slides to the next hour and returns to step (5).
[0042] S2. Obtain the distribution data and characteristic data of each monitoring station in the target area. The area where each monitoring station is located and the PM2.5 concentration distribution are as follows: Figure 2 The information of each monitoring station is shown in Table 1. Among them, the air pollutant data comes from the Environmental Protection and Testing Center of City A, and the meteorological data comes from the National Climate Data Center (NCDC) of the United States.
[0043] Table 1 Information of monitoring sites
[0044]
[0045] The GCN model is used to extract the spatial distribution trends of the 12 monitoring stations and learn the spatial relationships between PM2.5 concentrations. The distribution and feature data of the monitoring stations are input into the graph convolutional network (GCN) model, which outputs spatial features. The GCN model inputs include a station distribution map and station feature data. The geographic distance between stations can be calculated using their latitude and longitude data. The station feature data is represented by a feature matrix X, which consists of historical PM2.5 concentration data for each station.
[0046] S3. Fuse the filtered feature columns with the spatial features and output spatiotemporal sequence data.
[0047] S4. The Informer-GCN model (Informer-graph convolutional network) and LSTM-GCN (Long Short-Term Memory-graph convolutional network) model are used as base learners, and the XGBoost (Extreme Gradient Boosting Tree) model with attention mechanism is used as the second-layer learner. The base learner and the second-layer learner are integrated through the Stacking integration strategy, and the integrated model is trained according to the spatiotemporal sequence data, such as Figure 3 As shown in the figure, the concentration prediction model, namely the Informer-LSTM-GCN model, is obtained. In addition to implementing the Informer-LSTM-GCN model through the stacking integration strategy, the two strong learners are separated to form two independent Informer-GCN and LSTM-GCN models, which serve as comparison models for evaluating the Informer-LSTM-GCN model.
[0048] S5. Input the historical PM2.5 concentration data, site distribution data, and site characteristic data of the site to be predicted and its surrounding sites into the concentration prediction model, and output the predicted PM2.5 concentration value within a preset time period in the future.
[0049] The hourly air pollutant and meteorological data from the Guanyuan station in City A from January 1, 2021 to December 31, 2023, as well as PM2.5 concentration data from surrounding stations, were used to conduct a comparative experiment between the base learner Informer-GCN model and the LSTM-GCN model. The results show that for the spatiotemporal prediction of PM2.5 concentration in the next 3 hours, the integrated model Informer-LSTM-GCN has the smallest RMSE and MAE in the first, second, and third hours, and R 2 The prediction results show that the integrated Informer-LSTM-GCN model can accurately predict PM2.5 concentration, and the prediction accuracy is improved compared with the single learner Informer-GCN and LSTM-GCN. As the prediction time increases, the prediction effect of the three learners decreases, showing different "prediction hysteresis" phenomena. However, the evaluation indicators of the Informer-LSTM-GCN model are still the best. Figure 4 、 Figure 5 and Figure 6 As shown in the figure, the fitting results analysis of the predicted values and the true values corresponding to the 1st hour, 2nd hour and 3rd hour respectively.
[0050] Compared with the Informer-GCN model and the LSTM-GCN model, the Informer-LSTM-GCN model has the smallest RMSE and MAE in predicting the PM2.5 concentration in the next 3 hours. 2 The RMSE of the Informer-LSTM-GCN model for the prediction of PM2.5 concentration in the first hour was reduced by 14.4% compared with the Informer-GCN model and by 22.6% compared with the LSTM-GCN model; 2 Compared with the Informer-GCN model, it increased by 1.4%, and compared with the LSTM-GCN model, it increased by 2.7%. For the prediction of PM2.5 concentration in the second hour, the RMSE of the Informer-LSTM-GCN model decreased by 5.4% compared with the Informer-GCN model, and decreased by 14.6% compared with the LSTM-GCN model; 2 Compared with the Informer-GCN model, it increased by 1.2% and compared with the LSTM-GCN model, it increased by 4.0%. For the prediction of PM2.5 concentration at the 3rd hour, the RMSE of the Informer-LSTM-GCN model decreased by 3.3% and compared with the LSTM-GCN model, it decreased by 13.0%. 2This is a 1.2% increase compared to the Informer-GCN model and a 6.2% increase compared to the LSTM-GCN model, as shown in Table 2.
[0051] Table 2 Comparison of model evaluation indicators
[0052]
[0053]
[0054] The present invention also provides a PM2.5 concentration prediction system based on long-term spatiotemporal sequences, specifically comprising:
[0055] The feature acquisition module is used to obtain the atmospheric pollutant concentration data and meteorological data of the target site for a continuous time period, and obtain a time feature sequence based on the atmospheric pollutant concentration data and meteorological data of the target site for a continuous time period; filter out feature columns related to PM2.5 concentration in the time feature sequence; obtain the location distribution data and historical PM2.5 concentration data of the target site and its surrounding sites, input the site distribution data and site feature data into the graph convolutional network (GCN) model to obtain spatial features; fuse the filtered feature columns with the spatial features to output spatiotemporal series data.
[0056] The model building module is used to use the Informer-GCN model and LSTM-GCN model as base learners, the XGBoost model with attention mechanism as the second-layer learner, integrate the base learners and the second-layer learner through the Stacking integration strategy, and train the integrated model based on spatiotemporal sequence data to obtain a concentration prediction model.
[0057] The prediction module is used to use the site distribution data and site characteristic data of the site to be predicted and its surrounding sites to obtain the spatial characteristics to be predicted, fuse the spatial characteristics to be predicted with the historical PM2.5 concentration data of the site to be predicted to obtain the spatiotemporal characteristics to be predicted, input the spatiotemporal characteristics to be predicted into the concentration prediction model, and output the predicted PM2.5 concentration value within a preset time period in the future.
[0058] Each module in the aforementioned PM2.5 concentration prediction system based on long-term spatiotemporal sequences can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.
[0059] The present invention also provides a computer device comprising a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in an embodiment of a method for predicting PM2.5 concentration based on a long spatiotemporal sequence. The specific implementation method can be found in the method embodiment and will not be further described here.
[0060] Furthermore, the present invention also provides a non-temporary computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions, the instructions can be executed by a processor of a computer device to complete the above method. For example, the non-temporary computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device. When the computer program is executed by the processor, it can implement the steps in an embodiment of a method for predicting PM2.5 concentration based on a long spatiotemporal sequence. The specific implementation method can be found in the method embodiment, which will not be repeated here.
[0061] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0062] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0063] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.
[0064] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.
[0065] It should be pointed out that the specific implementation methods described above can enable those skilled in the art to understand the invention more comprehensively, but do not limit the invention in any way. Therefore, although the present specification and examples have described the invention in detail, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the scope of protection of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved. Any simple change or equivalent replacement of the technical solution that can be obviously obtained by any person familiar with the art within the technical scope disclosed in the present invention falls within the scope of protection of the present invention.
Claims
1. A PM2.5 concentration prediction method based on long-term spatiotemporal series, characterized in that: The following steps are involved: Obtain atmospheric pollutant concentration data and meteorological data for a continuous time period at a target site, and obtain a time feature sequence based on the atmospheric pollutant concentration data and meteorological data for the continuous time period at the target site; filter out feature columns related to PM2.5 concentration in the time feature sequence; obtain location distribution data and historical PM2.5 concentration data for the target site and its surrounding sites, and input the site distribution data and site feature data into a graph convolutional network (GCN) model to obtain spatial features; The filtered feature columns are fused with spatial features to output spatiotemporal sequence data; The Informer-GCN model and the LSTM-GCN model are used as base learners, and the XGBoost model including the attention mechanism is used as the second-layer learner. The base learners and the second-layer learners are integrated through the Stacking integration strategy. The integrated model is trained according to the spatiotemporal series data to obtain a concentration prediction model; The site distribution data and site characteristic data of the site to be predicted and its surrounding sites are used to obtain the spatial characteristics to be predicted, and the spatial characteristics to be predicted and the historical PM2.5 concentration data of the site to be predicted are fused to obtain the spatiotemporal characteristics to be predicted. The spatiotemporal characteristics to be predicted are input into the concentration prediction model to output the predicted PM2.5 concentration value within a preset time period in the future.
2. The PM2.5 concentration prediction method based on long-term spatiotemporal series according to claim 1, characterized in that: Before selecting feature columns highly correlated with PM2.5 concentrations, the Z-score method is used to detect outliers, which are then replaced using linear interpolation. Minimum-maximum normalization is then performed on each feature column and label column.
3. The PM2.5 concentration prediction method based on long-term spatiotemporal series according to claim 1, characterized in that: Feature columns related to PM2.5 concentration are screened out from the time feature sequence, specifically by using random forest and Pearson correlation coefficient.
4. The PM2.5 concentration prediction method based on long-term spatiotemporal series according to claim 1, characterized in that: It also includes the use of root mean square error RMSE, mean absolute error MAE and correlation coefficient R 2 The accuracy of the concentration prediction model was evaluated.
5. The PM2.5 concentration prediction method based on long-term spatiotemporal series according to claim 1, characterized in that: The time feature sequence specifically includes a timestamp column, a feature column, and a label column.
6. A PM2.5 concentration prediction system based on long-term spatial and temporal sequences, characterized in that: include: A feature acquisition module is used to obtain atmospheric pollutant concentration data and meteorological data for a continuous time period at a target site, and to obtain a time feature sequence based on the atmospheric pollutant concentration data and meteorological data for the continuous time period at the target site; to filter out feature columns related to PM2.5 concentration from the time feature sequence; to obtain location distribution data and historical PM2.5 concentration data for the target site and its surrounding sites, and to input the site distribution data and site feature data into a graph convolutional network (GCN) model to obtain spatial features; The filtered feature columns are fused with spatial features to output spatiotemporal sequence data; A model construction module is used to use the Informer-GCN model and the LSTM-GCN model as base learners, and the XGBoost model including the attention mechanism as the second-layer learner, integrate the base learners and the second-layer learners through the Stacking integration strategy, and train the integrated model according to the spatiotemporal series data to obtain a concentration prediction model; The prediction module is used to use the site distribution data and site feature data of the site to be predicted and its surrounding sites to obtain the spatial features to be predicted, fuse the spatial features to be predicted with the historical PM2.5 concentration data of the site to be predicted to obtain the spatiotemporal features to be predicted, input the spatiotemporal features to be predicted into the concentration prediction model, and output the predicted PM2.5 concentration value within a preset time period in the future.
7. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 4.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is loaded into a processor, it can execute the steps of the method according to any one of claims 1 to 4.