DeepSeek-based emergency treatment system and method for atmospheric pollution source influence

Abstract

The invention relates to an emergency processing system and method for atmospheric pollution source influence based on DeepSeek, and belongs to the technical field of environmental monitoring, and the method comprises the steps: collecting atmospheric pollutant concentration data, meteorological data and basic information data of surrounding buildings; preprocessing the data; carrying out deep analysis on the preprocessed data based on a DeepSeek model, and judging whether to trigger early warning or not and the early warning degree in combination with a pollutant exceeding standard knowledge base; when abnormal data is found, the constructed knowledge base is retrieved; combining the data retrieved by the knowledge base; the current atmospheric pollution and building influence conditions are retrieved and compared with existing plan cases; and a detailed alarm condition is output, and a targeted emergency plan is generated at the same time. According to the method, based on the DeepSeek data analysis and learning ability, the influence of multiple factors is fully considered, the position of the influenced building is accurately identified, the influence degree is evaluated, and a more reliable basis is provided for environment decision making.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring technology, and in particular to an emergency response system and method for the impact of air pollution sources based on DeepSeek. Background Technology

[0002] With rapid industrialization and urbanization, air pollution has become an increasingly serious problem, causing numerous adverse effects on the health of people living in surrounding buildings and on the buildings themselves. Accurately assessing the impact of air pollution sources on surrounding buildings is crucial for urban planning, environmental protection, and public health safeguards.

[0003] Traditional methods for analyzing the impact of air pollution have many limitations, such as low efficiency when processing massive amounts of environmental data and the inability to conduct timely and effective dynamic assessments and predictions of the impact of air pollution.

[0004] In recent years, the application of artificial intelligence (AI) technology in the environmental field has gradually increased. However, existing AI analysis methods still have shortcomings in terms of model accuracy, adaptability, and intelligence. Faced with complex air pollution scenarios, it is difficult to quickly and accurately analyze which buildings are affected by air pollution and to what extent. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, this invention provides an emergency response system and method for the impact of air pollution sources based on DeepSeek. By fully leveraging the technological advantages of DeepSeek and integrating multi-source data, it achieves accurate and efficient assessment of the impact of air pollution, identifies the location of affected buildings, and provides scientific and reliable solutions for environmental management, urban planning, and other decision-making.

[0006] Emergency response methods for the impact of air pollution sources based on DeepSeek include:

[0007] Step 1: Collect data on atmospheric pollutant concentrations, meteorological data, and basic information about surrounding buildings;

[0008] Step 2: Preprocess the data, including cleaning, transformation, and standardization.

[0009] Step 3: Based on the DeepSeek model, perform in-depth analysis on the preprocessed data, combine it with the knowledge base of pollutant exceedance standards, determine whether there is an anomaly, and trigger an early warning and determine the level of the early warning;

[0010] Step 4: When abnormal data is detected, the constructed knowledge base is retrieved. The constructed knowledge base includes an enterprise information knowledge base, a spatiotemporal characteristic knowledge base, and a material data knowledge base to determine the responsible parties for air pollutants, their distribution characteristics in time and space, and material data related to buildings.

[0011] Step 5: Combine the data retrieved from the knowledge base above;

[0012] Step 6: Compare the current air pollution and building impact situation with existing contingency plans and cases to provide a reference for the response measures to this air pollution situation;

[0013] Step 7: Output detailed alarm information, including pollution type, affected area and degree of impact, and generate targeted emergency plans to guide relevant departments and personnel to take effective measures to deal with the impact of air pollution sources on surrounding buildings.

[0014] Furthermore, the atmospheric pollutant concentration data includes sulfur dioxide, nitrogen oxides, and particulate matter; the meteorological data includes wind direction, wind speed, temperature, air pressure, and humidity; and the basic building information data includes building material, height, orientation, and geographical location.

[0015] Furthermore, step 2 specifically includes

[0016] Step 201: Use data cleaning algorithms to remove noise and outliers from the data. Use deep learning-based interpolation algorithms to predict and supplement missing data points based on the spatiotemporal correlation of the data and historical data features.

[0017] Step 202: Convert data from different formats and sources into feature vectors;

[0018] Step 203: Standardize the data to normalize all types of data to the same scale range.

[0019] Furthermore, the process of constructing the enterprise information knowledge base in step 4 includes:

[0020] First, multi-source data integration and knowledge graph construction are carried out. The data includes the integration of the environmental protection department's enterprise pollutant discharge permit database, environmental impact assessment reports, online monitoring data, satellite remote sensing data, and news and public opinion information. The DeepSeek model's named entity recognition technology is used to automatically extract enterprise names, geographical locations, production process types, major pollutant emission inventories, and emission data. With enterprises as the central nodes, a knowledge graph of the relationship between enterprises, production processes, pollutant emissions, and environmental impact is constructed through relationship extraction technology.

[0021] Finally, based on knowledge graphs and real-time monitoring data, information on potential air pollution emitting enterprises in the surrounding area is retrieved, including enterprise names, locations, production processes, types and amounts of pollutants emitted, and the relationship between enterprise emissions and air pollution and building impacts is analyzed to identify the main entities responsible for pollution.

[0022] Furthermore, the retrieval process of the spatiotemporal feature knowledge base is as follows: retrieve the distribution characteristics of air pollutants in time and space dimensions, analyze the pollution diffusion patterns in different time periods and spatial locations, as well as the spatiotemporal differences in the impact on surrounding buildings, combine meteorological data with a Gaussian diffusion model to predict the diffusion paths of pollutants under different spatiotemporal conditions, and clarify the spatiotemporal differences in the impact of air pollutants on surrounding buildings.

[0023] Furthermore, the material data knowledge base compiles collected construction-related material data into an Excel format and inputs it into the knowledge base. The material data includes performance parameters of various construction protection materials and equipment, providing basic data support for subsequent response measures.

[0024] Furthermore, during the construction of the knowledge base, the data is collected from third-party sources. After cleaning, word segmentation, and annotation, the text data is converted into vector form for storage. An index is established for fast retrieval and processing. Based on user feedback, the knowledge base structure is optimized, and the data is updated regularly to maintain the timeliness of the knowledge.

[0025] The DeepSeek-based emergency response system for atmospheric pollution source impacts includes a monitoring data acquisition unit, a data preprocessing unit, a data entry knowledge base unit, an atmospheric pollution source tracing unit, a pollutant concentration prediction and analysis unit, an anomaly detection and early warning unit, an affected building identification and risk assessment unit, and a results output unit.

[0026] The monitoring data acquisition unit collects atmospheric pollutant concentration data through sensors, obtains real-time meteorological data through meteorological monitoring stations, and obtains the location information of surrounding buildings through a geographic information system.

[0027] The data preprocessing unit cleans, transforms, and standardizes the collected data.

[0028] The data entry knowledge base unit collects enterprise and factory information, emergency plans, and material data related to the topic from third-party sources. After cleaning, word segmentation, and annotation, the text data is converted into vector form, and the vectorized data is stored in the database and an index is created.

[0029] The atmospheric pollution source tracing unit is based on DeepSeek, which combines atmospheric diffusion theory and Gaussian diffusion model to construct an atmospheric pollution diffusion model suitable for complex environments. By deeply mining and learning a large amount of historical meteorological data, pollution source emission data and pollutant concentration monitoring data, it automatically extracts key features and patterns from the data, thereby optimizing the parameter settings and calculation methods of traditional diffusion models.

[0030] The pollutant concentration prediction and analysis unit inputs the real-time collected pollution source information and meteorological data into the constructed atmospheric pollution diffusion model to predict the distribution of atmospheric pollutant concentrations at different times and locations in real time; by analyzing the prediction results, it draws a spatiotemporal variation map of pollutant concentrations; at the same time, it combines the location information of surrounding buildings to analyze the pollutant concentration at different building locations and assess the potential impact of air pollution on each building.

[0031] The affected building identification and risk assessment unit uses DeepSeek to identify and assess the risks of affected buildings based on the pollutant concentration prediction results and the sensitivity of surrounding buildings.

[0032] The output unit outputs pollutant concentration prediction data, the location of affected buildings, and risk assessment results in various intuitive formats.

[0033] Furthermore, the sensitivity of surrounding buildings is assessed by establishing a building sensitivity assessment index system, taking into account the building's function, the length of time people stay in the building, and the building's ventilation system. Each type of building is assigned a corresponding sensitivity weight, and by combining pollutant concentration with building sensitivity, the surrounding buildings are classified into risk levels.

[0034] Furthermore, the output results from the result output unit include a detailed analysis report, which contains various data charts, text descriptions, and conclusions and recommendations.

[0035] The beneficial effects of this invention are:

[0036] 1. Significantly Improved Accuracy: Leveraging DeepSeek's powerful data analysis and learning capabilities, this invention fully considers complex meteorological conditions, diverse pollution sources, and various factors such as surrounding buildings and topography. Compared to traditional methods, it can more accurately predict the diffusion paths and concentration distribution of air pollutants, accurately identify the location of affected buildings, and assess the degree of impact, providing a more reliable basis for environmental decision-making.

[0037] 2. Significantly improved efficiency: DeepSeek's rapid computing and parallel processing capabilities make data processing and analysis more efficient. It can complete the processing of large amounts of data and the calculation of complex models in a short time, enabling real-time monitoring and dynamic assessment of the impact of air pollution, and providing timely support for environmental management and emergency response.

[0038] 3. Visualization and Decision Support: Through intuitive visualization, complex information on the impact of air pollution is presented to users in a clear and easy-to-understand way. Whether it is professional environmental managers or urban planners, they can quickly understand the current status and trends of air pollution, as well as its impact on surrounding buildings. This enables them to formulate environmental protection policies more scientifically, make urban planning and building site selection decisions, and effectively reduce the negative impact of air pollution on society and the environment. Attached Figure Description

[0039] Figure 1 This is a system structure diagram of the present invention.

[0040] Figure 2 This is a flowchart of the method of the present invention.

[0041] Figure 3 This is an example diagram of the knowledge graph of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Example 1: Refer to Figure 1 This invention provides an emergency response system for the impact of air pollution sources based on DeepSeek, including a monitoring data acquisition unit, a data preprocessing unit, a data entry knowledge base unit, an air pollution source tracing unit, a pollutant concentration prediction and analysis unit, an anomaly detection and early warning unit, an affected building identification and risk assessment unit, and a result output unit.

[0044] The monitoring data acquisition unit utilizes high-precision sensors to collect real-time data on the concentration of air pollutants, including common pollutants such as sulfur dioxide, nitrogen oxides, and particulate matter. It also leverages meteorological monitoring stations to obtain real-time meteorological data, such as wind direction, wind speed, temperature, air pressure, and humidity; these meteorological factors play a crucial role in the dispersion of air pollutants. Simultaneously, it obtains detailed location information of surrounding buildings through a Geographic Information System (GIS), including precise coordinates and other parameters for various types of buildings such as hospitals, factories, residential areas, and schools.

[0045] The data preprocessing unit cleans, transforms, and standardizes the collected raw data. First, data cleaning algorithms remove noise and outliers, such as identifying and correcting abnormal data caused by sensor malfunctions or interference by setting reasonable data ranges and variation thresholds. For missing values, deep learning-based interpolation algorithms are used to fill in the gaps, leveraging the spatiotemporal correlation and historical data characteristics to predict and supplement missing data points. Then, data from different formats and sources are uniformly converted to a format suitable for analysis, such as converting text-based building information into numerical feature vectors. Finally, the data is standardized to normalize various data types to the same scale range, improving the training efficiency and accuracy of subsequent data analysis models.

[0046] Data entry into the knowledge base unit (including enterprise and factory information, emergency plans, and material data): Collect topic-related data from various sources (such as documents, web pages, and databases). Clean the data, removing noise and duplicate content, and perform word segmentation and annotation. Convert text data into vector form for easy retrieval and processing. Store the vectorized data in the database and create an index for fast retrieval. Optimize the knowledge base structure based on user feedback and regularly update the data to maintain the timeliness of the knowledge.

[0047] Air Pollution Source Tracing Unit: Based on DeepSeek, and combining classic atmospheric diffusion theory and Gaussian diffusion models, an atmospheric pollution diffusion model suitable for complex environments is constructed. Leveraging DeepSeek's powerful learning capabilities, it performs in-depth mining and learning on a large amount of historical meteorological data, pollution source emission data, and pollutant concentration monitoring data, automatically extracting key features and patterns from the data, thereby optimizing the parameter settings and calculation methods of traditional diffusion models.

[0048] The pollutant concentration prediction and analysis unit inputs real-time collected pollution source information and meteorological data into a pre-constructed atmospheric pollution diffusion model. Utilizing DeepSeek's rapid computational capabilities, it predicts the distribution of atmospheric pollutant concentrations at different times and locations in real time. By analyzing the prediction results, it generates spatiotemporal maps of pollutant concentration variations, visually demonstrating the diffusion process and concentration evolution trends of pollutants in the atmosphere. Simultaneously, by combining the location information of surrounding buildings, it analyzes the pollutant concentrations at different building locations, assessing the potential impact of air pollution on each building. For example, for buildings with high air quality requirements, such as hospitals, it focuses on whether the surrounding pollutant concentrations exceed relevant health standard thresholds, and the duration and frequency of these exceedances.

[0049] Affected Building Identification and Risk Assessment Unit: Based on the pollutant concentration prediction results and the sensitivity of surrounding buildings, DeepSeek is used to identify affected buildings and assess risks. An evaluation index system for building sensitivity is established, considering factors such as the usage function of the building (e.g., places with dense population and high sensitivity to air quality like hospitals and schools), the residence time of people in the building, and the ventilation system of the building. Corresponding sensitivity weights are assigned to each type of building. By combining pollutant concentrations with building sensitivity, risk grading is carried out for surrounding buildings to clarify which buildings are in high-risk areas and which are in medium- and low-risk areas, providing a basis for subsequent environmental management and the formulation of protection measures.

[0050] Result Output Unit: Output the results obtained from the air pollution analysis module, including pollutant concentration prediction data, the locations of affected buildings, and risk assessment results, in various forms. A detailed analysis report can be generated, which includes various data charts, text descriptions, and conclusion suggestions.

[0051] In another preferred embodiment, an emergency treatment method for the impact of air pollution sources based on DeepSeek is as Figure 2 shown, specifically as follows:

[0052] 1. Use various air monitoring devices, such as air quality monitoring stations and meteorological sensors, to collect real-time air pollutant concentrations (such as sulfur dioxide, nitrogen oxides, particulate matter, etc.), meteorological parameters (wind speed, wind direction, temperature, humidity, etc.), and basic information of surrounding buildings (building materials, height, orientation, etc.).

[0053] 2. Preprocess the data, including cleaning, conversion, and standardization;

[0054] Among them, when inputting, for some missing data (such as no wind speed data at a certain time), an interpolation algorithm based on deep learning is used for filling. Utilize the spatio-temporal correlation of the data and the characteristics of historical data to predict and supplement the missing data points; the specific interpolation calculation formula can be:

[0055]

[0056] Among them, (x0, x1) are the independent variables of known data points; (y0, y1) are the dependent variables corresponding to (x0, x1); x is the target independent variable for which interpolation is required, and x0 < x < x1; y is the estimated dependent variable corresponding to the target independent variable x;

[0057] 3. Based on the DeepSeek model, the preprocessed air pollution and building-related data are analyzed in depth. Combined with a knowledge base of pollutant exceedance standards, the system determines whether an early warning is triggered and the severity of the warning. The model compares the data with the current national and local air quality standards stored in the knowledge base (such as the concentration limits for each pollutant specified in the "Ambient Air Quality Standard" (GB3095-2012)). Taking sulfur dioxide as an example, if the knowledge base specifies a 24-hour average concentration limit of 150 μg / m³ for Class II areas... 3 When the DeepSeek model analyzes and finds that the 24-hour average concentration of sulfur dioxide at the current monitoring point exceeds the limit, it marks it as an abnormal data point and triggers an early warning.

[0058] 4. When abnormal data is detected, a knowledge base search is performed.

[0059] (I) Enterprise Information Knowledge Base: Information on potential air pollution emitting enterprises in the surrounding area is retrieved, including enterprise name, location, production process, types and amounts of pollutants emitted, etc. The relationship between enterprise emissions and air pollution and building impacts is analyzed to determine the main entities responsible for pollution. The specific retrieval process is as follows:

[0060] 1) Multi-source data integration and knowledge graph construction

[0061] Data sources: integration of environmental protection department enterprise discharge permit database, environmental impact assessment reports, online monitoring data (such as CEMS system), satellite remote sensing data, and news and public opinion information.

[0062] Knowledge Extraction: Utilizing the Named Entity Recognition (NER) technology of the DeepSeek model, automatically extract enterprise names, geographical locations (latitude and longitude coordinates), production process types (such as iron and steel smelting, chemical synthesis, etc.), major pollutant emission inventories (SO2, NOx, VOCs, etc.) and emission data.

[0063] Knowledge Graph Construction: Using enterprises as central nodes, a knowledge graph is constructed based on relation extraction techniques, focusing on "enterprise-production process-pollutant emissions-environmental impact." An example is shown below. Figure 3 As shown

[0064] 2) Pollution liability attribution algorithm

[0065] Develop a pollution source tracing algorithm based on knowledge graphs and real-time monitoring data:

[0066] Pollution contribution = f(emission intensity, distance attenuation, meteorological diffusion conditions)

[0067] Emission intensity: Standardized values ​​of pollutant emissions from enterprises;

[0068] Distance attenuation: Distance weight W calculated based on the Gaussian diffusion modeld The formula is:

[0069]

[0070] Where d is the distance from the enterprise to the monitoring point, and σ is the diffusion parameter;

[0071] Meteorological diffusion conditions: The impact of parameters such as wind speed, wind direction, and stability on the diffusion path of pollutants is analyzed using the DeepSeek model.

[0072] When pollutants in a certain area are detected to exceed the standard, the system automatically calculates the pollution contribution of each enterprise and ranks them to determine the main responsible parties.

[0073] (II) Spatiotemporal Feature Retrieval: Retrieve the distribution characteristics of air pollutants in time and space. Analyze the pollution diffusion patterns at different time periods (such as day and night, seasonal variations) and spatial locations (different regions, different altitudes), as well as the spatiotemporal differences in their impact on surrounding buildings.

[0074] (III) Material Data Retrieval: Retrieve material data related to construction, such as performance parameters and pollution resistance of building protective materials, to provide basic data support for subsequent response measures. Compile this data directly into an Excel file and input it into the knowledge base.

[0075] 5. Combine the data retrieved from the knowledge base mentioned above.

[0076] Enterprise Information Retrieval: The system automatically searches for potentially air-polluting enterprises in the surrounding area through environmental databases and enterprise pollution discharge permit information. It obtains information such as enterprise name, geographical location, production process, types and amounts of pollutants emitted. Using the DeepSeek model, it analyzes the correlation between enterprise emission data and air pollution monitoring data, and combines this with meteorological conditions to determine the extent of the impact of enterprise emissions on surrounding air pollution and buildings, thus identifying the main polluters responsible for pollution.

[0077] Spatiotemporal Feature Retrieval: In the temporal dimension, the system analyzes the variation patterns of pollutant concentrations at different times (e.g., day and night, seasons); in the spatial dimension, it uses a Geographic Information System (GIS) to divide the region into grids and retrieve the pollution distribution in each region and at different altitudes. Combining meteorological parameters such as wind speed and direction, a Gaussian diffusion model is used to predict the diffusion paths of pollutants under different spatiotemporal conditions, clarifying the spatiotemporal differences in their impact on surrounding buildings. The Gaussian diffusion model is a classic model for atmospheric pollution diffusion; its basic formula describes the concentration distribution of point-source emitted pollutants in the atmosphere.

[0078] Where C(x,y,z) is the pollutant concentration at spatial point (x,y,z); Q is the source strength (emissions per unit time); u is the average wind speed; sigma_y and sigma_z are the horizontal and vertical diffusion parameters, respectively; and H is the effective source height (actual height + lift height).

[0079] Material Data Retrieval: The system has a built-in database of building protection materials, containing performance parameters of various building protection materials (such as exterior wall coatings and sealing materials) and equipment (air purification devices), such as pollution resistance, corrosion resistance, and filtration efficiency. When an alert is triggered, it can quickly retrieve appropriate material data, providing basic data support for developing building protection and pollution response measures.

[0080] 6. Compare and search existing plans and cases: Compare the current situation of air pollution and building impact with existing plans and cases to provide a reference for the response measures to this air pollutant situation;

[0081] 7. Based on the DeepSeek model, the system integrates the analysis results from the previous layers and outputs detailed alarm information, including pollution type, scope of impact, and degree of impact. At the same time, it generates targeted emergency plans to guide relevant departments and personnel to take effective measures to deal with the impact of air pollution sources on surrounding buildings.

[0082] In the description of the embodiments of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "center," "top," "bottom," "top," "bottom," "inner," "outer," "inner side," and "outer side," etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. "Inner side" refers to the interior or enclosed area or space. "Outer perimeter" refers to the area surrounding a specific component or specific area.

[0083] In the description of embodiments of the present invention, the terms "first," "second," "third," and "fourth" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first," "second," "third," or "fourth" may explicitly or implicitly include one or more of that feature. In the description of the present invention, unless otherwise stated, "a plurality of" means two or more.

[0084] In the description of embodiments of the present invention, specific features, structures, materials or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0085] In the description of the embodiments of the present invention, it should be understood that "-" and "~" represent a range of two numerical values, and this range includes the endpoints. For example, "AB" represents a range greater than or equal to A and less than or equal to B. "A~B" represents a range greater than or equal to A and less than or equal to B.

[0086] In the description of embodiments of the present invention, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0087] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for emergency handling of atmospheric pollution source impact based on DeepSeek, characterized in that, Comprising Step 1: Collecting atmospheric pollutant concentration data, meteorological data, and basic information data of surrounding buildings; Step 2: Preprocessing the data, including cleaning, conversion, and standardization processing; Step 3: Deeply analyzing the preprocessed data based on the DeepSeek model, combining with the pollutant exceeding standard knowledge base, determining whether it is abnormal, and triggering the warning and warning degree; Step 4: When abnormal data is found, search the constructed knowledge base, including enterprise information knowledge base, spatio-temporal feature knowledge base, and material data knowledge base, to determine the atmospheric pollutant responsibility subject, the distribution characteristics in time and space dimensions, and the building-related material data; Step 5: Combine the data searched from the above knowledge base; Step 6: Search and compare the current atmospheric pollution and building impact with existing preplan cases to provide reference for the response measures for this atmospheric pollutant situation; Step 7: Output detailed alarm information, including pollution type, impact range, and impact degree, and generate targeted emergency plans to guide relevant departments and personnel to take effective measures to respond to the impact of atmospheric pollution sources on surrounding buildings.

2. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, The atmospheric pollutant concentration data includes sulfur dioxide, nitrogen oxides, and particulate matter; the meteorological data includes wind direction, wind speed, air temperature, air pressure, and humidity; and the building basic information data includes building material, height, orientation, and geographic location.

3. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, The step 2 specifically includes Step 201: Use data cleaning algorithm to remove noise and outliers in the data, and use deep learning interpolation algorithm to predict and supplement missing data points based on the spatio-temporal correlation and historical data characteristics of the data, Step 202: Convert data of different formats and sources into feature vectors; Step 203: Standardize the data by normalizing various types of data to the same scale range.

4. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, The construction process of the enterprise information knowledge base in step 4 includes: First, multi-source data integration and knowledge graph construction, data including integrated environmental protection department enterprise pollution discharge license database, environmental impact assessment report, online monitoring data, satellite remote sensing data and news public opinion information, using the named entity recognition technology of DeepSeek model to automatically extract enterprise name, geographic location, production process type, main pollutant emission list and emission data, taking enterprise as the center node, constructing the knowledge graph of enterprise-production process-pollutant emission-environmental impact relationship through relationship extraction technology; Finally, based on the knowledge graph and real-time monitoring data, search the information of surrounding possible atmospheric pollution emission enterprises, including enterprise name, location, production process, pollutant emission type and emission amount, analyze the relationship between enterprise emission and atmospheric pollution and building impact, and determine the main pollution responsibility subject.

5. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, The retrieval process of the spatio-temporal feature knowledge base is: retrieving the distribution characteristics of atmospheric pollutants in the time and space dimensions, analyzing the pollution diffusion law of different time periods and spatial positions, and the spatio-temporal difference of the influence on surrounding buildings, combining with the meteorological data using the Gaussian diffusion model, predicting the diffusion path of pollutants under different spatio-temporal conditions, and determining the spatio-temporal difference of the influence of atmospheric pollutants on surrounding buildings.

6. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, The material data knowledge base collects the collected building-related material data into excel format, enters the knowledge base, and the material data includes the performance parameters of various building protection materials and equipment, providing basic data support for subsequent response measures.

7. The DeepSeek-based emergency handling method for atmospheric pollution source impact according to claim 1, characterized in that, In the construction process of the knowledge base, the data is collected from third-party sources, and after cleaning, word segmentation and labeling, the text data is converted into vector form for storage, indexed for fast retrieval and processing, and the knowledge base structure is optimized according to usage feedback, and the data is updated regularly to maintain the timeliness of the knowledge.

8. An emergency handling system for atmospheric pollution source impact based on DeepSeek, characterized in that, It includes a monitoring data acquisition unit, a data preprocessing unit, a data entry knowledge base unit, an atmospheric pollution tracing unit, a pollutant concentration prediction and analysis unit, an affected building identification and risk assessment unit, and a result output unit; The monitoring data acquisition unit collects atmospheric pollutant concentration data through sensors, obtains real-time meteorological data through meteorological monitoring stations, and obtains location information of surrounding buildings through geographic information systems; The data preprocessing unit cleans, converts and standardizes the collected data; The data entry knowledge base unit collects information related to the theme from third-party sources, such as enterprise factory information, emergency plans and material data, and after cleaning, word segmentation and labeling, the text data is converted into vector form, and the vectorized data is stored in the database and indexed; The atmospheric pollution tracing unit takes DeepSeek as the core, combines atmospheric diffusion theory and Gaussian diffusion model, and constructs an atmospheric pollution diffusion model suitable for complex environments. Through deep mining and learning of a large amount of historical meteorological data, pollution source emission data and pollutant concentration monitoring data, the key features and rules in the data are automatically extracted, thereby optimizing the parameter settings and calculation methods of the traditional diffusion model; The pollutant concentration prediction and analysis unit inputs the real-time collected pollution source information and meteorological data into the constructed atmospheric pollution diffusion model to predict the atmospheric pollutant concentration distribution at different times and different locations in real time; through analysis of the prediction results, a pollutant concentration spatio-temporal variation diagram is drawn; at the same time, combined with the location information of surrounding buildings, the pollutant concentration at different building locations is analyzed, and the potential impact of atmospheric pollution on each building is evaluated; The affected building identification and risk assessment unit identifies and assesses the affected buildings using DeepSeek according to the pollutant concentration prediction results and the sensitivity of surrounding buildings; The result output unit outputs the pollutant concentration prediction data, the location of the affected building and the risk assessment results in various intuitive forms.

9. The DeepSeek-based emergency handling system for atmospheric pollution source impact according to claim 1, characterized in that, The sensitivity of the surrounding buildings is classified by establishing a building sensitivity evaluation index system, considering the use function of the building, the staying time of the personnel in the building and the ventilation system of the building, giving the corresponding sensitivity weight to each type of building, combining the pollutant concentration with the building sensitivity, thereby classifying the risk of the surrounding buildings.

10. The DeepSeek-based emergency handling system for atmospheric pollution source impact according to claim 1, characterized in that, The result output unit outputs the result including a detailed analysis report, and the report contains various data charts, textual descriptions and conclusions and suggestions.

Images