A knowledge-driven method, apparatus, and equipment for groundwater simulation and prediction

Abstract

This invention relates to the field of groundwater simulation technology, and discloses a knowledge-driven groundwater simulation and prediction method, apparatus, and equipment. The method includes: dividing a target area into multiple computational units according to preset rules, and obtaining current entity information of groundwater in the target area; determining the current entity relationships and current attributes of each entity based on the current entity information, and constructing a current groundwater knowledge graph of the target area using the current entity information, current entity relationships, and current attributes of each entity; and predicting the groundwater flow direction, flow rate, and water level of the target area at the next sampling time using a current prediction inference model based on the current groundwater knowledge graph of the target area. By constructing a groundwater knowledge graph of the target area, the stratigraphic structure of the groundwater system is accurately depicted. The groundwater knowledge graph drives the calculation and optimization of the prediction model, improving the flexibility and simulation efficiency of the prediction model while meeting the groundwater prediction accuracy requirements.

Description

Technical Field

[0001] This invention relates to the field of groundwater simulation technology, specifically to a knowledge-driven groundwater simulation and prediction method, apparatus, and equipment. Background Technology

[0002] A groundwater flow model is a mathematical tool or computer program used to simulate the movement and distribution of groundwater in aquifers. By combining physical laws, geological parameters, and boundary conditions, it quantitatively describes the water level, flow velocity, flow direction, and changes over time of groundwater.

[0003] Groundwater flow models mainly include analytical models, numerical models, and machine learning models. The characteristics of each type of model are as follows: analytical models have a clear structure and are simple to calculate, but are usually only applicable to simple geometries and homogeneous conditions; numerical models can provide refined calculation results, but have high requirements for basic data, complex parameter determination, and low computational efficiency; machine learning models have high flexibility and computational efficiency, but are heavily dependent on the quality of data samples and have problems with poor interpretability and insufficient generalization ability.

[0004] Therefore, in order to meet the increasingly sophisticated needs of groundwater management, how to accurately characterize the stratigraphic structure of the groundwater system and meet the support requirements of model applications to achieve precise, accurate, and rapid three-dimensional groundwater spatiotemporal dynamic evolution inference and prediction is a problem that groundwater flow calculation models urgently need to solve. Summary of the Invention

[0005] In view of this, the present invention provides a knowledge-driven groundwater simulation and prediction method, apparatus and equipment to solve the problem of inaccurate characterization of the stratigraphic structure of groundwater systems, and improves modeling flexibility and simulation efficiency while meeting the prediction accuracy requirements of model application support.

[0006] In a first aspect, the present invention provides a knowledge-driven groundwater simulation and prediction method, the method comprising:

[0007] The target area is divided into multiple calculation units according to preset rules, and the current entity information of groundwater in the target area is obtained based on the calculation units;

[0008] Based on the current entity information of each computing unit, the current entity relationships between entities and the current attributes of each entity are determined, and the current groundwater knowledge graph of the target area is constructed using the current entity information, current entity relationships and current attributes of each entity in each computing unit.

[0009] Based on the current groundwater knowledge graph of the target area, the current prediction inference model is used to predict the groundwater flow direction, flow rate and water level of the target area at the next sampling time, and the groundwater prediction results at the next sampling time are obtained.

[0010] The knowledge-driven groundwater simulation and prediction method provided by this invention integrates key information from traditional groundwater numerical models and measured data, summarizes and extracts groundwater knowledge graphs for target areas, efficiently simulates groundwater information in target areas, accurately depicts the stratigraphic structure of groundwater systems, and uses the groundwater knowledge graph to drive the calculation and optimization of prediction models. While meeting the requirements for groundwater prediction accuracy, it improves the flexibility of prediction models and the efficiency of simulation calculations.

[0011] In one alternative implementation, the process of determining the current predictive inference model includes:

[0012] Obtain the measured groundwater level data and predicted groundwater level results for the target area at the current sampling time;

[0013] Compare the predicted groundwater level at the current sampling time with the measured groundwater level data to determine the prediction error of the previous prediction inference model;

[0014] Based on the prediction error, the parameters of the previous prediction inference model are optimized and the error is corrected to obtain the current prediction inference model.

[0015] The knowledge-driven groundwater simulation and prediction method provided by this invention not only uses a predictive inference model to predict groundwater, but also uses predicted data and measured data to iteratively optimize the predictive inference model, thereby improving the performance of the predictive inference model and ensuring the accuracy of groundwater prediction.

[0016] In one optional implementation, entity information includes geographic entities and abstract entities. Based on the entity information of each computing unit, the relationships and attributes of each entity are determined, including:

[0017] Geographical entities are classified into the following types based on their underground media and stratigraphic structure: Quaternary aquifers, aquitards, faults, and karst aquifers.

[0018] Obtain the area of ​​each stratigraphic structure type in the target calculation unit, and compare the area size of each stratigraphic structure type;

[0019] The stratigraphic structure type with the largest area is used as the stratigraphic structure type of the target calculation unit;

[0020] Acquire basic data of the target area, and extract the relationship and attribute information of each geographic entity from the basic data by combining natural language processing technology and a preset information extraction model;

[0021] Based on the relationship and attribute information of each geographic entity, the relationships between each geographic entity and the relationships between abstract entities and geographic entities are determined as the entity relationships.

[0022] The knowledge-driven groundwater simulation and prediction method provided by this invention determines the stratigraphic structure type of each calculation unit and takes into account the impact of differences in stratigraphic structure type on groundwater prediction results. This makes the groundwater simulation of the target area more consistent with the actual situation and improves the accuracy of groundwater simulation and prediction for different stratigraphic structure types.

[0023] In one optional implementation, based on the current groundwater knowledge graph of the target area, the current prediction inference model is used to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time, including:

[0024] Based on the current groundwater knowledge graph of the target area, determine the current water level data of each calculation unit;

[0025] The groundwater flow direction inference rules for each computing unit are determined based on the type of geographic entity, and the flow direction relationship between each computing unit is determined by combining the current water level data of each computing unit.

[0026] The flow rate relationship of each calculation unit at the next sampling time is determined based on the flow direction relationship between each calculation unit and the current water level data;

[0027] The water level at the next sampling time can be inferred based on the flow relationship between each computing unit at the next sampling time.

[0028] The knowledge-driven groundwater simulation and prediction method provided by this invention utilizes groundwater knowledge graphs and groundwater flow direction, flow rate, and water level reasoning rules to predict the groundwater flow direction, flow rate, and water level at the next sampling time in advance. This method can not only effectively address the challenges in groundwater management but also help promote the rational development and protection of water resources.

[0029] In one optional implementation, determining the flow rate relationship of each computing unit at the next sampling time based on the flow direction relationship between each computing unit and the current water level data includes:

[0030] Based on the preset permeability coefficients of each stratigraphic structure type and the stratigraphic structure type of the target calculation unit, the permeability coefficient corresponding to the target calculation unit is determined.

[0031] Obtain the current groundwater movement characteristics of the target calculation unit, and determine the flow prediction formula based on the movement characteristics and flow direction relationship;

[0032] Based on the permeability coefficient corresponding to the target calculation unit and the current water level data of the target calculation unit, the groundwater flow rate of the target area at the next sampling time is determined using the flow prediction formula.

[0033] The knowledge-driven groundwater simulation and prediction method provided by this invention determines the permeability coefficient based on the stratigraphic structure type of the computational unit, thereby accurately predicting groundwater flow. This enhances the understanding of groundwater system behavior, facilitates more efficient water resource management, reduces unnecessary exploration and monitoring costs, provides scientific basis for decision-makers, and assists them in making more rational resource management and environmental protection decisions.

[0034] In one optional implementation, inferring the water level at the next sampling time based on the flow relationship between the computing units includes:

[0035] Based on the groundwater flow and source-sink data of the target area at the next sampling time, and combined with the water balance formula, the water level change at the next sampling time is determined.

[0036] Based on the water level change at the next sampling time and the water level data at the current sampling time, the water level data at the next sampling time is predicted.

[0037] In one optional implementation, based on the groundwater flow rate in the target area at the next sampling time and combined with the water balance formula, the water level change at the next sampling time is determined, including:

[0038]

[0039] Where Δh represents the water level change at the next sampling time, P represents precipitation, E represents evaporation or extraction, R represents replenishment, D represents discharge, and Q... in and Q out These represent the lateral inflow and outflow flows, respectively, with Δt representing the sampling time interval, and S... y Indicates the water supply level for diving;

[0040]

[0041] Where Δh represents the water level change at the next sampling time, P represents precipitation, E represents evaporation or extraction, R represents replenishment, D represents discharge, and Q... in and Q out Δt represents the lateral inflow and outflow flow rates, respectively, Δt represents the sampling time interval, and S represents the water storage coefficient of the confined water.

[0042] The knowledge-driven groundwater simulation and prediction method provided by this invention utilizes the inflow and outflow of groundwater (i.e., groundwater flow) and all factors that may affect the groundwater level (such as rainfall, extraction volume, ecological water replenishment, etc.) to more accurately predict the groundwater level at a future moment. By monitoring and predicting changes in groundwater level, the sustainable use of groundwater resources can be ensured. Detailed data analysis can help decision-makers formulate more reasonable groundwater management plans and effectively manage and use groundwater resources.

[0043] Secondly, the present invention provides a knowledge-driven groundwater simulation and prediction device, the device comprising:

[0044] The entity information acquisition module is used to divide the target area into multiple calculation units according to preset rules, and to acquire the current entity information of groundwater in the target area based on the calculation units;

[0045] The knowledge graph construction module is used to determine the current entity relationships between entities and the current attributes of each entity based on the current entity information of each computing unit, and to construct the current groundwater knowledge graph of the target area using the current entity information, current entity relationships and current attributes of each entity of each computing unit.

[0046] The groundwater prediction module is used to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time based on the current groundwater knowledge graph of the target area and the current prediction inference model, so as to obtain the groundwater prediction result at the next sampling time.

[0047] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the method described in the first aspect or any corresponding embodiment thereof.

[0048] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0049] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0050] Figure 1 This is a flowchart illustrating a knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention.

[0051] Figure 2 This is a flowchart illustrating another knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention.

[0052] Figure 3This is a schematic diagram illustrating the construction process of a groundwater knowledge graph in a knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention.

[0053] Figure 4 This is a schematic diagram of the knowledge-driven engine driving process for groundwater flow calculation in the knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention.

[0054] Figure 5 This is a schematic diagram of a groundwater knowledge graph in a specific embodiment of the knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention;

[0055] Figure 6 This is a structural block diagram of a knowledge-driven groundwater simulation and prediction device according to an embodiment of the present invention.

[0056] Figure 7 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0058] This invention provides a knowledge-driven groundwater simulation and prediction method. By constructing a groundwater knowledge graph and using a predictive reasoning model based on the knowledge graph, the method aims to accurately depict the stratigraphic structure of the groundwater system and improve prediction accuracy.

[0059] According to an embodiment of the present invention, a knowledge-driven groundwater simulation and prediction method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0060] This embodiment provides a knowledge-driven groundwater simulation and prediction method, which can be used in the aforementioned computer system. Figure 1 This is a flowchart of a knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0061] Step S101: Divide the target area into multiple calculation units according to preset rules, and obtain the current entity information of groundwater in the target area based on the calculation units.

[0062] Specifically, the diverse topography leads to a complex Earth groundwater system, including various types such as Quaternary loose pore water and carbonate rock fissure karst water. Due to various natural factors or human activities, the groundwater level may rise rapidly. Accurately predicting the groundwater dynamics within a predetermined range of the target area is an urgent need for refined groundwater management. However, due to the influence of various factors such as the heterogeneity of strata, traditional simulation methods are difficult to meet management needs.

[0063] Based on specific management needs and business rules, the division of groundwater calculation units should be flexible. In the division process, priority should be given to key areas within the management objectives, such as groundwater reserve areas, spring areas, over-extraction areas, and areas with leakage and buoyancy risks. Simultaneously, factors such as administrative divisions, stratigraphic distribution, and differences in water-bearing capacity should be considered to comprehensively assess geological conditions and hydrogeological characteristics for a reasonable division.

[0064] Based on key areas within the target region, and considering factors such as administrative divisions, stratigraphic distribution, and differences in water abundance, as well as geological conditions and hydrogeological characteristics, the target region is rationally divided into regional and stratigraphic units for groundwater calculation. The types of each calculation unit are generalized, and based on the characteristics of groundwater quality and stratigraphic structure, each calculation unit is specifically classified into Quaternary aquifers, aquitards, faults, and karst aquifers. Quaternary aquifers represent the main paths of water flow, aquitards are used to define isolated areas of water flow, and faults and karst aquifers represent special structures in groundwater flow.

[0065] Entities include geographic entities and abstract entities. Geographic entities are objectively existing objects such as underground media, geological structures, wells, rivers, springs, and faults based on various computing units, and are the basis for providing data, information, and logical reasoning. Abstract entities are non-objectively existing methods and rules such as calculation functions and constraints associated with geographic entities, and are the link between geographic entities and mathematical models. They are mainly used to evaluate and calculate the data and information characteristics of geographic entities.

[0066] Step S102: Based on the current entity information of each computing unit, determine the current entity relationship between entities and the current attributes of each entity, and use the current entity information, current entity relationship and current attributes of each computing unit to construct the current groundwater knowledge graph of the target area.

[0067] Specifically, the attributes of geographic entities include, but are not limited to, basic features, computational data and business objectives, and control indicators. Examples include the name, location, area, water level, and water level control indicators of groundwater computational units; the name, thickness, area, permeability coefficient, specific yield, and storage coefficient of three-dimensional strata; the name and characteristics of three-dimensional structures; the name, stratigraphy, location, and extraction volume of wells; the name, location, and flow rate of rivers; the name, location, area, and water level control indicators of springs; and the name and location of faults. The attributes of abstract entities include, but are not limited to, their functions and parameters, such as the name, description, form, and parameters of computational functions and constraints.

[0068] The relationships between entities include, but are not limited to, the relationships between geographic entities (such as belonging to, located at, adjacent to, inflow, outflow, etc.), and the relationships between abstract entities and geographic entities (such as applied to, derived).

[0069] The current groundwater knowledge graph is constructed using the current entity information, current entity relationships, and current attributes of each entity in each computing unit. The specific construction method of the knowledge graph is a mature existing technology and will not be elaborated here.

[0070] Step S103: Based on the current groundwater knowledge graph of the target area, use the current prediction inference model to predict the groundwater flow direction, flow rate and water level of the target area at the next sampling time, and obtain the groundwater prediction result at the next sampling time.

[0071] Specifically, the assessment characteristics of groundwater mainly include groundwater flow direction, flow rate, and water level. Therefore, based on the current groundwater knowledge graph of the target area, the current predictive inference model is used to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time. For example, the current groundwater knowledge graph can be used to predict the groundwater situation for the next year. This is just an example and is not a limitation. When new groundwater entity information is obtained, the predictive inference model is optimized using the new entity information data.

[0072] Based on the water levels of the upstream and downstream calculation units at the current sampling time, the flow direction relationship between each calculation unit is inferred. Simultaneously, the lateral / vertical inflow and outflow rates at the next sampling time are calculated. These are then superimposed with the source and sink data for the next sampling time. Based on the principle of water balance, the water level variation at the next sampling time is obtained, and the water level value for the next time time is then inferred. At each sampling time, the calculation is recursively performed from upstream to downstream according to the flow field. After calculating one time, the same process is repeated to calculate the next time.

[0073] The calculation results are evaluated for accuracy using multiple evaluation metrics, and the evaluation results are fed back to the parameter optimization module in real time to support iterative improvement of the model.

[0074] Based on the intelligent optimization mechanism triggered by the error threshold, the particle swarm optimization algorithm is integrated to dynamically correct key parameters such as the permeability coefficient. After optimization, the equivalent time is used to verify the stability of the parameters, forming a closed loop of "evaluation-optimization-verification" to achieve autonomous optimization of model parameters.

[0075] The knowledge-driven groundwater simulation and prediction method provided in this embodiment integrates key information from traditional groundwater numerical models and measured data, summarizes and extracts groundwater knowledge graphs for the target area, efficiently simulates groundwater information in the target area, accurately depicts the stratigraphic structure of the groundwater system, and uses the groundwater knowledge graph to drive the calculation and optimization of the prediction model. While meeting the groundwater prediction accuracy requirements, it improves the flexibility of the prediction model and the efficiency of simulation calculation.

[0076] This embodiment provides a knowledge-driven groundwater simulation and prediction method, which can be used in the aforementioned computer system. Figure 2 This is a flowchart of a knowledge-driven groundwater simulation and prediction method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0077] Step S201: Divide the target area into multiple calculation units according to preset rules, and obtain the current entity information of groundwater in the target area based on the calculation units. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0078] Step S202: Based on the current entity information of each computing unit, determine the current entity relationship between entities and the current attributes of each entity, and use the current entity information, current entity relationship and current attributes of each computing unit to construct the current groundwater knowledge graph of the target area.

[0079] Specifically, entity information includes geographic entities and abstract entities. Step S202 above, which determines the relationships and attributes of each entity based on the entity information of each computing unit, includes:

[0080] Step S2021: Based on the different underground media and stratigraphic structures, the types of geographical entities are divided into: Quaternary aquifers, aquitards, faults, and karst aquifers.

[0081] Specifically, the characteristics of groundwater quality and stratigraphic structure for different types of calculation units are as follows:

[0082] Quaternary aquifers: Piedmont areas (no good aquifers, water present in localized areas, but the water volume is small and the water-bearing capacity varies), single-structure sandy gravel aquifers (at a drawdown of 5m, the water yield per well is generally greater than 5000m³). 3 / d), a two- to three-layered sand and gravel aquifer (with a drawdown of 5m, the water output of a single well can generally reach 3000 to 5000 m³ / h). 3 / d), a multi-layered aquifer consisting of gravel and sand with a small amount of sand (when the drawdown is 5m, the water output of a single well can generally reach 1500-3000m³). 3 / d), a multi-layered sandy aquifer with a small amount of gravel (at a drawdown of 5m, the single well yield is 1500m³). 3 (approximately / d).

[0083] Aquitard: Quaternary bottom clay layer (thickness 5-20m), Ordovician limestone interbedded with mudstone layer.

[0084] Fault: A fracture that causes water blockage or conduction.

[0085] Karst aquifers: The main water-bearing strata are the Ordovician aquifer groups (including exposed areas, buried areas, and Ordovician strata that are in direct contact with Quaternary aquifers).

[0086] The type of calculation unit is determined according to the characteristics of different types of groundwater quality and stratigraphic structure.

[0087] Step S2022: Obtain the area of ​​each stratum structure type in the target calculation unit and compare the area size of each stratum structure type.

[0088] Specifically, the area of ​​different stratigraphic structures in each calculation unit can be obtained through methods such as field measurements, and the area size of each stratigraphic structure type can be compared.

[0089] Step S2023: Select the stratigraphic structure type with the largest area as the stratigraphic structure type of the target calculation unit.

[0090] Specifically, if a certain calculation unit contains multiple types, the type of the calculation unit is generalized according to the area with the largest proportion. For example, if a certain calculation unit includes three types: Quaternary aquifer, aquitard, and fault, and the area of ​​the Quaternary aquifer is determined to be X, the area of ​​the aquitard to be Y, and the area of ​​the fault to be Z, and X>Z>Y, then the calculation unit is determined to be a Quaternary aquifer. This is just an example, but it is not a limitation.

[0091] Step S2024: Obtain basic data of the target area, and extract the relationship information and attribute information of each geographic entity from the basic data by combining natural language processing technology and preset information extraction model.

[0092] Specifically, basic data includes, but is not limited to, texts and atlases such as hydrogeological concepts, technical reports, policy documents, relevant literature, hydrogeological maps, and regional distribution maps, as well as various semi-structured and structured basic data. For example... Figure 3The diagram illustrates the construction process of a groundwater knowledge graph. For basic data, data cleaning and organization are performed. Natural Language Processing (NLP) technology is used, integrating predefined rules, machine learning, deep learning, and other knowledge extraction methods to extract the relational and attribute information of each geographic entity. This information is then represented in a standardized triplet format. The relational information of each geographic entity is in the form of "entity 1, relation, entity 2", and the attribute information is in the form of "entity, attribute, attribute value".

[0093] Step S2025: Based on the relationship information and attribute information of each geographic entity, determine the relationships between each geographic entity and the relationships between abstract entities and geographic entities as the relationships between entities.

[0094] Specifically, the relationships between geographic entities are determined from the extracted relational and attribute information. Based on the actual situation and the relationships between geographic entities, the relationships between abstract entities and geographic entities are determined. For example, a certain calculation formula (abstract entity) is applied to a certain calculation unit (geographic entity), and the relationships between geographic entities and the relationships between abstract entities and geographic entities are used as the relationships between entities in the groundwater knowledge graph.

[0095] The knowledge-driven groundwater simulation and prediction method provided in this embodiment determines the stratigraphic structure type of each calculation unit and takes into account the impact of differences in stratigraphic structure type on groundwater prediction results. This makes the groundwater simulation of the target area more consistent with the actual situation and improves the accuracy of groundwater simulation and prediction for different stratigraphic structure types.

[0096] Step S203: Based on the current groundwater knowledge graph of the target area, use the current prediction inference model to predict the groundwater flow direction, flow rate and water level of the target area at the next sampling time, and obtain the groundwater prediction result at the next sampling time.

[0097] In some alternative implementations, the process of determining the current predictive inference model includes:

[0098] Obtain the measured groundwater level data and predicted groundwater level results for the target area at the current sampling time.

[0099] By comparing the predicted groundwater level at the current sampling time with the measured groundwater level data, the prediction error of the previous prediction inference model is determined.

[0100] Based on the prediction error, the parameters of the previous prediction inference model are optimized and the error is corrected to obtain the current prediction inference model.

[0101] Specifically, the predictive inference model is iteratively optimized based on measured data, which can avoid large deviations in long-term unchanging predictive inference models and ensure the accuracy of inference predictions.

[0102] This invention utilizes Python to construct entity extraction models, attribute extraction models, and three-dimensional variable flow direction topology reasoning models, groundwater flow reasoning models, and groundwater level change reasoning models. It also uses Java to construct and integrate the knowledge graph database, groundwater flow dynamic evolution calculation rule base, and knowledge-driven engine used in this invention. Figure 4 The diagram shown illustrates the workflow of a knowledge-driven engine for groundwater flow calculation. The main steps are as follows:

[0103] Step X1: Based on the rule-based reasoning engine developed using Apache Jena, perform topological reasoning of groundwater flow direction based on a knowledge graph database according to requirements.

[0104] Step X2 involves inferring the groundwater flow topology and its contained entities based on the inference engine, and then calling the groundwater flow dynamic evolution calculation rule base and calculation knowledge according to the entity object attribute values.

[0105] Step X3: After completing the rule application and knowledge matching for the dynamic evolution calculation of groundwater flow for each entity object, the knowledge-driven engine will gradually call the matched model for calculation according to the direction of groundwater flow for each entity object.

[0106] Based on the current water levels of the upstream and downstream calculation units, the flow direction relationships between the nodes of the calculation units are inferred. Simultaneously, the lateral / vertical inflow and outflow rates for the next time moment are calculated. These are then superimposed with the source and sink data for the next time moment. Based on the principle of water balance, the water level variation for the next time moment is obtained, and the water level value for the next time moment is then inferred. At each time moment, the calculation is recursively performed from upstream to downstream according to the flow field. After calculating one time moment, the same process is repeated to recursively calculate the next time moment.

[0107] Step X4 involves evaluating the accuracy of the calculation results using multiple evaluation metrics and feeding the evaluation results back to the parameter optimization module in real time to support iterative improvement of the model.

[0108] Step X5 involves triggering an intelligent optimization mechanism based on an error threshold, integrating a particle swarm optimization algorithm to dynamically correct key parameters such as the permeability coefficient, and then performing an equivalent time-based simulation to verify parameter stability, forming an "evaluation-optimization-verification" closed loop to achieve autonomous optimization of model parameters.

[0109] The predictive inference model can employ intelligent optimization algorithms to automatically assess and adjust the water level results at each time step. By combining measured data, it can infer the starting point of the problem and use intelligent optimization algorithms to optimize parameters and correct errors at the starting point of the problem. Then, it can continue the inference process from the starting point of the problem, thus achieving a dynamic optimization process.

[0110] The knowledge-driven groundwater simulation and prediction method provided in this embodiment not only uses a predictive inference model to predict groundwater, but also uses predicted data and measured data to iteratively optimize the predictive inference model, thereby improving the performance of the predictive inference model and ensuring the accuracy of groundwater prediction.

[0111] Specifically, geographic entities include: underground media, geological structures, wells, rivers, springs, and faults. Step S203 above, based on the current groundwater knowledge graph of the target area, uses the current predictive inference model to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time, including:

[0112] Step S2031: Based on the current groundwater knowledge graph of the target area, determine the current water level data of each calculation unit.

[0113] Specifically, water level data is an attribute of a geographic entity in the groundwater knowledge graph. Each computational unit includes at least one observation well, which can be used to measure the observed water level of the corresponding computational unit. The average of multiple observed water levels is taken as the water level of the corresponding computational unit. Therefore, the groundwater level data at the current sampling time can be determined using the observation wells. Based on the groundwater runoff knowledge graph, hydrogeological characteristics, and water level data, the flow direction between different groundwater runoff computational units can be accurately determined.

[0114] Step S2032: Determine the groundwater flow direction inference rules for each computing unit based on the type of geographic entity, and determine the flow direction relationship between each computing unit by combining the current water level data of each computing unit.

[0115] Specifically, based on the characteristics of the four types of geological units—Quaternary aquifers, impermeable layers, faults, and karst aquifers—three-dimensional flow direction inference is performed, taking into account both hydraulic gradient and inter-unit transport characteristics. Table 1 shows the flow direction inference rules for the four types of geological structures.

[0116] Table 1

[0117]

[0118] Quaternary aquifers are the main channels for groundwater flow. Due to their high hydraulic conductivity, the flow direction within Quaternary aquifers is primarily influenced by the water levels at adjacent nodes. Generally, water flows along the hydraulic gradient from high to low water levels. However, the actual flow path is also affected by the permeability of the Quaternary aquifer. In highly permeable Quaternary aquifers, water flows faster and is more easily transported; while in low-permeability Quaternary aquifers, the flow is slower and may even be restricted. Therefore, in addition to the hydraulic gradient, the permeability of the Quaternary aquifer must be considered when inferring the flow direction to accurately determine the flow direction and path.

[0119] An impermeable layer primarily serves to block groundwater flow, preventing water from crossing the connections between different strata. The permeability of an impermeable layer is typically very low, even bordering on being impermeable. When groundwater encounters an impermeable layer, its transmission path is effectively blocked, preventing the water from passing through the layer. Therefore, when determining flow direction, if an impermeable layer exists between adjacent nodes, the continuity of the water flow must be broken, directly cutting off the flow direction inference. This layer usually forms a natural barrier for water flow, making the flow relationship between upstream and downstream nodes completely independent, and the flow path completely isolated.

[0120] Faults are a special geological structure in groundwater flow, and their impact on water flow depends on the fault's permeability. Faults can guide water flow, especially when there are fractures or karstification within the fault, allowing water to flow through these fractures. Normally, faults have low permeability, hindering water flow on either side. However, in the presence of fractures or karstification, faults can act as channels for water flow. Therefore, inferring flow direction requires assessing the fault's hydrogeological characteristics to determine if water can flow along it, and considering the water level difference on either side of the fault, as well as the fault's permeability (e.g., fracture width, fracture permeability). In some cases, if the fault has high permeability, water may flow rapidly through fractured areas, altering the flow direction and pattern.

[0121] Karst aquifers typically exist in soluble rocks such as limestone and dolomite. Water can rapidly infiltrate through these fissures, forming high-speed flow paths. Karst aquifers generally have high permeability, allowing water to propagate quickly through these fissures, especially when fissures are well-developed and karstification is high, where the flow velocity can be extremely fast. However, the flow paths of water in karst aquifers are usually irregular and complex, thus allowing for simplified calculations. The direction of water flow between two adjacent nodes can be directly inferred from the water level.

[0122] The knowledge-driven groundwater simulation and prediction method provided in this embodiment uses a groundwater knowledge graph and groundwater flow direction reasoning rules to predict groundwater flow direction and predict the groundwater flow direction at the next sampling time in advance. This not only effectively addresses the challenges in groundwater management but also helps to promote the rational development and protection of water resources.

[0123] Step S2033: Determine the flow rate relationship of each calculation unit at the next sampling time based on the flow direction relationship between each calculation unit and the current water level data.

[0124] In some optional implementations, step S2033 above includes:

[0125] Step a1: Combine the preset permeability coefficients of each stratigraphic structure type with the stratigraphic structure type of the target calculation unit to determine the permeability coefficient corresponding to the target calculation unit.

[0126] Specifically, while reasoning about the flow direction, it is necessary to set different permeability coefficients based on empirical values ​​according to the flow rates between various calculation units such as water level, distance, and area, and select appropriate formulas for calculation based on four types of calculation units: Quaternary aquifers, impermeable layers, faults, and karst aquifers.

[0127] Step a2: Obtain the current groundwater movement characteristics of the target calculation unit, and determine the flow prediction formula based on the movement characteristics and flow direction relationship.

[0128] Specifically, the target calculation unit is used to determine the current groundwater movement characteristics, and the flow prediction formula is determined based on the movement characteristics. The usage conditions of different flow prediction formulas are shown in Table 2.

[0129] Table 2

[0130]

[0131] Step a3: Based on the permeability coefficient corresponding to the target calculation unit and the current water level data of the target calculation unit, the groundwater flow rate of the target area at the next sampling time is determined using the flow prediction formula.

[0132] Specifically, based on the permeability coefficient corresponding to the target calculation unit and the groundwater level in the target area at the current sampling time, an appropriate flow prediction formula is selected to calculate the groundwater flow rate in the target area at the next sampling time. The expressions for each flow prediction formula are shown below.

[0133] (1) Darcy's Law: Based on the water level at the previous moment, the lateral and vertical flow rates of groundwater at this moment can be calculated.

[0134]

[0135] In the formula, K is the permeability coefficient, A is the cross-sectional area, Δh is the water level difference between the two stations, and L is the distance between the two stations.

[0136] (2) Dubuis formula: describes radial flow in unconfined aquifers.

[0137]

[0138] In the formula, Q is the pumping rate of the well (unit: m3 / s), K is the permeability coefficient of the aquifer (unit: m / s), h1 and h2 represent the water level height of the aquifer at different radii r1 and r2 (unit: m), and r1 and r2 are the distances from the measurement point to the center of the well (unit: m).

[0139] (3) Theis formula: used to describe unsteady flow in confined aquifers.

[0140]

[0141] The cumulative pumping influence function W(u) is defined as follows:

[0142]

[0143] In the formula, s represents the drawdown of the well (unit: m), Q represents the pumping rate (unit: m3 / s), T represents the permeability of the aquifer (unit: m2 / s, T = K·b, where b is the aquifer thickness), W(u) represents the Theis influence function, and u represents the dimensionless time factor, defined as r represents the distance from the measuring point to the pumping well (unit: m), S represents the water storage coefficient (dimensionless), and t represents the pumping time (unit: s).

[0144] (4) Recharge through a confined aquifer: The process by which a confined aquifer is recharged through an impermeable layer, used for the analysis of multi-layered aquifer systems. The permeability of the impermeable layer determines the recharge rate.

[0145]

[0146] Where q represents the flow rate per unit area (unit: m / s); K b h represents the gravity permeability coefficient of the impermeable layer (unit: m / s); u The overlying aquifer head (unit: m); h l b represents the head of the confined aquifer (unit: m); b represents the thickness of the impermeable layer (unit: m).

[0147] The knowledge-driven groundwater simulation and prediction method provided in this embodiment determines the permeability coefficient based on the stratigraphic structure type of the computing unit, thereby accurately predicting groundwater flow. This enhances the understanding of groundwater system behavior, facilitates more efficient water resource management, reduces unnecessary exploration and monitoring costs, provides scientific basis for decision-makers, and assists them in making more rational resource management and environmental protection decisions.

[0148] Step S2034: Infer the water level at the next sampling time based on the flow relationship between each calculation unit at the next sampling time.

[0149] Specifically, based on the geological structure type and current groundwater level data of each computational unit, the groundwater knowledge graph is dynamically updated, and the water level at the next sampling time is predicted. By traversing all the smallest computational units and comprehensively considering the flow characteristics within and between different strata, the flow direction relationships between adjacent computational units are quickly determined, updating the relationships between entities in the groundwater runoff knowledge graph and forming a complete three-dimensional groundwater flow topology network. Using graph search algorithms in the knowledge graph (such as breadth-first search, depth-first search, or graph traversal algorithms), nodes and their flow direction relationships within the management area can be efficiently queried, thus revealing the spatial distribution and path of groundwater flow more intuitively.

[0150] In some optional implementations, the step S2034 above, which infers the water level at the next sampling time based on the flow relationship between each computing unit at the next sampling time, includes:

[0151] Step b1: Based on the groundwater flow and source-sink data of the target area at the next sampling time, and combined with the water balance formula, determine the water level change at the next sampling time.

[0152] Specifically, the change in groundwater level at a certain moment can be calculated using the water balance formula, as follows:

[0153] ΔW=(P-E+R-D+Q in -Q out )·Δt (6)

[0154] Where P is precipitation, E is evaporation or extraction, R is replenishment, D is discharge, and Q is... in and Q out These represent the lateral inflow and outflow rates, respectively, where Δt is the time step, and the specific yield (S) for the unconfined aquifer (free aquifer) is... y The formula can be written as:

[0155]

[0156] Where Δh represents the change in groundwater level, S yFor the specific yield, the storage coefficient (S) is used for confined aquifers (confined water bodies), and the formula can be written as:

[0157]

[0158] Where Δh is the pressure head change of the confined water, and S is the water storage coefficient.

[0159] In some alternative implementations, considering the different situations of unconfined and confined water, and incorporating existing water balance formulas, step b1 above includes:

[0160]

[0161] Where Δh represents the water level change at the next sampling time, P represents precipitation, E represents evaporation or extraction, R represents replenishment, D represents discharge, and Q... in and Q out These represent the lateral inflow and outflow flows, respectively, with Δt representing the sampling time interval, and S... y This indicates the water supply level for diving.

[0162]

[0163] Where Δh represents the water level change at the next sampling time, P represents precipitation, E represents evaporation or extraction, R represents replenishment, D represents discharge, and Q... in and Q out Δt represents the lateral inflow and outflow flow rates, respectively, Δt represents the sampling time interval, and S represents the water storage coefficient of the confined water.

[0164] Depending on the actual hydrological conditions, the water level change at the next sampling time can be calculated using either the specific yield or the storage coefficient.

[0165] Step b2: Based on the water level change at the next sampling time and the water level data at the current sampling time, predict the water level data at the next sampling time.

[0166] Specifically, the inference rules based on the water balance formula can calculate the water level change at the current sampling time for each observation well, and further infer the water level value at the current sampling time. During the calculation, the flow rate between each node calculated using groundwater flow inference is used as the lateral inflow / outflow. The values ​​at the current sampling time corresponding to rain gauges, extraction wells, and ecological water replenishment monitoring points within or near the calculation unit are called, and combined with the water balance formula, the water level change at the next sampling time is calculated. The water level value at the current sampling time is superimposed to infer the water level value at the next sampling time, and this inferred water level update is saved to the current water level data of the corresponding calculation unit.

[0167] The knowledge-driven groundwater simulation and prediction method provided in this embodiment utilizes the inflow and outflow of groundwater (i.e., groundwater flow) and all factors that may affect the groundwater level (such as rainfall, extraction volume, ecological water replenishment, etc.) to more accurately predict the groundwater level at a future moment. By monitoring and predicting changes in groundwater level, the sustainable use of groundwater resources can be ensured. Detailed data analysis can help decision-makers formulate more reasonable groundwater management plans and effectively manage and use groundwater resources.

[0168] In one specific implementation, the process of groundwater inference prediction for a target area includes the following steps.

[0169] Step 1, Calculation Unit Division: Based on the 1:50,000 hydrogeological map and borehole stratification data, GIS spatial overlay analysis is used to divide the calculation unit grid according to the differences in strata permeability. For example, the first calculation unit contains 70% sand and gravel layer and 20% clay layer, which is generalized as Quaternary aquifer. The second calculation unit has 65% karst fissures, which is generalized as karst aquifer. This is just an example, but not a limitation.

[0170] Step 2, Entity Classification: For geographic entities, select representative datasets such as project reports, documents, and geographic information manuals for the target area and label them. For example, administrative regions are labeled with the first letter of their pinyin, such as Shijingshan being labeled as "SJS". Further distinctions can be made between mountainous and plain areas, with mountainous areas labeled as "SJS-M" and plain areas as "SJS-P". Groundwater observation wells are labeled as "GWO", rain gauges as "PS", extraction wells as "EW", and ecological water replenishment monitoring stations as "RS". Strata are labeled as "GS", underground media as "SM", hydrogeological profiles as "HP", aquifers as "AQ", unconfined groundwater as "PW", confined aquifers as "CW", springs as "SP", faults as "F", recharge areas as "RA", discharge areas as "DA", and porosity as "N". For the location of geographic entities in text, “B” can be used at the beginning of the entity text, “I” can be used in the middle, and “O” can be used for irrelevant information.

[0171] Abstract entities are non-objective methods and rules such as computational functions and constraints associated with geographic entities. Abstract entities are usually stored in rule-based hydrological datasets. A semantic dictionary of abstract entities can be constructed based on the entity content. For example, the dictionary shown in Table 3 can be defined, but this is not a limitation.

[0172] Table 3

[0173]

[0174] Based on entity categories and attributes, and according to the 3D groundwater flow direction and entity location attributes, appropriate relationships are designed to correctly link each entity node. These relationships include those between geographic entities (e.g., belonging to, located at, adjacent, inflow, outflow), and those between abstract entities and geographic entities (e.g., applied to, derived). For example: entity nodes of adjacent administrative regions and groundwater observation wells are linked according to the "adjacent" relationship; for calculation units with explicit inflow / outflow relationships, they are linked according to the relationship "Calculation Unit A — Inflow / Outflow —> Calculation Unit B"; for the stratigraphic structures where each calculation unit is located, the "located at" relationship is used to link them to the corresponding stratigraphic structures according to the entity location attributes; and for calculation functions, constraints, and other rules associated with calculation units, the "applied to" relationship is used to link them to the corresponding calculation units.

[0175] Step 3, Entity Extraction: For semi-structured text data containing watershed engineering information and natural water area information that includes geographic entities, a trained BERT+BiLSTM+CRF model can be used for entity extraction and annotation to obtain geographic entities with clear categories and accurate content. For structured Excel data containing abstract entities, abstract entities with clear categories and accurate content can be obtained through rule matching using a semantic dictionary.

[0176] Step 4, Geographic Entity Attribute Information Extraction: After obtaining the complete entity set, the corresponding entity attributes are extracted and filled in. Entity attribute extraction has many similarities to geographic entity extraction and labeling. For example, the name attribute can be labeled "NAME", the area attribute "Area", the distance attribute "Distance", the water level attribute "WL", the precipitation attribute "Precipitation", the extraction volume attribute "Exploitation", the ecological water replenishment flow attribute "Replenishment", the observation well depth attribute "Depth", and the permeability coefficient attribute "K". After training the attribute extraction model using the labeled data, the model can be used for attribute extraction and entity attribute filling.

[0177] Step 5: Knowledge Extraction for 3D Groundwater Flow Calculation. First, a rich scenario library is constructed to support subsequent 3D groundwater flow calculations. The data in the scenario library includes groundwater level time series, rainfall time series, extraction time series, and ecological water replenishment flow time series. The data source is initially based on historical measured hydrological data such as water level, rainfall, extraction, and water replenishment. When the measured data has a limited span or insufficient data dimensions to support 3D groundwater flow calculations, traditional hydrological models are used for simulation to enrich the data dimensions and expand the data scale. The 3D groundwater flow scenario library for different regions and strata needs to be labeled using key-value pairs to prepare for the subsequent conversion of the 3D groundwater flow scenario library into knowledge.

[0178] By labeling each feature relationship, we can trace back to the strata, 3D structure, and computational unit type to which each feature relationship belongs. The Word2Vec model is used to match the object name with the map entity object name, thus completing the association between the feature relationship and the map entity object. After the association is completed, the feature relationships are filled as attributes of the graph entity objects to obtain a standardized triplet form and structured expression: (Quaternary aquifer flow direction law -> applied to -> Quaternary aquifer), (all smallest computational units in the Quaternary aquifer -> Quaternary aquifer flow direction law -> flow direction of all smallest computational units in the Quaternary aquifer), (Quaternary aquifer flow inference law -> applied to -> Quaternary aquifer), (all smallest computational units in the Quaternary aquifer -> Quaternary aquifer flow inference law -> flow between all smallest computational units in the Quaternary aquifer), (Quaternary aquifer water level inference law -> applied to -> Quaternary aquifer), (all smallest computational units in the Quaternary aquifer -> Quaternary aquifer water level inference law -> water level of all smallest computational units in the Quaternary aquifer at the next moment).

[0179] Step Six, Storage: The defined entities, relationships, and attributes are sequentially written into the graph database for storage, such as... Figure 5 The image shown is a schematic diagram of the groundwater knowledge graph for the target area.

[0180] A three-dimensional groundwater flow calculation rule base is constructed, including a flow direction topology reasoning rule base, a groundwater flow rate reasoning rule base, and a groundwater level change reasoning rule base.

[0181] Step 7: Construction of the Flow Direction Topology Inference Rule Base: Based on the geological characteristics of the target area, three-dimensional flow direction inference rules are designed for four types of calculation units: Quaternary aquifers, impermeable layers, faults, and karst aquifers, forming logical expressions, as shown in Table 4. Combined with actual data from the target area, rule triggering conditions and parameter thresholds are constructed: Quaternary aquifer permeability coefficient threshold: classified according to aquifer type (sand and gravel layer threshold K≥50m / d, fractured karst layer K≥10m / d); Fault hydraulic conductivity critical value: hydraulic conductivity ≥100m.2 When the value is / d, it is determined to be a water-conducting fault; otherwise, it is a water-blocking fault. Simplified conditions for karst aquifers: when the fracture rate is ≥8%, the water level gradient direct determination rule is enabled.

[0182] Table 4

[0183]

[0184] Based on the existing groundwater knowledge graph database, the unit water level, stratigraphic attributes, and spatial topological relationships in the knowledge graph are loaded. According to the above rules, triggering conditions, and parameter thresholds, adjacent units are traversed, and corresponding rules are matched according to geological type. A recursive algorithm is used to dynamically calculate the water level gradient, permeability coefficient, and fault conductivity to generate a three-dimensional flow direction relationship. The flow direction relationship is written into the graph database in the form of (unit A, flow direction, unit B). Finally, all calculation units are linked together from upstream to downstream in the order of water flow direction.

[0185] Step 8, Construction of Groundwater Flow Inference Rule Base: Construct a general water balance model, mainly considering water supply and demand balance. Combining Darcy's Law, Jubuy's Formula, Theis Formula, and overflow recharge formula, design groundwater flow calculation rules, as shown in Table 5. Based on actual data, construct rule triggering conditions and parameter thresholds: Quaternary aquifer permeability coefficient: sand and gravel layer K = 72 m / d, fissured karst layer K = 25 m / d; impermeable layer permeability coefficient: mudstone interlayer K = 0.005 m / d; fault hydraulic conductivity: A fault zone 200 m 2 / d, B fault zone 50m 2 / d; Equivalent permeability coefficient of karst aquifer: K = 25m / d when the fracture rate is 12%.

[0186] Table 5

[0187]

[0188] Based on the above rules, triggering conditions, and parameter thresholds, a recursive algorithm is adopted to achieve real-time updates of traffic relationships through the Neo4j graph database.

[0189] Step 3: Construction of the groundwater level change inference rule base: Based on the principle of water balance and combined with the dynamic characteristics of groundwater in a certain area, groundwater level change inference rules are designed. For details, please refer to the relevant content in formulas (6)-(10) in the above embodiments, which will not be repeated here.

[0190] The empirical formulas for source and sink terms include the formula for calculating atmospheric precipitation infiltration recharge, river and canal seepage recharge, and irrigation return recharge, etc. The calculations are based on the actual conditions of the study area. The specific formulas and calculation processes are mature existing technologies and will not be elaborated here.

[0191] This embodiment also provides a knowledge-driven groundwater simulation and prediction device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0192] This embodiment provides a knowledge-driven groundwater simulation and prediction device, such as... Figure 6 As shown, it includes:

[0193] The entity information acquisition module 601 is used to divide the target area into multiple calculation units according to preset rules, and to acquire the current entity information of groundwater in the target area based on the calculation units.

[0194] The knowledge graph construction module 602 is used to determine the current entity relationships between entities and the current attributes of each entity based on the current entity information of each computing unit, and to construct the current groundwater knowledge graph of the target area using the current entity information, current entity relationships and current attributes of each entity of each computing unit.

[0195] The groundwater prediction module 603 is used to predict the groundwater flow direction, flow rate and water level of the target area at the next sampling time based on the current groundwater knowledge graph of the target area and the current prediction inference model, so as to obtain the groundwater prediction result at the next sampling time.

[0196] In some alternative implementations, the knowledge graph construction module 602 includes:

[0197] Type classification unit, used to classify the types of geographical entities according to different underground media and stratigraphic structures: Quaternary aquifers, aquitards, faults and karst aquifers.

[0198] The area comparison unit is used to obtain the area of ​​each stratum structure type in the target calculation unit and compare the area size of each stratum structure type.

[0199] The target type determination unit is used to select the stratigraphic structure type with the largest area as the target calculation unit.

[0200] The information extraction unit is used to acquire basic data of the target area and, in conjunction with natural language processing technology and a preset information extraction model, extract the relationship and attribute information of each geographic entity from the basic data.

[0201] The relationship determination unit is used to determine the relationships between geographic entities and the relationships between abstract entities and geographic entities based on the relationship information and attribute information of each geographic entity.

[0202] In some alternative implementations, the groundwater prediction module 603 includes:

[0203] The water level data determination unit is used to determine the current water level data of each calculation unit based on the current groundwater knowledge graph of the target area.

[0204] The flow direction prediction unit is used to determine the groundwater flow direction inference rules for each calculation unit based on the type of geographic entity, and to determine the flow direction relationship between each calculation unit by combining the current water level data of each calculation unit.

[0205] The flow prediction unit is used to determine the flow relationship of each calculation unit at the next sampling time based on the flow direction relationship between each calculation unit and the current water level data.

[0206] The water level prediction unit is used to infer the water level at the next sampling time based on the flow relationship between each calculation unit at the next sampling time.

[0207] In some alternative implementations, the traffic prediction unit includes:

[0208] The permeability coefficient determination sub-unit is used to determine the permeability coefficient corresponding to the target calculation unit by combining the preset permeability coefficients of each stratigraphic structure type and the stratigraphic structure type of the target calculation unit.

[0209] The flow prediction formula determines the sub-unit, which is used to obtain the current groundwater movement characteristics of the target calculation unit, and determines the flow prediction formula based on the movement characteristics and flow relationship.

[0210] The flow prediction subunit is used to determine the groundwater flow rate of the target area at the next sampling time based on the permeability coefficient corresponding to the target calculation unit and the current water level data of the target calculation unit, using the flow prediction formula.

[0211] In some alternative implementations, the water level prediction unit includes:

[0212] The water level change calculation subunit is used to determine the water level change at the next sampling time based on the groundwater flow and source-sink data of the target area at the next sampling time, combined with the water balance formula.

[0213] The water level prediction subunit is used to predict the water level data at the next sampling time based on the water level change at the next sampling time and the water level data at the current sampling time.

[0214] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0215] In this embodiment, the knowledge-driven groundwater simulation and prediction device is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0216] This invention also provides a computer device having the above-described features. Figure 6 The device shown is a knowledge-driven groundwater simulation and prediction device.

[0217] Please see Figure 7 , Figure 7 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 7 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 7 Take a processor 10 as an example.

[0218] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0219] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0220] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0221] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0222] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0223] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0224] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A knowledge-driven groundwater simulation and prediction method, characterized in that, The method includes: The target area is divided into multiple calculation units according to preset rules, and the current entity information of groundwater in the target area is obtained based on the calculation units. Based on the current entity information of each computing unit, the current entity relationships between entities and the current attributes of each entity are determined, and the current groundwater knowledge graph of the target area is constructed using the current entity information, current entity relationships and current attributes of each entity in each computing unit. The entity information includes geographic entities and abstract entities. Based on the entity information of each computing unit, the relationships and attributes of each entity are determined, including: classifying geographic entities into Quaternary aquifers, impermeable layers, faults, and karst aquifers according to different underground media and stratigraphic structures; obtaining the area of ​​each stratigraphic structure type in the target computing unit and comparing the area sizes of each stratigraphic structure type; selecting the stratigraphic structure type with the largest area as the stratigraphic structure type of the target computing unit; obtaining basic data of the target area and extracting the relationship and attribute information of each geographic entity from the basic data using natural language processing technology and a preset information extraction model; and determining the relationships between geographic entities and between abstract entities and geographic entities based on the relationship and attribute information of each geographic entity. Based on the current groundwater knowledge graph of the target area, the current prediction inference model is used to predict the groundwater flow direction, flow rate and water level of the target area at the next sampling time, and the groundwater prediction results at the next sampling time are obtained.

2. The method according to claim 1, characterized in that, The process of determining the current predictive inference model includes: Obtain the measured groundwater level data and predicted groundwater level results for the target area at the current sampling time; Compare the predicted groundwater level at the current sampling time with the measured groundwater level data to determine the prediction error of the previous prediction inference model; Based on the prediction error, the parameters of the previous prediction inference model are optimized and the error is corrected to obtain the current prediction inference model.

3. The method according to claim 1, characterized in that, Based on the current groundwater knowledge graph of the target area, the current prediction and inference model is used to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time, including: Based on the current groundwater knowledge graph of the target area, determine the current water level data of each calculation unit; The groundwater flow direction inference rules for each computing unit are determined based on the type of geographic entity, and the flow direction relationship between each computing unit is determined by combining the current water level data of each computing unit. The flow rate relationship of each calculation unit at the next sampling time is determined based on the flow direction relationship between each calculation unit and the current water level data; The water level at the next sampling time can be inferred based on the flow relationship between each computing unit at the next sampling time.

4. The method according to claim 3, characterized in that, Inferring the water level at the next sampling time based on the flow relationship between each computing unit at the next sampling time includes: Based on the groundwater flow and source-sink data of the target area at the next sampling time, and combined with the water balance formula, the water level change at the next sampling time is determined. Based on the water level change at the next sampling time and the water level data at the current sampling time, the water level data at the next sampling time is predicted.

5. The method according to claim 4, characterized in that, Based on the groundwater flow rate in the target area at the next sampling time, and combined with the water balance formula, the water level change at the next sampling time is determined, including: in, This indicates the change in water level at the next sampling time. Indicates precipitation. This indicates the amount of evaporation or extraction. Indicates the amount of supplies. Indicates the amount of excretion. and These represent the lateral inflow and outflow flows, respectively. Indicates the sampling time interval. Indicates the water supply level for diving; in, This indicates the change in water level at the next sampling time. Indicates precipitation. This indicates the amount of evaporation or extraction. Indicates the amount of supplies. Indicates the amount of excretion. and These represent the lateral inflow and outflow flows, respectively. Indicates the sampling time interval. This indicates the water storage coefficient of confined water.

6. The method according to claim 3, characterized in that, The flow rate relationship of each calculation unit at the next sampling time is determined based on the flow direction relationship between each calculation unit and the current water level data, including: Based on the preset permeability coefficients of each stratigraphic structure type and the stratigraphic structure type of the target calculation unit, the permeability coefficient corresponding to the target calculation unit is determined. Obtain the current groundwater movement characteristics of the target calculation unit, and determine the flow prediction formula based on the movement characteristics and the flow direction relationship; Based on the permeability coefficient corresponding to the target calculation unit and the current water level data of the target calculation unit, the groundwater flow rate of the target area at the next sampling time is determined using the flow prediction formula.

7. A knowledge-driven groundwater simulation and prediction device, characterized in that, The device includes: The entity information acquisition module is used to divide the target area into multiple calculation units according to preset rules, and acquire the current entity information of groundwater in the target area based on the calculation units; The knowledge graph construction module is used to determine the current entity relationships between entities and the current attributes of each entity based on the current entity information of each computing unit, and to construct the current groundwater knowledge graph of the target area using the current entity information, current entity relationships and current attributes of each entity of each computing unit. The entity information includes geographic entities and abstract entities. Based on the entity information of each computing unit, the relationships and attributes of each entity are determined, including: classifying geographic entities into Quaternary aquifers, impermeable layers, faults, and karst aquifers according to different underground media and stratigraphic structures; obtaining the area of ​​each stratigraphic structure type in the target computing unit and comparing the area sizes of each stratigraphic structure type; selecting the stratigraphic structure type with the largest area as the stratigraphic structure type of the target computing unit; obtaining basic data of the target area and extracting the relationship and attribute information of each geographic entity from the basic data using natural language processing technology and a preset information extraction model; and determining the relationships between geographic entities and between abstract entities and geographic entities based on the relationship and attribute information of each geographic entity. The groundwater prediction module is used to predict the groundwater flow direction, flow rate, and water level of the target area at the next sampling time based on the current groundwater knowledge graph of the target area and the current prediction inference model, so as to obtain the groundwater prediction result at the next sampling time.

8. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method of any one of claims 1 to 6.

Images