Non-point source pollution simulation method and non-point source pollution intelligent interaction method based on intelligent agent

Abstract

The invention relates to the technical field of non-point source pollution decision, and discloses a non-point source pollution simulation method and a non-point source pollution intelligent interaction method based on an agent, and the simulation method comprises the steps: extracting land and river attribute data of a target region, and constructing a nitrogen emission coefficient database; associating the land attribute type with a nitrogen emission coefficient library, calculating the plot emission amount through a reference area correction model nitrogen emission coefficient, converting the plot emission amount into emission amount grid data, and generating a nitrogen emission distribution map; converting the emission raster data into point data, calculating each pollution point and a river reach influence area, and constructing a river directed graph network; calculating a river reach absorption coefficient by adopting an exponential decay model, and calculating a pollution load transmission quantity and a river reach absorption quantity step by step along a river digraph network topology through breadth-first search by combining the river reach absorption coefficient based on the digraph network; the total consumption amount of the influence river reach of each pollution point is accumulated, the net discharge load distribution diagram and the consumption amount distribution diagram are generated, the calculation process is high in real-time performance, and the method is suitable for various application scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of non-point source pollution decision-making, and in particular to a non-point source pollution simulation and an intelligent agent-based non-point source pollution intelligent interaction method. Background Art

[0002] When simulating non-point source pollution, traditional methods usually use fixed values ​​for nitrogen emission coefficients, which cannot reflect the dynamic changes in land use characteristics or pollutant types in different regions. They are mostly designed for a single region or specific pollutants and are difficult to extend to different soil types or policy scenarios. When calculating along-the-line absorption, process models are often used. This model uses many parameters and generalizes the water bodies in the region into a large water body. It is difficult to consider the distribution impact of small water body sections. At the same time, it cannot distinguish land use types, nor does it consider the impact of landscape fragmentation. The calculation complexity is high and it is difficult to support real-time analysis. Summary of the Invention

[0003] In view of this, the present invention provides a non-point source pollution simulation and an intelligent agent-based non-point source pollution intelligent interaction method to address the shortcomings of the existing non-point source pollution simulation method, which has poor real-time performance and cannot adapt to different areas or pollutant types.

[0004] In a first aspect, the present invention provides a method for simulating non-point source pollution, the method comprising:

[0005] Extract the land parcel boundaries and attribute types of the target area's land use vector data and the river attributes of the river surface vector data, and build a nitrogen emission coefficient database to store the unit area nitrogen emission coefficients corresponding to different land attribute types;

[0006] The land attribute type of the target area is associated with the nitrogen emission coefficient database, and the basic nitrogen emission coefficient in the database is assigned to each land attribute type plot. The model nitrogen emission coefficient is corrected by the benchmark area, and the plot emission in the target area is calculated based on the corrected nitrogen emission coefficient. The emission is converted into emission raster data and a nitrogen emission distribution map is generated.

[0007] Convert the discharge raster data into point data, divide the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and construct a river directed graph network based on the river vector data;

[0008] The exponential decay model is used to calculate the river section absorption coefficient. Based on the directed graph network of river nodes and the river section absorption coefficient, the pollution load transmission and river section absorption capacity are calculated step by step along the river directed graph network topology through breadth-first search.

[0009] The total absorption capacity of each pollution point in the affected river section is accumulated, and the net emission load is obtained by subtracting the total absorption capacity from the land emission capacity and converted into raster data to generate the net emission load distribution map and the absorption capacity distribution map.

[0010] The non-point source pollution simulation method provided by the embodiment of the present invention extracts vector data such as land use and river boundaries and constructs a structured database to achieve rapid call and batch calculation of pollution parameters. It uses breadth-first search to traverse the river directed graph network, combined with an exponential decay model (such as rapid calculation of the absorption coefficient) to achieve hierarchical recursive calculation of pollution load transmission, avoiding the transmission path distortion caused by traditional "full-basin hybrid calculation"; the topological structure (node-edge model) constructed based on river vector data is compatible with any river network form, without the need to remodel for a specific basin, only the specific parameters need to be adjusted. It can be applied to scenarios such as basin non-point source pollution prevention and control, agricultural green development planning (fertilizer and pesticide reduction zoning), urban waterlogging and non-point source pollution coordinated management, etc., providing an efficient technical tool for "precise pollution control and scientific pollution control."

[0011] In an optional embodiment, the method of correcting the model nitrogen emission coefficient by reference area includes:

[0012] Based on the actual area of ​​the plot A current With unit area A base , the emission coefficient is corrected by the power exponential relationship:

[0013] C adj =C base ×(A current A base ) k

[0014] Among them, k is the empirical correction factor, which is used to reflect the nonlinear effect of plot area on emission intensity.

[0015] In contrast to the traditional method of assigning a uniform emission coefficient to the same land attribute type, the embodiment of the present invention ignores the actual emission intensity differences caused by the area differences between plots. The empirical correction factor k can be flexibly adjusted to be closer to the actual pollution load, which helps to improve the calculation accuracy.

[0016] In an optional embodiment, the calculation of the river section absorption coefficient using an exponential decay model includes:

[0017] η uv =(1-Ef) Luv

[0018] Where η uv is the corrected absorption coefficient, Ef is the absorption efficiency per kilometer of the river section, L uv is the actual length of the river section, which is obtained based on the river attributes of the river surface vector data.

[0019] The embodiment of the present invention realizes efficient modeling, accurate simulation and dynamic regulation of the river section absorption process through the exponential decay model, becoming a key technical link connecting "land emissions" and "river transmission" in non-point source pollution simulation.

[0020] In an optional embodiment, the method of calculating the pollution load transmission amount and the river section absorption capacity step by step along the river directed graph network topology based on the river node directed graph network combined with the river section absorption coefficient by breadth-first search includes:

[0021] The flow ratio of each branch of the river is obtained through river attributes to calculate the diversion ratio of pollutants;

[0022] Calculate the load increment transmitted from the current pollution point to the downstream:

[0023] Q new =Q ucurrent ×η uv ×α uv

[0024] Calculate the current river section's absorption capacity:

[0025] D uv =Q ucurrent ×(1-η uv )×α uv

[0026] Among them, Q ucurrent represents the real-time pollution load of the current node u, η uv Represents the self-purification efficiency coefficient of the river section (u→v); α uv Indicates the diversion ratio of the river section (u→v), Q new Indicates the pollutants flowing downstream from the current node, D uv Indicates the pollutants absorbed by the current node;

[0027] Through the river node relationship of the river node directed graph network, based on the above steps, the pollutant absorption and pollutant accumulation of each river section are calculated step by step through breadth-first search.

[0028] The embodiment of the present invention achieves topological fidelity, precise diversion, efficient calculation, and clear traceability of the pollution load transmission and absorption process through the combination of directed graph networks and breadth-first search, which significantly improves the scientific nature and decision-making support capabilities of non-point source pollution simulation.

[0029] In a second aspect, the present invention provides an agent-based intelligent interaction method for non-point source pollution, comprising:

[0030] Receive user text or multimodal input information;

[0031] Analyze user input information through a large language model and determine whether the user's demand type is normal conversation or non-point source pollution simulation demand;

[0032] When it is determined that the simulation needs are for non-point source pollution, the key parameters and target areas related to non-point source pollution in the user input information are extracted;

[0033] The non-point source pollution simulation agent obtained by the method of the first aspect is called through an interface, and based on the key parameters related to the non-point source pollution and the target area, at least one of the emissions, net emission load and absorption capacity of the target area is calculated;

[0034] The calculation results are converted into natural language to answer users, and vector data result graphs are generated in a visual way.

[0035] The agent-based intelligent interaction method for non-point source pollution provided in an embodiment of the present invention breaks the technical barriers of traditional environmental models by deeply integrating the semantic understanding ability of a large language model with the non-point source pollution simulation agent. It realizes the full-chain intelligence of "demand acquisition-model calculation-decision support" and provides a new generation of "convenient, accurate and efficient" tools for non-point source pollution control.

[0036] In a third aspect, the present invention provides a non-point source pollution simulation system, the system comprising:

[0037] The data preparation module is used to extract the land parcel boundaries and attribute types of the land use vector data of the target area and the river attributes of the river water surface vector data, and to build a nitrogen emission coefficient database to store the nitrogen emission coefficients per unit area corresponding to different land attribute types;

[0038] The nitrogen emission calculation module is used to associate the land attribute type of the target area with the nitrogen emission coefficient library, assign a basic nitrogen emission coefficient in the database to each land attribute type plot, correct the model nitrogen emission coefficient by the benchmark area, calculate the plot emission of the target area based on the corrected nitrogen emission coefficient, and convert it into emission raster data to generate a nitrogen emission distribution map;

[0039] The river section impact area and river directed graph construction module is used to convert the discharge raster data into point data, divide the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and construct a river directed graph network based on the river vector data;

[0040] The absorption and load calculation module is used to calculate the river section absorption coefficient using an exponential decay model. Based on the river node directed graph network combined with the river section absorption coefficient, the pollution load transmission volume and the river section absorption capacity are calculated step by step along the river directed graph network topology through breadth-first search;

[0041] The distribution map generation module is used to accumulate the total amount of absorption of the affected river section for each pollution point, subtract the total absorption amount from the land emission amount to obtain the net emission load and convert it into raster data to generate the net emission load distribution map and the absorption amount distribution map.

[0042] In a fourth aspect, the present invention provides an agent-based intelligent interactive system for non-point source pollution, comprising:

[0043] User information acquisition module, used to obtain user text or multimodal input information;

[0044] The demand analysis module is used to analyze user input information through a large language model and determine whether the user's demand type is a normal conversation or a non-point source pollution simulation demand;

[0045] The non-point source pollution simulation demand triggering module is used to extract key parameters and target areas related to non-point source pollution from the user input information when it is determined that there is a non-point source pollution simulation demand;

[0046] A non-point source pollution simulation module, configured to call a non-point source pollution simulation agent obtained by the method of any embodiment of the first aspect through an interface, and calculate at least one of the emissions, net emission load, and absorption capacity of a plot in the target area based on key parameters related to non-point source pollution and the target area input by the user;

[0047] The result feedback module is used to convert the calculation results into natural language to answer users and generate vector data result graphs in a visual way.

[0048] In a fifth aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, computer instructions stored in the memory, and the processor executing the computer instructions to thereby execute the non-point source pollution simulation method of the first aspect or any corresponding embodiment thereof or the non-point source pollution intelligent interaction method based on an intelligent agent as described in the second aspect or any corresponding embodiment thereof.

[0049] In the sixth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the non-point source pollution simulation method of the first aspect or any corresponding embodiment thereof, or the non-point source pollution intelligent interaction method based on an intelligent agent as described in the second aspect or any corresponding embodiment thereof.

[0050] In the seventh aspect, the present invention provides a computer program product comprising computer instructions, the computer instructions being used to enable a computer to execute the non-point source pollution simulation method of the first aspect or any corresponding embodiment thereof, or the non-point source pollution intelligent interaction method based on an intelligent agent as described in the second aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 is a flow chart of a method for simulating non-point source pollution according to an embodiment of the present invention;

[0053] Figure 2 1 is a flow chart of an intelligent interactive method for non-point source pollution based on an intelligent agent according to an embodiment of the present invention;

[0054] Figure 3 is a structural block diagram of a non-point source pollution simulation system according to an embodiment of the present invention;

[0055] Figure 4 is a structural block diagram of an intelligent interactive system for non-point source pollution based on an agent according to an embodiment of the present invention;

[0056] Figure 5 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0058] In this embodiment, a method for simulating non-point source pollution is provided. Figure 1 : is a flowchart of non-point source pollution simulation according to an embodiment of the present invention. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer device such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here. Figure 1 As shown, the process includes the following steps:

[0059] Step S101 , extracting the land parcel boundaries and attribute types of the land use vector data of the target area and the river attributes of the river surface vector data, and constructing a nitrogen emission coefficient database to store the nitrogen emission coefficients per unit area corresponding to different land attribute types.

[0060] Specifically, by obtaining a current land use map, all land boundaries and attributes (such as land type and area) are extracted; by obtaining a river surface map, the attributes of all river boundaries in the region (such as absorption coefficient, flow rate, and flow direction) are extracted. A database containing unit area nitrogen emission coefficients (such as kg / ha·yr) corresponding to different land use types is constructed. By integrating land use vector data with river surface vector data, a full chain data linkage of "pollution source-transmission path-absorption carrier" is achieved. The mechanism for constructing a nitrogen emission coefficient library can be directly transferred to other pollutants (such as phosphorus and pesticides) by simply replacing the unit area emission coefficient in the database without modifying the model logic.

[0061] Step S102: associate the land attribute type of the target area with the nitrogen emission coefficient library, assign a basic nitrogen emission coefficient in the database to each land attribute type plot, correct the model nitrogen emission coefficient by the benchmark area, calculate the plot emission of the target area according to the corrected nitrogen emission coefficient, and convert it into emission raster data to generate a nitrogen emission distribution map.

[0062] Specifically, the traditional method uses a uniform emission coefficient for the same type of land (e.g., all farmland uses the same nitrogen emission coefficient), ignoring the variation in emission intensity caused by differences in plot area. In the embodiment of the present invention, the nitrogen emission coefficient of the model is corrected by the reference area:

[0063] Based on the actual area of ​​the plot A current With unit area A base , the emission coefficient C is corrected by the power exponential relationship adj :

[0064] C adj =C base ×(A current A base ) k

[0065] Among them, C base is the basic nitrogen emission coefficient, and k is the empirical correction factor (default k = 0.3), which is used to reflect the nonlinear effect of plot area on emission intensity.

[0066] Furthermore, the emission per unit plot E is calculated based on the revised emission coefficient. total :E total =C adj ×A current .

[0067] Based on this, the present invention converts the emission data from vector plots into a spatially continuous, high-resolution raster (e.g., TIFF format). The raster pixel values ​​represent the nitrogen emission intensity per unit area (kg / ha). Area-weighted or kriging interpolation algorithms are also used to ensure a smooth transition in emission gradients at the boundaries of the raster plots, avoiding sudden changes in emission intensity due to differences in properties between adjacent plots (e.g., jumps in raster values ​​at the border between cultivated land and forest land).

[0068] By adjusting the unit area Abase and the empirical correction factor k, the present invention can adapt to land use patterns in different regions. For example, in plain agricultural areas, where plots are large, Abase is set to 50 hectares and k = 0.2 (emission intensity decreases slowly with increasing area); in hilly and mountainous areas, where plots are fragmented, Abase is set to 2 hectares and k = 0.5 (emission intensity increases sharply with decreasing area). Compared to traditional models, this model can be extended from the North China Plain to the Jiangnan hilly region simply by adjusting parameters, without requiring new algorithm development, reducing application costs.

[0069] Step S103: converting the discharge raster data into point data, dividing the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and constructing a river directed graph network based on the river vector data;

[0070] Specifically, the embodiment of the present invention converts raster data (continuous space) into point data (discrete pollution sources), and then realizes the spatial mapping of "pollution source-absorbing water body" by calculating the shortest Euclidean distance from each pollution point to each river section in the region. This clarifies the direct impact range of each river section and avoids errors caused by the "rough division" of traditional watershed scales (such as classifying mountainous areas far away from rivers into the impact area). When a river is diverted or a new river section is added, it is only necessary to recalculate the distance to adjust the impact area without reconstructing the model. The affected river section of each pollution point is uniquely linked to the nearest river section through the Euclidean distance, avoiding confusion in the calculation of the absorption capacity caused by the sharing of pollution loads by multiple river sections, and ensuring the spatial correspondence between the absorption capacity and the load capacity (such as the absorption capacity of a certain river section only comes from the pollution points within its influence area).

[0071] The embodiment of the present invention extracts nodes (starting and ending points of river sections) and directed edges (flow direction of river sections) based on river vector data, and constructs a directed graph network of river networks (such as tree-like river networks and ring-like river networks) to achieve accurate modeling of water flow direction and tributary confluence relationships. Weights are assigned to directed graph edges in combination with river attributes (flow rate, flow direction) (such as river sections with large flow rates have stronger transmission capabilities), so that the model can simulate differences in pollution diffusion under different hydrological conditions. The directed graph network records the upstream and downstream relationships between river sections (such as river section A → river section B → river section C). When pollution exceeds the standard at the downstream monitoring point, the potential pollution source river section and the corresponding plot can be quickly traced along the reverse topological direction (C → B → A).

[0072] This step provides a high-precision and high-efficiency underlying framework for the transmission simulation of non-point source pollution through the technical combination of "spatial point-line association + topological network construction". It is the core foundation for realizing the integration of "pollution source positioning-transmission path tracking-absorption capacity assessment".

[0073] In step S104, the exponential decay model is used to calculate the river section absorption coefficient, and based on the river node directed graph network combined with the river section absorption coefficient, the pollution load transmission amount and the river section absorption amount are calculated step by step along the river directed graph network topology through breadth-first search.

[0074] Specifically, the exponential decay model is used to calculate the river section absorption coefficient, including:

[0075] ηuv=(1-Ef) Luv

[0076] Where η uv is the corrected absorption coefficient, Ef is the absorption efficiency per kilometer of the river section, L uv is the actual length of the river section, derived from the river attributes of the river surface vector data. This formula is based on the exponential decay of pollutants with river section length and conforms to the actual mechanism of river self-purification.

[0077] This embodiment of the present invention leverages the node-edge relationships of a directed graph network to transfer pollution loads from upstream to downstream, taking into account the current river section's absorption coefficient and diversion weight at each level. This enables a chained simulation of the "transmission-absorption" process. A breadth-first search algorithm, starting from the river's source node and accessing downstream nodes layer by layer, ensures that each river section is calculated only once, avoiding duplicate processing (such as the loop path problem in a ring river network).

[0078] The specific process includes:

[0079] Step a1: Calculate the diversion ratio of pollutants by obtaining the flow ratio of each branch of the river through river attributes.

[0080] This embodiment of the present invention distributes pollution loads based on the flow ratio of river branches, simulating the diversion phenomenon of real river networks (for example, the ratio of the flow of the main stream to its tributaries determines the pollution diversion ratio). In one application scenario, in a multi-branch river network in a delta, if a river section accounts for 70% of the flow, it will receive 70% of the pollution load from upstream, which is more consistent with the laws of hydrodynamics than the "equally proportional diversion" assumption.

[0081] Step a2: Calculate the load increment transmitted from the current pollution point to the downstream:

[0082] Q new =Q ucurrent ×η uv ×α uv

[0083] Calculate the current river section's absorption capacity:

[0084] D uv =Q ucurrent ×(1-η uv )×α uv

[0085] Among them, Q ucurrent represents the real-time pollution load of the current node u, η uv Represents the self-purification efficiency coefficient of the river section (u→v); α uv Indicates the diversion ratio of the river section (u→v), Q new Indicates the pollutants flowing downstream from the current node, D uv Indicates the pollutants absorbed by the current node;

[0086] Step a3, through the river node relationship of the river node directed graph network, based on the above steps a1-a2, the pollutant absorption and pollutant accumulation of each river section are calculated step by step through breadth-first search.

[0087] The embodiment of the present invention can simultaneously generate a transmission load contribution matrix (recording the pollution contribution of river section u to river section v) and an ecological absorption matrix (recording the self-purification amount of river section u) through a step-by-step calculation process, providing a quantitative basis for pollution tracing and ecological assessment.

[0088] Step S105: For each pollution point, the total amount of absorption in the affected river section is accumulated, the land discharge amount minus the total absorption amount is used to obtain the net discharge load and convert it into raster data to generate a net discharge load distribution map and an absorption amount distribution map.

[0089] Specifically, the embodiment of the present invention can intuitively reflect the proportion of pollutants discharged from different plots that are actually absorbed by the river ecosystem during the transmission process by accumulating the total amount of absorption capacity of the river section corresponding to each pollution point. For example, the plot close to the downstream of the river may have a higher net emission due to the short absorption path, while the upstream plot may have a lower net emission due to the long-distance absorption, thus realizing a refined conversion from total emission to actual effective load. The net emission load and absorption capacity are inverted to a spatial grid and a thematic distribution map (such as a heat map, a graded color map) is generated, which can intuitively present the spatial aggregation characteristics of the pollution load (such as high-load areas concentrated in farmland near the city) and the spatial differences in the river absorption capacity (such as the absorption capacity of wide and shallow river sections in the middle and upper reaches is higher). This type of visualization result makes it easy for managers to quickly locate key areas for governance (such as high-value net emission areas) and formulate zoning management and control strategies.

[0090] The embodiment of the present invention realizes the full-process quantification of non-point source pollution from source analysis to process simulation and then to effect evaluation through the closed loop of "pollution load accounting-absorption process simulation-spatial visualization". It not only provides precise spatial targeting for pollution control, but also provides a scientific and intuitive decision-making basis for ecological protection and restoration, and coordinated watershed management by revealing the dynamic relationship between emissions and absorption.

[0091] The non-point source pollution simulation method provided by the embodiment of the present invention extracts vector data such as land use and river boundaries and constructs a structured database to achieve rapid call and batch calculation of pollution parameters. It uses breadth-first search to traverse the river directed graph network, combined with an exponential decay model (such as rapid calculation of the absorption coefficient) to achieve hierarchical recursive calculation of pollution load transmission, avoiding the transmission path distortion caused by traditional "full-basin hybrid calculation"; the topological structure (node-edge model) constructed based on river vector data is compatible with any river network form and does not require remodeling for specific basins. It can be applied to scenarios such as basin non-point source pollution prevention and control, agricultural green development planning (fertilizer and pesticide reduction zoning), and coordinated management of urban waterlogging and non-point source pollution, providing an efficient technical tool for "precise pollution control and scientific pollution control."

[0092] Traditional non-point source pollution decision-making systems rely on professional software and fixed parameter inputs. Users need to have professional knowledge, which makes the software system interaction dependent on professional operations. In addition, the system mostly adopts the "preset scene-fixed output" mode, which is difficult to adapt to the user's personalized and refined needs. This makes the interaction threshold of the existing non-point source pollution system high and difficult to adapt to the flexible deduction of complex scenarios. Based on this, the embodiment of the present invention also provides an intelligent interaction method for non-point source pollution based on intelligent agents, which combines a large language model with a non-point source pollution model to make the decision-making system more intelligent, wherein the non-point source pollution model adopts Figure 1 The non-point source pollution simulation method shown in the figure can enable users to initiate complex environmental decision-making requests in everyday language and support multi-modal input and multi-modal output. Figure 2 As shown, the following steps are included:

[0093] Step S201: receiving user text or multimodal input information.

[0094] Specifically, integrated speech recognition technology (such as ASR) can be used to support users to input requirements through voice commands, such as "calculate the amount of agricultural non-point source pollution absorbed in the Yangtze River Basin in 2023", which can be used to further optimize speech semantic analysis and realize the recognition of fuzzy expressions in natural language, such as "help me see the pollution situation of upstream farmland". At the same time, users are allowed to upload satellite remote sensing images, drone aerial photos or on-site photos. The system automatically extracts land use types, river boundaries and other information through image recognition technology (such as deep learning semantic segmentation model), supplements or corrects vector data, such as identifying newly added farmland areas or river diversions, and supports users to import common GIS data formats (such as Shapefile, GeoJSON, KML), directly delineate analysis areas or modify basic geographic features, such as users uploading customized watershed boundary data to quickly carry out local area pollution assessment.

[0095] User-entered data (such as custom parameters) is formatted and logically validated. For example, if a user enters a negative emission coefficient, the system automatically displays an error message and suggests a reasonable range. Ambiguous input is intelligently completed, so if a user enters "last year's data," it will be interpreted as "2024 data." The system also automatically completes missing parameters based on historical conversation history and sets intelligent defaults for unclear parameters, streamlining user operations.

[0096] Step S202 : Analyze the user input information through a large language model and determine whether the user demand type is a normal conversation or a non-point source pollution simulation demand.

[0097] Specifically, the large language models in the embodiments of the present invention can use the relatively mature large language model API. The large language model can understand the fuzzy expressions, contextual associations and domain terms in the user input. For example, if the user inputs "help me see the pollution situation of farmland in the upper reaches of the Yangtze River", the model can automatically extract the keywords "upper reaches of the Yangtze River", "farmland" and "pollution", triggering the non-point source pollution simulation process without the user having to manually select parameters or areas. The demand types of traditional systems need to be defined in advance (such as fixed "calculation", "query" and "visualization" buttons), while the large speech model can continuously identify new demand patterns through learning. For example, if the user inputs "simulate the impact of climate change on non-point source pollution in the next ten years", the model can automatically classify it as a "scenario simulation" demand, triggering parameter adjustment and long-term prediction modules without manual code updates.

[0098] The embodiment of the present invention realizes user intention recognition by pre-training a large language model in advance, such as "You are a non-point source decision-making intelligent agent. When a user asks questions about wetland site selection or planting structure adjustment, the questions raised by the user need to be standardized and organized into the following format for answering: 'Function: ***, Region: **, Wetland efficiency: **, Emission reduction target: **'. For example, if a user asks 'I want to reduce emissions by 20% in a certain area, and the wetland efficiency is 10%, how should I plan the wetland?', the answer is 'Function: Wetland site selection, Region: a certain area, Wetland efficiency: 10,, Emission reduction target: 20'; if the user does not mention the emission reduction target and wetland efficiency, the answer is 10. If the region is not specified, the historical records will be searched and the region where the user asked the question last time will be used. For example, if the user asked the question last time, "I want to reduce emissions by 20% in a certain area, and the wetland efficiency is 10%. How should I plan the wetland?" and the question this time is "How should the wetland efficiency be increased by 10%?", the user's answer should be "Function: Wetland site selection, Region: a certain area, Wetland efficiency: 20, Emission reduction target: 20." If the user does not ask questions about wetland site selection or planting structure adjustment, the answer is "Ordinary dialogue." This pre-training method can standardize the user's needs and process them into a form that is easy for the program to recognize, which is convenient for subsequent non-point source pollution simulation intelligent agent calls.

[0099] Step S203 : When it is determined that the simulation demand is for non-point source pollution, key parameters and target areas related to non-point source pollution are extracted from the user input information.

[0100] Specifically, the large speech model can automatically identify the core calculation parameters in the user input, such as:

[0101] Spatial parameters: target area (e.g., "Yangtze River Delta," "a certain county or town"), analysis scope (e.g., "watershed scale," "village scale");

[0102] Model parameters: emission coefficient type (such as "dryland nitrogen emission coefficient" and "paddy field nitrogen emission coefficient"), correction factor (such as benchmark area, empirical coefficient).

[0103] It can identify necessary parameters that are not explicitly entered by the user and automatically complete them based on default values ​​or historical data. For example, when the user does not specify the empirical correction factor k, the default value of 0.3 is automatically adopted; when the unit absorption efficiency of the river is not provided, the preset parameter library is called according to the regional type (such as urban river section, natural river section).

[0104] Step S204, the non-point source pollution simulation intelligent entity obtained by the non-point source pollution simulation method in steps S101-SS104 called by the interface calculates at least one of the emission amount, net emission load and absorption amount of the target area based on the key parameters related to non-point source pollution and the target area.

[0105] Specifically, the embodiment of the present invention separates the front-end interactive system from the back-end computing model through an interface call mode, allowing both to be upgraded and iterated independently: the interactive layer can independently optimize the semantic parsing capabilities of the large language model (such as supporting dialects or terminology) without modifying the computing model code; the model layer can independently update the algorithm (such as introducing a more advanced emission coefficient correction model) and is backward compatible with the old version of the interactive system through the interface. For example, an environmental protection department upgraded its computing model to version 3.0 (with a new phosphorus pollution calculation module) by simply updating the interface parameter definition, and the front-end interactive interface did not need to be redeveloped. This not only retains the interactive flexibility of the large language model, but also ensures the professionalism and efficiency of non-point source pollution simulation.

[0106] Step S205 , converting the calculation result into natural language to answer the user, and generating a vector data result graph in a visual manner.

[0107] The embodiment of the present invention uses natural language generation (NLG) technology to convert professional environmental model calculation results (such as pollution load values ​​and absorption capacity) into easy-to-understand sentence descriptions, allowing non-professional users (such as government managers and the public) to intuitively understand complex pollution analysis results. At the same time, it automatically generates conclusive descriptions containing key data, allowing users to understand the results more quickly. Combined with a visualization engine, vector result maps (such as nitrogen emission distribution maps, net emission load distribution maps, and absorption capacity distribution maps) are dynamically generated for intuitive display. The collaborative display of natural language answers and visual graphics integrates the output format of both pictures and texts in a way that conforms to human cognitive habits.

[0108] The agent-based intelligent interaction method for non-point source pollution provided in an embodiment of the present invention breaks the technical barriers of traditional environmental models by deeply integrating the semantic understanding ability of a large language model with the non-point source pollution simulation agent. It realizes the full-chain intelligence of "demand acquisition-model calculation-decision support" and provides a new generation of "convenient, accurate and efficient" tools for non-point source pollution control.

[0109] In this embodiment, a non-point source pollution simulation system is also provided. This system is used to implement the above-mentioned embodiments and preferred embodiments, and the details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0110] This embodiment provides a non-point source pollution simulation system, such as Figure 3 Shown, including:

[0111] Data preparation module 301 is used to extract the land parcel boundaries and attribute types of the land use vector data of the target area and the river attributes of the river surface vector data, and to build a nitrogen emission coefficient database to store the nitrogen emission coefficients per unit area corresponding to different land attribute types;

[0112] The nitrogen emission calculation module 302 is used to associate the land attribute type of the target area with the nitrogen emission coefficient library, assign a basic nitrogen emission coefficient in the database to each land attribute type plot, correct the model nitrogen emission coefficient by the base area, calculate the plot emission of the target area based on the corrected nitrogen emission coefficient, and convert it into emission raster data to generate a nitrogen emission distribution map;

[0113] The river section impact area and river directed graph construction module 303 is used to convert the discharge raster data into point data, divide the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and construct a river directed graph network based on the river vector data;

[0114] The absorption and load calculation module 304 is used to calculate the river section absorption coefficient using an exponential decay model, and based on the river node directed graph network combined with the river section absorption coefficient, calculate the pollution load transmission volume and the river section absorption capacity step by step along the river directed graph network topology through breadth-first search;

[0115] The distribution map generation module 305 is used to accumulate the total absorption capacity of the affected river section for each pollution point, subtract the total absorption capacity from the land discharge to obtain the net discharge load and convert it into raster data to generate the net discharge load distribution map and the absorption capacity distribution map.

[0116] In some optional embodiments, the nitrogen emission calculation module 302 corrects the model nitrogen emission coefficient using the reference area, including:

[0117] Based on the actual area of ​​the plot A current With unit area A base , the emission coefficient C is corrected by the power exponential relationship adj :

[0118] C adj =C base ×(A current A base ) k

[0119] Among them, C base is the basic nitrogen emission coefficient, and k is the empirical correction factor, which is used to reflect the nonlinear effect of plot area on emission intensity.

[0120] In some optional implementations, the water absorption and load calculation module 304 uses an exponential decay model to calculate the river absorption coefficient, including:

[0121] η uv =(1-Ef)^L uv

[0122] Where η uv is the corrected absorption coefficient, Ef is the absorption efficiency per kilometer of the river section, L uv is the actual length of the river section, which is obtained based on the river attributes of the river surface vector data.

[0123] In some optional embodiments, the pollution load and load calculation module 304 calculates the pollution load transmission and the river section absorption capacity step by step along the river directed graph network topology based on the river node directed graph network and the river section absorption coefficient through breadth-first search, including:

[0124] The flow ratio of each branch of the river is obtained through river attributes to calculate the diversion ratio of pollutants;

[0125] Calculate the load increment transmitted from the current pollution point to the downstream:

[0126] Q new =Q ucurrent ×η uv ×α uv

[0127] Calculate the current river section's absorption capacity:

[0128] D uv =Q ucurrent ×(1-η uv )×α uv

[0129] Among them, Q ucurrent represents the real-time pollution load of the current node u, η uv Represents the self-purification efficiency coefficient of the river section (u→v); α uv Indicates the diversion ratio of the river section (u→v), Q new Indicates the pollutants flowing downstream from the current node, D uv Indicates the pollutants absorbed by the current node;

[0130] Through the river node relationship of the river node directed graph network, based on the above steps, the pollutant absorption and pollutant accumulation of each river section are calculated step by step through breadth-first search.

[0131] The embodiment of the present invention also provides an intelligent interactive system for non-point source pollution based on intelligent agents, such as Figure 4 Shown, including:

[0132] User information acquisition module 401, used to obtain user text or multimodal input information;

[0133] The demand analysis module 402 is used to analyze the user input information through a large language model and determine whether the user demand type is a normal conversation or a non-point source pollution simulation demand;

[0134] The non-point source pollution simulation demand triggering module 403 is used to extract key parameters and target areas related to non-point source pollution from the user input information when it is determined that there is a non-point source pollution simulation demand;

[0135] The non-point source pollution simulation module 404 is configured to call the non-point source pollution simulation method through the interface to obtain a non-point source pollution simulation agent, and calculate at least one of the emission amount, net emission load, and absorption amount of the target area based on the key parameters related to non-point source pollution input by the user and the target area;

[0136] The result feedback module 405 converts the calculation results into natural language to answer the user and generates a vector data result graph in a visual way.

[0137] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0138] The non-point source pollution simulation system and the agent-based non-point source pollution intelligent interaction system in this embodiment are presented in the form of functional units, where the units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0139] The embodiment of the present invention also provides a computer device having the non-point source pollution simulation system 3 shown in the above figure or Figure 4 The agent-based intelligent interactive system for non-point source pollution is shown.

[0140] See also Figure 5 , Figure 5 Schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0141] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0142] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0143] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0144] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0145] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0146] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0147] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0148] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for simulating non-point source pollution, characterized in that: include: Extract the land parcel boundaries and attribute types of the target area's land use vector data and the river attributes of the river surface vector data, and build a nitrogen emission coefficient database to store the unit area nitrogen emission coefficients corresponding to different land attribute types; The land attribute type of the target area is associated with the nitrogen emission coefficient database, and the basic nitrogen emission coefficient in the database is assigned to each land attribute type plot. The model nitrogen emission coefficient is corrected by the benchmark area, and the plot emission in the target area is calculated based on the corrected nitrogen emission coefficient. The emission is converted into emission raster data and a nitrogen emission distribution map is generated. Convert the emission raster data into point data, divide the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and construct a river directed graph network based on the river vector data; The exponential decay model is used to calculate the river section absorption coefficient. Based on the directed graph network of river nodes and the river section absorption coefficient, the pollution load transmission and river section absorption capacity are calculated step by step along the river directed graph network topology through breadth-first search. The total absorption capacity of each pollution point in the affected river section is accumulated, and the net emission load is obtained by subtracting the total absorption capacity from the land emission capacity and converted into raster data to generate the net emission load distribution map and absorption capacity distribution map.

2. The method according to claim 1, characterized in that The nitrogen emission coefficient of the model corrected by the reference area includes: Based on the actual area of ​​the plot A current With unit area A base , the emission coefficient C is corrected by the power exponential relationship adj : C adj =C base ×(A current A base ) k Among them, C base is the basic nitrogen emission coefficient, and k is the empirical correction factor, which is used to reflect the nonlinear effect of plot area on emission intensity.

3. The method according to claim 1, characterized in that The exponential decay model is used to calculate the river section absorption coefficient, including: or uv =(1-Ef) Luv Where η uv is the corrected absorption coefficient, Ef is the absorption efficiency per kilometer of the river section, L uv is the actual length of the river section, which is obtained based on the river attributes of the river surface vector data.

4. The method according to claim 1, wherein The method is based on the directed graph network of river nodes and the river section absorption coefficient, and calculates the pollution load transmission amount and the river section absorption amount step by step along the river directed graph network topology through breadth-first search, including: The flow ratio of each branch of the river is obtained through river attributes to calculate the diversion ratio of pollutants; Calculate the load increment transmitted from the current pollution point to the downstream: Q new =Q ucurrent ×η uv ×α uv Calculate the current river section's absorption capacity: D uv =Q ucurrent ×(1-n) uv )×a uv Among them, Q ucurrent represents the real-time pollution load of the current node u, η uv Represents the self-purification efficiency coefficient of the river section (u→v); α uv Indicates the diversion ratio of the river section (u→v), Q new Indicates the pollutants flowing downstream from the current node, D uv Indicates the pollutants absorbed by the current node; Through the river node relationship of the river node directed graph network, based on the above steps, the pollutant absorption and pollutant accumulation of each river section are calculated step by step through breadth-first search.

5. An intelligent interactive method for non-point source pollution based on an intelligent agent, characterized in that: include: Receive user text or multimodal input information; Analyze user input information through a large language model and determine whether the user's demand type is normal conversation or non-point source pollution simulation demand; When it is determined that the simulation needs are for non-point source pollution, the key parameters and target areas related to non-point source pollution in the user input information are extracted; The non-point source pollution simulation intelligent body obtained by the method according to any one of claims 1 to 4 is called through an interface, and based on the key parameters related to the non-point source pollution and the target area, at least one of the emission amount, net emission load and absorption amount of the target area is calculated; The calculation results are converted into natural language to answer users, and vector data result graphs are generated in a visual way.

6. A non-point source pollution simulation system, characterized in that: include: The data preparation module is used to extract the land parcel boundaries and attribute types of the land use vector data of the target area and the river attributes of the river water surface vector data, and to build a nitrogen emission coefficient database to store the nitrogen emission coefficients per unit area corresponding to different land attribute types; The nitrogen emission calculation module is used to associate the land attribute type of the target area with the nitrogen emission coefficient library, assign a basic nitrogen emission coefficient in the database to each land attribute type plot, correct the model nitrogen emission coefficient by the benchmark area, calculate the plot emission of the target area based on the corrected nitrogen emission coefficient, and convert it into emission raster data to generate a nitrogen emission distribution map; The river section impact area and river directed graph construction module is used to convert the discharge raster data into point data, divide the river section impact area according to the Euclidean distance between each pollution point and the nearest river section in the point data, and construct a river directed graph network based on the river vector data; The absorption capacity and load calculation module is used to calculate the river section absorption coefficient using an exponential decay model. Based on the river node directed graph network combined with the river section absorption coefficient, the pollution load transmission capacity and the river section absorption capacity are calculated step by step along the river directed graph network topology through breadth-first search; The distribution map generation module is used to accumulate the total amount of absorption of the affected river section for each pollution point, subtract the total absorption amount from the land emission amount to obtain the net emission load and convert it into raster data to generate the net emission load distribution map and the absorption amount distribution map.

7. An intelligent interactive system for non-point source pollution based on an intelligent agent, characterized in that: include: User information acquisition module, used to obtain user text or multimodal input information; The demand analysis module is used to analyze user input information through a large language model and determine whether the user's demand type is a normal conversation or a non-point source pollution simulation demand; The non-point source pollution simulation demand triggering module is used to extract key parameters and target areas related to non-point source pollution from the user input information when it is determined that there is a non-point source pollution simulation demand; A non-point source pollution simulation module, configured to call, through an interface, a non-point source pollution simulation agent obtained by the method according to any one of claims 1 to 4, and calculate at least one of the emissions, net emission load, and absorption capacity of the target area based on the key parameters related to non-point source pollution input by the user and the target area; The result feedback module is used to convert the calculation results into natural language to answer users and generate vector data result graphs in a visual way.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the non-point source pollution simulation method according to any one of claims 1 to 4 or the intelligent agent-based non-point source pollution intelligent interaction method according to claim 5 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, which are used to enable a computer to execute the non-point source pollution simulation method described in any one of claims 1 to 4 or the agent-based non-point source pollution intelligent interaction method described in claim 5.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the non-point source pollution simulation method according to any one of claims 1 to 4 or the intelligent agent-based non-point source pollution intelligent interaction method according to claim 5.

Images