Atmospheric pollutant early warning method and system based on multi-source isomerism

Abstract

The invention relates to an atmospheric pollutant early warning method and system based on multi-source isomerism, and the method comprises the steps: obtaining multi-source isomerism atmospheric pollutant data, carrying out the data arrangement of the atmospheric pollutant data, and obtaining the arranged atmospheric pollutant data; carrying out unified representation on the sorted air pollutant data to obtain air pollutant data subjected to unified representation, and carrying out intelligent detection by adopting a gridding edge calculation method according to the air pollutant data subjected to unified representation to obtain an air pollutant detection result; according to the detection result of the atmospheric pollutants, performing forecasting and early warning by adopting a method based on big data analysis to obtain a forecasting and early warning result of the atmospheric pollutants; and the forecasting and early warning result of the atmospheric pollutants is displayed through an artificial intelligence GIS platform, so that the purposes of accurately monitoring, forecasting and early warning the atmospheric pollutants, improving the atmospheric pollution treatment level, assisting in environmental protection decision making and promoting development of related technologies are achieved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an atmospheric pollutant early warning method and system based on multi-source heterogeneity. Background Art

[0002] The core technical challenge facing atmospheric pollutant monitoring and early warning is how to effectively integrate and utilize massive amounts of heterogeneous data from multiple sources. Traditional methods are difficult to fully mine and correlate pollutant data from different sources and in different formats, resulting in incomplete monitoring results and untimely early warnings. Specifically, there are problems such as uneven data quality and a large number of missing values ​​in the data collection link; it is difficult to uniformly characterize the characteristics of multi-source heterogeneous data in the data processing stage; the intelligent detection link is limited by computing resources and it is difficult to achieve large-scale real-time processing; the accuracy and timeliness of the forecast and early warning model need to be improved. In addition, how to intuitively display the complex distribution of pollutants and realize pollution source tracing and weight analysis are also difficult problems that need to be solved urgently. These problems are interrelated and progressive, and together constitute the systemic challenges of intelligent monitoring and early warning of atmospheric pollutants. Summary of the invention

[0003] The main purpose of the present invention is to provide an atmospheric pollutant early warning method and system based on multi-source heterogeneity, so as to achieve accurate monitoring and forecasting of atmospheric pollutants, improve the level of atmospheric pollution control, assist environmental protection decision-making and promote the development of related technologies.

[0004] To achieve the above object, the present invention provides an atmospheric pollutant early warning method based on multi-source heterogeneity, comprising the following steps:

[0005] Acquiring multi-source heterogeneous atmospheric pollutant data, and sorting the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data;

[0006] The collated air pollutant data are uniformly characterized to obtain uniformly characterized air pollutant data, and a grid-based edge computing method is used to perform intelligent detection based on the uniformly characterized air pollutant data to obtain detection results of air pollutants;

[0007] According to the detection results of the air pollutants, a forecast and warning method based on big data analysis is used to obtain forecast and warning results of air pollutants;

[0008] The forecast and warning results of the atmospheric pollutants are displayed through an artificial intelligence GIS platform.

[0009] Furthermore, the step of collating the multi-source heterogeneous atmospheric pollutant data includes:

[0010] Performing data exchange processing on the multi-source heterogeneous atmospheric pollutant data to obtain clustering feature information;

[0011] Combining multi-view clustering and integrated clustering methods, data sharing and data repair are performed on the cluster feature information;

[0012] The repaired data are cleaned, summarized and transformed to obtain sorted atmospheric pollutant data.

[0013] Furthermore, the step of uniformly characterizing the sorted atmospheric pollutant data includes:

[0014] Using a generative adversarial network model to perform unbalanced data processing on the sorted atmospheric pollutant data;

[0015] Based on the multi-objective optimization algorithm, the processed atmospheric pollutant data are adaptively calibrated;

[0016] A deep learning representation model is constructed to extract features from the calibrated atmospheric pollutant data to obtain uniformly represented atmospheric pollutant data.

[0017] Furthermore, the step of using the grid edge computing method to perform intelligent detection to obtain the detection results of air pollutants includes:

[0018] Dividing the uniformly characterized atmospheric pollutant data into a plurality of grid cells;

[0019] Perform edge computing on the atmospheric pollutant data in each grid cell to screen and filter duplicate data;

[0020] According to the calculation results, the atmospheric pollutants are intelligently identified and tracked to obtain the detection results of the atmospheric pollutants.

[0021] Furthermore, the step of using the grid edge computing method to perform intelligent detection to obtain the detection results of air pollutants includes:

[0022] Dividing the uniformly characterized atmospheric pollutant data into a plurality of grid cells;

[0023] Perform edge computing on the atmospheric pollutant data in each grid cell to screen and filter duplicate data;

[0024] According to the calculation results, the atmospheric pollutants are intelligently identified and tracked to obtain the detection results of the atmospheric pollutants.

[0025] Furthermore, the step of establishing an intelligent monitoring and reasoning decision model for air pollutant sources based on the temporal and spatial dynamic change trend information includes:

[0026] According to the pollution cloud map simulated by numerical simulation, pollutants at a specified time and place can be queried to trace the source of pollutants;

[0027] Calculate the proportion of pollution caused by a designated pollution source to a designated location and obtain the pollution degree equation;

[0028] When the same location is affected by multiple pollution sources, the weights of different pollution sources are determined according to the ratio of pollution degrees;

[0029] Determine the control parameters of air pollutants by determining the allowable pollution concentration, the allowable emission concentration of each pollution source and the emission reduction;

[0030] The emission concentration of each pollution source is calculated based on real-time monitoring data, and an early warning model is formed in combination with the allowable emission concentration of the pollution source. The early warning threshold is determined based on the actual measured value of the early warning indicator. When the corresponding threshold is exceeded, corresponding measures are taken, which include increasing the monitoring frequency, special inspections, and comprehensive evaluations.

[0031] Based on the pollution degree equation, the pollution source weight, the control parameters, the early warning mode and the threshold setting information, an intelligent monitoring and reasoning decision model for the source of air pollutants is established.

[0032] Furthermore, the step of establishing an intelligent monitoring and reasoning decision model for air pollutant sources based on the temporal and spatial dynamic change trend information includes:

[0033] According to the pollution cloud map simulated by numerical simulation, pollutants at a specified time and place can be queried to trace the source of pollutants;

[0034] Calculate the proportion of pollution caused by a designated pollution source to a designated location and obtain the pollution degree equation;

[0035] When the same location is affected by multiple pollution sources, the weights of different pollution sources are determined according to the ratio of pollution degrees;

[0036] Determine the control parameters of air pollutants by determining the allowable pollution concentration, the allowable emission concentration of each pollution source and the emission reduction;

[0037] The emission concentration of each pollution source is calculated based on real-time monitoring data, and an early warning model is formed in combination with the allowable emission concentration of the pollution source. The early warning threshold is determined based on the actual measured value of the early warning indicator. When the corresponding threshold is exceeded, corresponding measures are taken, which include increasing the monitoring frequency, special inspections, and comprehensive evaluations.

[0038] Based on the pollution degree equation, the pollution source weight, the control parameters, the early warning mode and the threshold setting information, an intelligent monitoring and reasoning decision model for the source of air pollutants is established.

[0039] The present invention also provides an atmospheric pollutant early warning system based on multi-source heterogeneity, comprising:

[0040] A data acquisition module is used to obtain multi-source heterogeneous atmospheric pollutant data, and to sort the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data;

[0041] A data processing module is used to uniformly characterize the collated air pollutant data to obtain uniformly characterized air pollutant data, and to perform intelligent detection using a grid-based edge computing method based on the uniformly characterized air pollutant data to obtain detection results of air pollutants;

[0042] A data analysis module is used to perform forecast and warning according to the detection results of the atmospheric pollutants by using a method based on big data analysis to obtain forecast and warning results of atmospheric pollutants;

[0043] The information display module is used to display the forecast and warning results of the atmospheric pollutants through an artificial intelligence GIS platform.

[0044] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned multi-source heterogeneous based atmospheric pollutant early warning method when executing the computer program.

[0045] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned atmospheric pollutant early warning method based on multi-source heterogeneity are implemented.

[0046] The multi-source heterogeneous atmospheric pollutant early warning method and system provided by the present invention has the following beneficial effects: the present invention integrates a variety of advanced technologies and applies them to the entire process of atmospheric pollution monitoring and early warning, thus realizing the intelligence from data acquisition, processing to analysis and early warning. The pollutant distribution is intuitively displayed on the GIS platform, providing strong support for atmospheric pollution prevention and control decision-making, which has important practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flow chart of an atmospheric pollutant early warning method based on multi-source heterogeneity in one embodiment of the present invention;

[0048] Figure 2 It is a structural block diagram of an atmospheric pollutant early warning system based on multi-source heterogeneity in one embodiment of the present invention;

[0049] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0050] 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

[0051] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] Reference Figure 1 , is a flow chart of a method for early warning of atmospheric pollutants based on multi-source heterogeneity proposed by the present invention, comprising the following steps:

[0053] S1, acquiring multi-source heterogeneous atmospheric pollutant data, and sorting the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data;

[0054] S2, uniformly characterizing the sorted atmospheric pollutant data to obtain uniformly characterized atmospheric pollutant data, and using a grid-based edge computing method to perform intelligent detection based on the uniformly characterized atmospheric pollutant data to obtain detection results of atmospheric pollutants;

[0055] S3, according to the detection results of the air pollutants, using a method based on big data analysis to perform forecast and warning, and obtain forecast and warning results of air pollutants;

[0056] S4, displaying the forecast and warning results of the atmospheric pollutants through an artificial intelligence GIS platform.

[0057] As described in step S1, multi-source heterogeneous atmospheric pollutant data are obtained; and the multi-source heterogeneous atmospheric pollutant data are sorted to obtain sorted atmospheric pollutant data.

[0058] The ways to obtain multi-source heterogeneous atmospheric pollutant data are as follows: The atmospheric pollutant early warning system based on multi-source heterogeneity includes a data acquisition module, which is equipped with optical sensors and electrochemical sensors. Optical sensors are used to monitor the concentration of particulate matter and dust in the atmosphere. Their working principle is based on the detection of light scattering or absorption characteristics. They can obtain real-time information on the concentration of particulate matter in the atmosphere without changing the physical and chemical properties of suspended dust. Electrochemical sensors are used to monitor pollutants such as hydrocarbons, sulfides, sulfur oxides, and nitrogen oxides. They monitor by generating electrical signals related to the concentration of pollutants through electrochemical reactions. Obtaining real-time concentration data of major atmospheric pollutants such as sulfur dioxide, nitrogen dioxide, inhalable particulate matter (PM10), fine particulate matter (PM2.5), carbon monoxide, and ozone constitutes multi-source heterogeneous atmospheric pollutant data. The acquired multi-source heterogeneous atmospheric pollutant data needs to be sorted and preprocessed for subsequent analysis and application.

[0059] The data quality is controlled to remove abnormal and erroneous values. The monitoring equipment may have drift, failure and other problems that may cause data distortion, which requires calibration and correction. Secondly, data cleaning and conversion are required to convert data in different formats into a unified standard format, such as scientific data formats such as NetCDF, HDF or the commonly used CSV format. For emission inventory data in text format, parsing and extraction are required. Finally, data fusion is performed to match data from different sources to a unified spatiotemporal benchmark. The monitoring station data is point data, which can be spatially interpolated to form grid data; the spatial resolution of remote sensing data and model data is low, and downscaling is required; the time resolution of emission data and actual monitoring data may be inconsistent, and time registration is required. In the process of data fusion, the complementarity between different data sources can also be used, such as using ground monitoring data to correct satellite remote sensing data, and using emission data to constrain model simulation results. Finally, the sorted data should be stored and managed, and a database and metadata should be established to facilitate subsequent calls and analysis. Metadata records important information such as the source, spatiotemporal range, resolution, and uncertainty of the data. The amount of atmospheric pollutant data is large, and distributed storage and parallel computing technologies are used for optimized management and processing. After this series of data collation work, multi-source heterogeneous atmospheric pollutant data are integrated into a unified data framework, providing high-quality data support for subsequent analysis applications.

[0060] As described in step S2, the sorted atmospheric pollutant data are uniformly characterized to obtain the atmospheric pollutant data after uniform characterization; based on the atmospheric pollutant data after uniform characterization, a grid-based edge computing method is used to perform intelligent detection to obtain the detection results of atmospheric pollutants.

[0061] The study area is divided into regular grids. The size of each grid can be set according to actual needs, for example, it can be set to 1km×1km. The atmospheric pollutant data of each data source after unified characterization is mapped to the corresponding grid. For example, if the spatial resolution of the AOD data of satellite remote sensing at a certain moment is 10km, it will be evenly distributed to the corresponding 100 1km grids; the PM2.5 data of the ground monitoring station is directly allocated to the 1km grid where it is located. After this processing, each 1km grid has pollutant data from different data sources. Intelligent analysis and detection of pollutant data in each grid. Since the data has been uniformly characterized and the features of each data source are distributed in the same vector space, it is convenient to use machine learning algorithms for modeling and analysis. Taking the detection of PM2.5 concentration exceeding the standard as an example, the multi-source data in the grid can be input into the pre-trained PM2.5 concentration prediction model (such as a deep neural network model). The model outputs the PM2.5 concentration value of the grid, and then compares it with the national standard value to determine whether the grid exceeds the standard. The detection results of all grids form a gridded pollutant distribution map, which can intuitively reflect the spatial distribution characteristics of the pollution status within the study area, and provide a basis for further pollution source tracing, early warning and forecasting, etc. This gridded edge computing method can make full use of local computing resources, reduce bandwidth pressure for data transmission, and improve analysis and detection efficiency.

[0062] As described in step S3, according to the detection results of the atmospheric pollutants, a forecast and warning method based on big data analysis is adopted to obtain forecast and warning results of atmospheric pollutants.

[0063] According to the detection results of the atmospheric pollutants, a method based on big data analysis is used to forecast and warn, and the forecast and warning results of atmospheric pollutants are obtained. Specifically, it is necessary to collect and organize the monitoring data of various atmospheric pollutants, such as the concentration values ​​of sulfur dioxide, nitrogen dioxide, and inhalable particulate matter. These data usually come from air quality monitoring stations in various places, forming a large amount of spatiotemporal data. After obtaining massive monitoring data, it is necessary to use big data analysis technology to process and mine it. Common methods include time series analysis, spatial interpolation, machine learning, etc. Taking time series analysis as an example, the changing trend and periodicity of the concentration of each pollutant at a certain monitoring point can be analyzed to determine whether there are characteristics such as seasonality and holiday effects. Spatial interpolation can infer the distribution of pollutant concentrations in the entire area based on the geographical location and pollutant concentration values ​​of each monitoring point. Machine learning algorithms such as support vector machines and random forests build prediction models by training historical data to forecast the concentration of pollutants in the future. On the basis of the forecast, it is also necessary to combine the ambient air quality standards to warn of possible pollution. For example, when it is predicted that the average daily concentration of PM2.5 may exceed 75 micrograms / cubic meter in the next 24 hours, a mild pollution warning can be issued; when it exceeds 115 micrograms / cubic meter, a moderate pollution warning can be issued; when it exceeds 150 micrograms / cubic meter, a severe pollution warning can be issued. Warning information needs to be released to the public in a timely manner through the Internet, television, radio and other channels to remind the public to take necessary protective measures. At the same time, the warning results are also an important basis for decision-making on air pollution control. According to the warning level, emergency measures such as traffic restrictions, suspension of work, and spraying of dust suppressants can be taken in advance to control pollutant emissions from the source. Big data analysis can also reveal the causes of pollution in depth. For example, through the analysis of pollutant components, it is found that coal burning is the main source of PM2.5, and the use of coal should be controlled.

[0064] As described in step S4, the forecast and warning results of the atmospheric pollutants are displayed through the artificial intelligence GIS platform.

[0065] The forecast and early warning results of atmospheric pollutants can be displayed through the artificial intelligence GIS platform, which can intuitively and efficiently present the dynamic changes of atmospheric pollution. Specifically, the GIS platform integrates multi-source heterogeneous data such as atmospheric pollutant concentration data and meteorological data at different time and space scales, and reveals the temporal and spatial distribution laws and evolution trends of pollutants through data mining and machine learning algorithms. On this basis, artificial intelligence technologies such as deep learning networks are used to establish an atmospheric pollution forecast and early warning model to simulate the atmospheric pollution process and predict changes in pollution conditions in the future. The results of forecasts and early warnings can be presented on the GIS platform in a variety of visualization methods, such as using thematic maps to display the spatial distribution of regional pollutant concentrations, using color scales or contour lines to represent different pollution levels; generating time series curves to represent the dynamic trend of the pollution process; and even using three-dimensional visualization technology to build virtual reality scenes to simulate the diffusion process of atmospheric pollution, allowing users to experience changes in air quality in an immersive way. The artificial intelligence GIS platform displays the results of atmospheric pollution forecasts and early warnings, which can achieve real-time monitoring, dynamic display, intelligent early warning and precise decision-making of pollution conditions, and provide strong support for the prevention and control of atmospheric pollution. Through spatial mapping, time series analysis, 3D simulation and other visualization methods, the temporal and spatial distribution and evolution of air pollution are revealed intuitively, making it easier for the government and the public to obtain air quality information in a timely manner. At the same time, with the help of artificial intelligence algorithms, pollution hotspots can be intelligently identified to improve the timeliness and accuracy of early warnings, thereby guiding air pollution control work more accurately and efficiently.

[0066] In one embodiment, the atmospheric pollutant data is acquired in real time by optical sensors and electrochemical sensors. For example, data collection is performed in a certain park, and monitoring data is combined and arranged within a certain area (for example, with the center of the park as the origin and a radius of 5 kilometers) based on the big data platform. Ten monitoring points are set up in the area, and each monitoring point is equipped with the above two types of sensors. During the data collection process, the data drift and equipment abnormalities of the sensor are monitored in real time. The big data platform collects and analyzes the sensor data once every 5 minutes. Once abnormal data fluctuations are found, the early warning mechanism is immediately activated to notify the technicians to check. At the same time, the sensor equipment is regularly quality controlled and calibrated once a week using a mobile calibration vehicle to ensure the accuracy and reliability of the monitoring data. The acquired multi-source heterogeneous atmospheric pollutant data (including component composition data set, volume data set and image data set) are sorted. In the data sorting unit, the data exchange processing module processes these data. For example, for atmospheric pollutant data from different monitoring points and different sensor types, their feature information is integrated through a specific data exchange algorithm. Taking the gas pollutant data (including carbon monoxide, sulfur dioxide and other component data) collected at a monitoring point in the park and the particulate matter image data collected at another monitoring point as examples, after data exchange processing, their feature information is associated and fused according to certain rules to obtain cluster feature information. For example, carbon monoxide and sulfur dioxide data with similar concentration change trends are clustered together, and the particulate matter image data in the areas related to these gas pollutant emission sources are also associated with the same cluster.

[0067] The cluster feature information is processed through the data sharing and repair module combined with multi-perspective clustering and integrated clustering methods. Since the sensor data of different monitoring points may have data redundancy or missing data, this module uses multi-perspective clustering methods to analyze data features from multiple perspectives, such as clustering analysis from the time dimension (changes in pollutant concentrations in different time periods) and the spatial dimension (pollutant distribution at different monitoring point locations). At the same time, the integrated clustering method integrates multiple clustering results to achieve data sharing and complementarity. For parts with missing data, they are automatically repaired through the similarity and complementarity of clustering features. For example, if the sulfur dioxide concentration data of a certain monitoring point in a certain period of time is missing, the system will reasonably estimate and repair the missing data based on the sulfur dioxide concentration data of other monitoring points with similar environmental characteristics (such as surrounding pollution source types, meteorological conditions, etc.) and the concentration data of other related pollutants (such as nitrogen oxides) at the same monitoring point. After data sharing and repair, the data sets enter the data cleaning, aggregation and conversion process to clean the data and remove obviously erroneous data (such as data that exceeds the normal concentration range, such as data with carbon monoxide concentration greater than 100ppm, which is considered abnormal data and eliminated). Then, data of different formats and sources are aggregated, such as converting the raw data formats from different sensors into a standard format for subsequent processing (such as converting the current signal output by the electrochemical sensor into pollutant concentration data in mg / m³). Finally, the processed data is uploaded to the system database to complete the data sorting process and obtain sorted atmospheric pollutant data.

[0068] The sorted atmospheric pollutant data enters the unified data representation unit and is first processed by the unbalanced data processing module using the generative adversarial network (GANs) model. In the atmospheric pollutant monitoring data of the park, there are some pollutants with large data volumes (such as particulate matter concentration data, a large number of samples can be collected every day), while some pollutants have small data volumes (such as some volatile organic compound data, which are difficult to collect and have relatively small sample volumes), which is the data imbalance problem. The generator in the GANs model captures the potential distribution of real data samples (including a large amount of particulate matter concentration data and a small amount of volatile organic compound data) and generates new sample data, while the discriminator is used to determine whether the input sample is real or generated. Through continuous training, the samples generated by the generator gradually approach the real samples, thereby increasing the data volume of a small number of sample pollutants and realizing the unbalanced processing of atmospheric pollutant data. Based on the multi-objective optimization algorithm, the adaptive calibration module calibrates the processed atmospheric pollutant data. In different areas of the park (such as industrial production areas, residential areas, commercial areas, etc.), the atmospheric environment varies greatly, and the sensor measurement results are easily affected by environmental factors and produce deviations. By combining monitoring data in different areas, for example, setting up multiple monitoring points in industrial production areas, and using national standard method instruments and gas sensors to combine and arrange the points. The sensor equipment obtains monitoring data in real time, and compares it with the monitoring data of the national standard method instrument, and calibrates the data adaptively according to the comparison results. For example, when the nitrogen oxide concentration measured by the electrochemical sensor deviates from the measurement result of the national standard method instrument by more than 5%, the system automatically adjusts the calibration parameters of the sensor to ensure the accuracy of the measurement results, ensure the long-term stable operation of the system and the effectiveness and reliability of the monitoring results.

[0069] Extract features from the calibrated atmospheric pollutant data. The spatiotemporal transformation mapping method is used to convert the one-dimensional pollutant concentration data (such as carbon monoxide concentration data in time series) obtained by the observation network into a two-dimensional representation, so as to obtain the location distribution information of adjacent collection points, extract the connection features between the collected data, and unify them with other observation data (such as temperature and humidity data in meteorological data). The unsupervised deep learning model is constructed using the deep learning auto-encoder (AE) framework, and the model is improved for different observation data types (such as gas pollutant data and particulate matter data). The feature self-learning of the observation data is realized by unsupervised learning, and then a new deep learning model is constructed. The model is trained in a supervised manner using labeled data (such as sample data with known pollution source type and concentration). The unsupervised learning method is combined with the supervised learning method to obtain features with stronger representation capabilities, realize the unified representation processing of multi-source heterogeneous observation data of atmospheric pollutants, and obtain the atmospheric pollutant data after unified representation. The atmospheric pollutant data after unified representation enters the intelligent detection unit, and the grid division module divides it into multiple grid units. Taking the park in Jiaxing City as an example, the park is divided into 100 square grid units with a side length of 500 meters according to the geographical scope and monitoring needs of the park. The atmospheric pollutant data in each grid unit is processed independently, and the edge computing processing module performs edge computing processing on the data in each grid unit. Due to the large scale of atmospheric pollution source data and high data collection efficiency requirements, this module can remove duplicate information from the data in the grid unit and filter out useless information. For example, in a certain grid unit, multiple sensors may collect similar pollutant concentration data in the same time period. The edge computing processing module retains the most representative data through data comparison and screening, reduces the amount of data transmission, and improves computing efficiency.

[0070] According to the edge computing processing results, the atmospheric pollutants are intelligently identified and tracked. There are different types of pollution sources in the park, such as industrial clusters, catering clusters, and road traffic. By analyzing the pollutant data in the grid unit, the pollutant characteristics of different types of pollution sources can be identified (such as specific chemical components in industrial waste gas, particulate matter characteristics in catering oil fume, etc.). For key pollution sources, the system performs hierarchical protection and technical tracking. For example, industrial enterprises with large emissions and excessive pollutant concentrations are marked as key monitoring objects, and their pollutant emissions are tracked in real time. When the pollutant concentration increases abnormally, an alarm is issued in time, and the emission source of the pollutant is traced back to obtain the detection results of atmospheric pollutants. The obtained atmospheric pollutant detection results are transmitted to the forecast and early warning unit, and dynamic information on the time domain and airspace dynamic change trend of multi-source heterogeneous atmospheric pollutants is obtained based on the GIS platform. In the park of the embodiment, atmospheric pollution monitoring data is collected at multiple monitoring points (a total of 30 monitoring points) arranged in the designated area, and pollution sources are identified and classified into pollution point sources (such as factory chimney emissions), pollution line sources (such as road traffic emissions), and pollution surface sources (such as large-scale construction dust). Based on the GIS platform, the location information of these monitoring points is combined with pollutant concentration data and meteorological data (temperature, humidity, wind speed and direction, etc.) to analyze dynamic information such as concentration change trends and diffusion directions of atmospheric pollutants in different time periods (such as every hour, every day) and different areas (such as different functional areas in the park).

[0071] Based on the dynamic change trend information obtained, an intelligent monitoring and reasoning decision model for air pollutant sources is established. For example:

[0072] According to the pollution cloud map simulated by numerical simulation, query the pollutants (such as inhalable particulate matter PM2.5) at the specified time (such as 10 am) and the specified location (such as near a school in the park), and trace the source of the pollutants (such as dust emissions from a nearby construction site). Calculate the proportion of pollution caused by the specified pollution source (construction site dust) to the specified location (near the school) to obtain the pollution degree equation. For example, by analyzing the relationship between the change in pollutant concentration at the location and the intensity of dust emissions from the construction site over a certain period of time, a pollution degree equation is established to quantify the contribution of the pollution source to the pollution at the location. When the same location (near the school) is affected by multiple pollution sources (such as construction site dust and surrounding road traffic emissions), the weights of different pollution sources are determined according to the ratio of pollution degrees. For example, if it is calculated that the contribution of construction site dust to PM2.5 pollution at the location is 60%, and the contribution of road traffic emissions is 40%, then their weights are determined to be 0.6 and 0.4 respectively. By determining the permissible pollution concentration (according to relevant national and local environmental standards, the permissible concentration of PM2.5 in this area is 75μg / m³), the permissible emission concentration of each pollution source (such as the permissible emission concentration of dust from construction sites is 1mg / m³, and the permissible emission concentration of road traffic emissions is 0.5mg / m³) and the reduction of emissions. For pollution sources that exceed the permissible emission concentration, calculate the emissions that need to be reduced. For example, if the actual emission concentration of dust from construction sites is 1.5mg / m³, which exceeds the permissible emission concentration of 0.5mg / m³, it is necessary to formulate an emission reduction plan and determine the emission reduction target. The emission concentration of each pollution source is calculated based on real-time monitoring data, and combined with the permissible emission concentration of the pollution source, a four-level early warning mode is formed. The warning levels are green (qualified, such as pollutant concentration is lower than 80% of the allowed concentration), blue (general pollution, such as pollutant concentration is between 80% and 100% of the allowed concentration), yellow (severe pollution, such as pollutant concentration is between 100% and 150% of the allowed concentration), and red (particularly serious, such as pollutant concentration exceeds 150% of the allowed concentration), corresponding to different warning threshold ranges (such as the green threshold is a pollutant concentration of less than 60μg / m³, the blue threshold is 60-75μg / m³, etc.). The yellow and red warning thresholds are determined according to the measured values ​​of the warning indicators. When the corresponding thresholds are exceeded, different measures are taken, including increasing the monitoring frequency (such as the monitoring frequency from once an hour to once every 30 minutes during the yellow warning), special inspections (such as the red warning, a comprehensive inspection of the pollution source, including equipment operation, implementation of pollution control measures, etc.), and comprehensive assessment (comprehensive assessment is conducted based on factors such as the type of pollution source, emission history, and surrounding environmental impact, and targeted governance and control plans are formulated).Based on the above-obtained pollution degree equation, pollution source weight, control parameters, early warning mode and threshold setting information, an intelligent monitoring and reasoning decision model for air pollutant sources is established for intelligent monitoring, reasoning decision-making and subsequent forecast and early warning analysis of air pollutants. For example, when the emission concentration of a certain pollution source is monitored to increase, and according to the model calculation, the pollutant concentration in a certain area may exceed the early warning threshold, the system automatically issues an early warning and provides corresponding decision-making suggestions, such as adjusting the production process of the pollution source and strengthening the operation of pollution control facilities.

[0073] Finally, the result analysis module analyzes the detection results of atmospheric pollutants according to the intelligent monitoring and reasoning decision model to obtain the forecast and warning results of atmospheric pollutants. For example, according to the current pollution source emissions, meteorological conditions (such as wind speed and direction, rainfall probability, etc.) and model predictions, it is predicted that the PM2.5 concentration in a certain area of ​​the park will reach the yellow warning level in the next 6 hours. The system notifies relevant departments and enterprises in advance to take corresponding emission reduction and protection measures, such as requiring enterprises to reduce production and increase the frequency of road watering and dust reduction, so as to reduce the concentration of atmospheric pollutants and prevent further deterioration of pollution. The forecast and warning results enter the display platform unit. The GIS map platform development module develops a lightweight scalable GIS (LSGIS) map platform based on Amap to support artificial intelligence applications. The platform uses REST to provide map services for the front end, is compatible with the latest ARCGIS REST service, and supports the import of multiple GIS data formats and standard Web OGC service protocols. The forecast and warning results of atmospheric pollutants are integrated with the GIS map to form a pollutant concentration distribution map. On the GIS map of the Jiaxing Park, different colored areas represent different levels of pollutant concentration (such as green areas for low concentrations and red areas for high concentrations), and the location of pollution sources (such as factories, traffic arteries, etc.) and meteorological information (such as wind direction arrows, temperature and humidity markings, etc.) are marked to intuitively display the distribution of atmospheric pollutants in the park. By developing an artificial intelligence GIS application interface, spatial analysis and query functions of pollutant concentration distribution maps are realized. Through this interface, users can click on any area on the map to query detailed pollutant information in the area (such as pollutant types, concentration values, source analysis, etc.), conduct spatial analysis (such as calculating the average value of pollutant concentrations in different areas, analyzing the spatial gradient changes of pollutant concentrations, etc.), and customize route planning according to needs (such as planning driving routes to avoid highly polluted areas). At the same time, a heat map is generated to show the hot spots of pollutant concentrations, providing a decision-making basis for the park's environmental management, enterprise production arrangements, and residents' travel. For example, the park management department can reasonably adjust the production layout of enterprises and optimize the traffic flow of the park based on the pollutant concentration distribution map to reduce the emission of atmospheric pollutants and the impact on residents. Residents can use mobile phone applications to query the pollution situation of the surrounding environment and reasonably arrange outdoor activity time and routes.

[0074] Reference Figure 2 , is a structural block diagram of an atmospheric pollutant early warning system based on multi-source heterogeneity in one embodiment of the present invention, including:

[0075] A data acquisition module is used to obtain multi-source heterogeneous atmospheric pollutant data, and to sort the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data;

[0076] A data processing module is used to uniformly characterize the collated air pollutant data to obtain uniformly characterized air pollutant data, and to perform intelligent detection using a grid-based edge computing method based on the uniformly characterized air pollutant data to obtain detection results of air pollutants;

[0077] A data analysis module is used to perform forecast and warning according to the detection results of the atmospheric pollutants by using a method based on big data analysis to obtain forecast and warning results of atmospheric pollutants;

[0078] The information display module is used to display the forecast and warning results of the atmospheric pollutants through an artificial intelligence GIS platform.

[0079] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer 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 the corresponding data in this embodiment. The network 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 above method is implemented.

[0080] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0081] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0082] In summary, the present invention obtains multi-source heterogeneous atmospheric pollutant data, organizes the multi-source heterogeneous atmospheric pollutant data, and obtains organized atmospheric pollutant data; uniformly characterizes the organized atmospheric pollutant data to obtain uniformly characterized atmospheric pollutant data, and uses a gridded edge computing method to perform intelligent detection based on the uniformly characterized atmospheric pollutant data to obtain detection results of atmospheric pollutants; uses a big data analysis-based method to perform forecasting and warning based on the detection results of atmospheric pollutants to obtain forecasting and warning results of atmospheric pollutants; and displays the forecasting and warning results of atmospheric pollutants through an artificial intelligence GIS platform to achieve accurate monitoring and forecasting and warning of atmospheric pollutants, improve the level of atmospheric pollution control, assist environmental protection decision-making, and promote the development of related technologies.

[0083] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods 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 memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0084] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0085] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for early warning of atmospheric pollutants based on multi-source heterogeneity, characterized in that: The following steps are involved: Acquiring multi-source heterogeneous atmospheric pollutant data, and sorting the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data; The collated air pollutant data are uniformly characterized to obtain uniformly characterized air pollutant data, and a grid-based edge computing method is used to perform intelligent detection based on the uniformly characterized air pollutant data to obtain detection results of air pollutants; According to the detection results of the air pollutants, a forecast and warning method based on big data analysis is used to obtain forecast and warning results of air pollutants; The forecast and warning results of the atmospheric pollutants are displayed through an artificial intelligence GIS platform.

2. According to the multi-source heterogeneous atmospheric pollutant early warning method of claim 1, the step of collating the multi-source heterogeneous atmospheric pollutant data comprises: Performing data exchange processing on the multi-source heterogeneous atmospheric pollutant data to obtain clustering feature information; Combining multi-view clustering and integrated clustering methods, data sharing and data repair are performed on the cluster feature information; The repaired data are cleaned, summarized and transformed to obtain sorted atmospheric pollutant data.

3. The method for early warning of atmospheric pollutants based on multi-source heterogeneity according to claim 1 is characterized in that: The step of uniformly characterizing the sorted atmospheric pollutant data includes: Using a generative adversarial network model to perform unbalanced data processing on the sorted atmospheric pollutant data; Based on the multi-objective optimization algorithm, the processed atmospheric pollutant data are adaptively calibrated; A deep learning representation model is constructed to extract features from the calibrated atmospheric pollutant data to obtain uniformly represented atmospheric pollutant data.

4. The method for early warning of atmospheric pollutants based on multi-source heterogeneity according to claim 1 is characterized in that: The step of using the grid edge computing method to perform intelligent detection to obtain the detection results of air pollutants includes: Dividing the uniformly characterized atmospheric pollutant data into a plurality of grid cells; Perform edge computing on the atmospheric pollutant data in each grid cell to screen and filter duplicate data; According to the calculation results, the atmospheric pollutants are intelligently identified and tracked to obtain the detection results of the atmospheric pollutants.

5. The method for early warning of atmospheric pollutants based on multi-source heterogeneity according to claim 1 is characterized in that: The steps of using the method based on big data analysis to perform forecasting and early warning include: Based on the GIS platform, obtain the temporal and spatial dynamic change trend information of multi-source heterogeneous atmospheric pollutants; According to the temporal and spatial dynamic change trend information, an intelligent monitoring and reasoning decision model for air pollutant sources is established; According to the intelligent monitoring and reasoning decision model, the detection results of the atmospheric pollutants are analyzed to obtain forecast and warning results of the atmospheric pollutants.

6. The method for early warning of atmospheric pollutants based on multi-source heterogeneity according to claim 5 is characterized in that: The step of establishing an intelligent monitoring and reasoning decision model for air pollutant sources based on the temporal and spatial dynamic change trend information includes: According to the pollution cloud map simulated by numerical simulation, pollutants at a specified time and place can be queried to trace the source of pollutants; Calculate the proportion of pollution caused by a designated pollution source to a designated location and obtain the pollution degree equation; When the same location is affected by multiple pollution sources, the weights of different pollution sources are determined according to the ratio of pollution degrees; Determine the control parameters of air pollutants by determining the allowable pollution concentration, the allowable emission concentration of each pollution source and the emission reduction; Calculate the emission concentration of each pollution source based on real-time monitoring data, combine it with the allowable emission concentration of the pollution source, form an early warning model, and determine the early warning threshold based on the measured value of the early warning indicator. When the corresponding threshold is exceeded, take corresponding measures, which include increasing the monitoring frequency, special inspections, and comprehensive assessments; Based on the pollution degree equation, the pollution source weight, the control parameters, the early warning mode and the threshold setting information, an intelligent monitoring and reasoning decision model for the source of air pollutants is established.

7. The method for early warning of atmospheric pollutants based on multi-source heterogeneity according to claim 1 is characterized in that: The step of displaying the forecast and warning results of air pollutants through the artificial intelligence GIS platform includes: Through a lightweight and scalable GIS map platform, the forecast and warning results of the atmospheric pollutants are integrated with the GIS map to form a pollutant concentration distribution map.

8. An atmospheric pollutant early warning system based on multi-source heterogeneity, characterized in that: include: A data acquisition module is used to obtain multi-source heterogeneous atmospheric pollutant data, and to sort the multi-source heterogeneous atmospheric pollutant data to obtain sorted atmospheric pollutant data; A data processing module is used to uniformly characterize the collated air pollutant data to obtain uniformly characterized air pollutant data, and to perform intelligent detection using a grid-based edge computing method based on the uniformly characterized air pollutant data to obtain detection results of air pollutants; A data analysis module is used to perform forecast and warning according to the detection results of the atmospheric pollutants by using a method based on big data analysis to obtain forecast and warning results of atmospheric pollutants; The information display module is used to display the forecast and warning results of the atmospheric pollutants through an artificial intelligence GIS platform.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the multi-source heterogeneous atmospheric pollutant early warning method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-source heterogeneous atmospheric pollutant early warning method described in any one of claims 1 to 7 are implemented.

Images