A knowledge-driven water network system optimization scheduling decision method and device

Abstract

This invention relates to the field of water network system scheduling and decision-making technology, and discloses a knowledge-driven method and apparatus for optimizing water network system scheduling and decision-making. The method includes: acquiring water network system knowledge; constructing a water network system knowledge network based on the water network system knowledge; constructing water network system state reasoning knowledge based on the water network system knowledge network; constructing water network multi-objective optimization scheduling and decision-making knowledge; the water network multi-objective optimization scheduling and decision-making knowledge includes the water network optimization scheduling objective function, constraints, multi-objective optimization solution algorithms, and multi-attribute decision algorithms; and constructing a multi-level knowledge-fusion intelligent decision-making reasoning engine for water network system optimization scheduling based on the water network system knowledge network, water network system state reasoning knowledge, and water network multi-objective optimization scheduling and decision-making knowledge to generate water network system scheduling and decision-making schemes. This invention significantly improves the intelligence level, computational efficiency, and decision accuracy of water network system optimization scheduling and decision-making.

Description

Technical Field

[0001] This invention relates to the field of water network system scheduling and decision-making technology, specifically to a knowledge-driven water network system optimization scheduling and decision-making method and apparatus. Background Technology

[0002] With the continuous advancement of urban development and water system connectivity construction, the scale of water network systems is increasing day by day. As the water network system itself has the typical characteristics of multiple water sources, multiple paths, and multiple projects, and at the same time has multiple functions such as water source allocation, flood control and drainage, water supply security, and river and lake interconnection, the complexity of its scheduling and decision-making has increased significantly.

[0003] Currently, the optimization and scheduling decisions of water network systems mainly face the following technical bottlenecks: First, it is difficult to integrate and apply multi-source knowledge. Various types of knowledge, such as design data, physical parameters, operational data, scheduling procedures, and expert experience, are scattered and independent, generally exhibiting problems such as data silos and fragmented knowledge. There is a lack of a unified knowledge representation and organization method suitable for the field of water network scheduling, making it difficult to comprehensively utilize multi-source heterogeneous information in scheduling decisions. Second, traditional simulation and decision-making methods are inefficient. Simply relying on traditional hydraulic models for simulation and pre-planning is difficult to update when facing complex situations such as water network structure adjustments and changes in physical parameters. At the same time, existing scheduling decisions still mainly rely on manually preset scenarios for "forward pre-planning," resulting in low efficiency in multi-scenario simulation calculations and difficulty in quickly generating optimal scheduling schemes based on scenario requirements and scheduling objectives. Summary of the Invention

[0004] This invention provides a knowledge-driven method and apparatus for optimizing the scheduling of water network systems, in order to solve the problem of low accuracy and efficiency in the optimization scheduling of water network systems in the prior art.

[0005] In a first aspect, the present invention provides a knowledge-driven water network system optimization scheduling decision-making method, the method comprising: Acquire knowledge of water network systems; Construct a knowledge network for water network systems based on knowledge of water network systems; Based on the knowledge network of water network systems, construct state reasoning knowledge of water network systems; Constructing multi-objective optimization scheduling decision-making knowledge for water networks; this knowledge includes the objective function, constraints, multi-objective optimization solution algorithms, and multi-attribute decision-making algorithms for water network optimization scheduling. Based on the knowledge network of water network system, the state reasoning knowledge of water network system, and the multi-objective optimization scheduling decision knowledge of water network, a multi-level knowledge fusion intelligent decision reasoning engine for water network system optimization scheduling is constructed to generate water network system scheduling decision schemes.

[0006] This invention constructs a knowledge network for water network systems, state reasoning knowledge, and multi-objective optimization scheduling decision-making knowledge to form a multi-level knowledge fusion intelligent decision-making reasoning engine. It can effectively integrate multi-source heterogeneous knowledge and achieve unified organization and invocation, solving problems such as knowledge dispersion, poor model adaptability, and low decision-making efficiency in traditional scheduling. At the same time, relying on knowledge-driven methods, it realizes multi-model collaborative inference and multi-objective optimization, which can quickly respond to different scheduling scenarios and generate optimal scheduling schemes. It significantly improves the intelligence level, computational efficiency, and decision accuracy of water network system optimization scheduling decisions, and can better meet the complex scheduling needs of multi-source, multi-path, and multi-functional water network systems.

[0007] In one optional implementation, a water network system knowledge network is constructed based on water network system knowledge, including: Extract the physical entities of water network projects, the entity relationships between physical entities of water network projects, and the entity attributes of physical entities of water network projects from the knowledge of water network systems; Based on the physical entities and relationships of water network engineering, a physical water network knowledge graph is constructed. Embed entity attributes into physical entities of water network engineering in the physical water network knowledge graph; Encode the physical water network knowledge graph with embedded entity attributes to obtain the water network system knowledge network.

[0008] By extracting physical entities, entity relationships, and attributes from water network projects to construct a physical water network knowledge graph, and then embedding entity attributes into the graph and encoding them to form a water network system knowledge network, a unified representation and structured organization of multi-source heterogeneous information such as water network topology and physical parameters can be achieved. This effectively breaks down data silos and knowledge fragmentation problems, provides standardized and efficient knowledge support for subsequent state reasoning, model calculation, and scheduling decisions, and improves the efficiency of knowledge retrieval in the water network system.

[0009] In one optional implementation, based on the water network system knowledge network, state reasoning knowledge of the water network system is constructed, including: To obtain the underlying mechanisms, empirical relationships, input parameters, and output variables for calculating the hydrodynamic processes of a water network system; Based on mechanistic formulas, empirical relationships, input parameters, and output variables, the computational entity and its characteristic attributes are determined. By associating computational entities with physical entities of water network engineering, and establishing a mapping relationship between the characteristic attributes of computational entities and the entity attributes of physical entities of water network engineering, we can obtain state reasoning knowledge of the water network system.

[0010] By acquiring the mechanistic formulas, empirical relationships, and input / output parameters required for hydrodynamic process calculations, the computational entity and its characteristic attributes are determined. A mapping relationship is established between the computational entity and the physical entity of the water network project, and between the characteristic attributes of the computational entity and the attributes of the physical entity. This constructs knowledge for water network system state reasoning, enabling flexible association between hydrodynamic calculation logic and the physical entity of the water network. Under conditions such as water network structure adjustments and changes in simulation range, physical parameter matching and hydrodynamic calculations can be quickly achieved. Simultaneously, the core elements and associated logic of hydrodynamic calculation and decision optimization are standardized, allowing for flexible selection of computational objects, calculation methods, and adjustment of scheduling strategies. This solves the problems of deep binding between traditional hydrodynamic calculation and modeling methods and modeling ranges, and the separation between simulation pre-playing and decision optimization. It enables rapid modeling in multiple scenarios, providing accurate and efficient model support for water network state reasoning, scenario pre-playing, and scheduling decisions. While maintaining the accuracy of water network system state prediction and reasoning, it significantly improves computational efficiency, laying a solid model foundation for intelligent scheduling decisions.

[0011] In one alternative implementation, multi-objective optimization scheduling decision-making knowledge for water networks is constructed, including: Define the objectives for optimizing water network scheduling; Determine the objective function for water network optimization scheduling for each water network optimization scheduling objective; Based on the knowledge network of water network system and the state reasoning knowledge of water network system, the constraints are determined. An improved multi-objective particle swarm optimization algorithm was determined and used as the multi-objective optimization solution algorithm. The multi-attribute decision algorithm is used to determine the weight of the decision target by subjective and objective weighting method, normalize the non-dominated solution set, and determine the proximity based on weighted Euclidean distance to realize the ranking of scheduling decision schemes.

[0012] In one alternative implementation, the improved multi-objective particle swarm optimization algorithm includes: A dimensional decoupling strategy of segmented day pre-storage and single-granularity flow optimization is adopted. The day parameter is decoupled from the particle optimization dimension and pre-stored. The particles only optimize the single-granularity flow dimension. The high-dimensional vector is reconstructed by pre-stored days in the fitness calculation stage. A flexible constraint handling strategy based on violation degree and penalty function is used to adaptively penalize particles that violate constraints by quantifying the degree of constraint violation of particles.

[0013] By constructing multi-objective optimization scheduling decision knowledge for water networks, clarifying scheduling objectives and corresponding objective functions, and accurately determining constraints based on the knowledge network and state reasoning knowledge of water network systems, and employing an improved multi-objective particle swarm optimization algorithm as the solution algorithm and a multi-attribute decision algorithm as the scheme ranking algorithm, the problems of single optimization objectives, insufficient multi-objective solution efficiency, and insufficient global optimization capability in traditional water network scheduling are effectively solved.

[0014] In one optional implementation, a multi-level knowledge fusion-based water network system optimization scheduling decision reasoning engine is constructed based on the water network system knowledge network, water network system state reasoning knowledge, and water network multi-objective optimization scheduling decision knowledge, including: A knowledge graph information fusion layer is constructed based on the knowledge network of the water network system to enable rapid organization and retrieval of water network topology, physical parameters, rule constraints and real-time status; A predictive inference collaborative computing layer is constructed based on the state reasoning knowledge of the water network system. The predictive inference task is dynamically invoked and responded to based on the data provided by the knowledge graph information fusion layer, so as to provide candidate solutions for decision-making. Based on the knowledge of multi-objective optimization scheduling in water networks, a decision optimization reasoning layer is constructed. Based on the prediction and deduction of the water network system state in the prediction and deduction collaborative computing layer, the weight coefficients of multiple optimization objectives are dynamically adjusted according to the real-time scenario and scheduling strategy. The multi-objective optimization solution algorithm is used to solve the problem, generate a Pareto optimal solution set, and recommend and rank the decision solutions based on the scheduling strategy. The water network system optimization scheduling decision reasoning engine includes a knowledge graph information fusion layer, a prediction and inference collaborative computing layer, and a decision optimization reasoning layer.

[0015] By constructing a multi-layered knowledge fusion and reasoning engine comprising a knowledge graph information fusion layer, a prediction and inference collaborative computing layer, and a decision optimization and reasoning layer, a unified organization and efficient retrieval of multi-source heterogeneous knowledge in water networks can be achieved. It can dynamically complete water network topology reasoning, state prediction and inference, and scheduling scheme generation based on real-time scenarios. Furthermore, it can dynamically adjust target weights based on scheduling strategies to perform multi-objective optimization and scheme ranking, effectively improving the intelligence level and response speed of water network scheduling decisions, and enhancing the timeliness, accuracy, and practicality of decisions under changing conditions. It can perform flexible modeling, efficient simulation calculations, and multi-objective decision optimization based on a knowledge base, offering advantages such as flexible modeling, high computational efficiency, and accurate decision-making.

[0016] Secondly, the present invention provides a knowledge-driven water network system optimization scheduling decision-making device, the device comprising: The acquisition module is used to acquire knowledge about the water network system. The water network system knowledge network construction module is used to construct a water network system knowledge network based on water network system knowledge. The water network system state reasoning knowledge construction module is used to construct water network system state reasoning knowledge based on the water network system knowledge network. The module for constructing knowledge for multi-objective optimization scheduling of water networks is used to build knowledge for multi-objective optimization scheduling of water networks. This knowledge includes the objective function for water network optimization scheduling, constraints, multi-objective optimization solution algorithms, and multi-attribute decision-making algorithms. The intelligent decision-making reasoning engine construction module for water network system optimization scheduling is used to build a multi-level knowledge fusion intelligent decision-making reasoning engine for water network system optimization scheduling based on water network system knowledge network, water network system state reasoning knowledge, and water network multi-objective optimization scheduling decision knowledge, and generate water network system scheduling decision schemes.

[0017] Thirdly, the present invention provides an electronic device, comprising: 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 computer instructions to perform the knowledge-driven water network system optimization scheduling decision-making method described in the first aspect or any corresponding embodiment.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the knowledge-driven water network system optimization scheduling decision-making method described in the first aspect or any corresponding embodiment above.

[0019] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the knowledge-driven water network system optimization scheduling decision-making method described in the first aspect or any corresponding embodiment above.

[0020] It should be noted that the knowledge-driven water network system optimization scheduling decision-making device, electronic equipment, computer-readable storage medium, and computer program product provided by this invention correspond to the knowledge-driven water network system optimization scheduling decision-making method described above. Therefore, for the beneficial effects of the knowledge-driven water network system optimization scheduling decision-making device, electronic equipment, computer-readable storage medium, and computer program product, please refer to the description of the corresponding beneficial effects of the knowledge-driven water network system optimization scheduling decision-making method above, and will not be repeated here. Attached Figure Description

[0021] 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.

[0022] Figure 1 This is a schematic diagram of the first step of the knowledge-driven water network system optimization scheduling decision-making method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a water network system knowledge network according to an embodiment of the present invention; Figure 3 This is a schematic diagram of an intelligent decision-making reasoning engine for water network optimization scheduling according to an embodiment of the present invention; Figure 4 This is a 4-dimensional schematic diagram of the Pareto solution set for a multi-objective optimization solution in a scheduling scenario according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the optimal decision-making result for a certain scheduling scenario according to an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the specific scheduling process of different strategies under a certain scheduling scenario according to an embodiment of the present invention. Figure 7 This is a structural block diagram of a knowledge-driven water network system optimization scheduling decision-making device according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0023] 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.

[0024] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0025] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0026] According to an embodiment of the present invention, a knowledge-driven water network system optimization scheduling decision 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.

[0027] This embodiment provides a knowledge-driven water network system optimization scheduling decision-making method, which can be used on servers, terminals, mobile terminals, etc. Figure 1 This is a flowchart of a knowledge-driven water network system optimization scheduling decision-making method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain knowledge of the water network system.

[0028] Knowledge of water network systems can be derived from water network engineering design data, operation and monitoring data, scheduling procedures, industry standards, mechanistic models, and expert experience. Specific knowledge of water network systems can include planning and design reports, construction drawings, as-built data, 3D models, water network topology, engineering parameters, and equipment information; it can also include real-time monitoring data such as water level, flow rate, water quality, gate opening, and pumping station operating conditions; it can also include regulatory and normative documents such as water conservancy engineering design standards, scheduling procedures, flood control and drainage plans, water supply security schemes, and river and lake connectivity project operation and management methods; it can also include mechanistic models and empirical knowledge such as hydrological and hydraulic mechanism formulas, engineering empirical relationships, historical scheduling cases, and decision-making rules; and it can also include external environmental information such as meteorological forecast data, hydrological forecast data, topographic data, and watershed / regional hydrological data.

[0029] Step S102: Based on the knowledge of the water network system, construct a knowledge network for the water network system.

[0030] Based on knowledge of water network systems, physical water networks can be abstracted into a network topology structure stored in a NebulaGraph graph database (an open-source database specifically designed for storing and processing relational data), expressed through "nodes" and "edges." Specifically, physical entities of water network projects such as reservoirs, rivers, canals, sluice gates, pumping stations, and water plants can be abstracted as "nodes" in the network topology, while the spatial logical relationships between entities, such as water flow direction, connection, control, and membership, can be abstracted as "edges."

[0031] Based on the "nodes" and "edges" of the water network system topology, engineering data, hydraulic parameters, scheduling rules, and variables such as flow and head are embedded as feature attributes of entity nodes, while spatial logical relationships such as inflow, connection, and control are embedded as feature attributes of edges. Using this topological network structure as an index, all types of feature attributes are uniformly encoded and stored in a graph database, forming a unified underlying knowledge network of the water network system, integrating "physical-rule-state," i.e., the water network system knowledge network. The feature attributes of the physical entity nodes in the water network engineering include engineering attributes, rule attributes, and state attributes. Engineering attributes are the engineering design parameters of the physical entity in the water network engineering; rule attributes are the engineering operation constraints and boundary rules; and state attributes are historical, real-time, or simulated data of the water network operation.

[0032] Step S103: Based on the knowledge network of the water network system, construct the state reasoning knowledge of the water network system.

[0033] Based on the physical entities of water network engineering in the knowledge network of water network systems, computational entities are created. Mechanism formulas (such as Muskingen's formula and Saint-Venant's equation), empirical relationships (such as the flow transmission time and flow loss of upstream and downstream sections of river segments, as well as the relationship between key characteristic indicators such as seepage, channel storage, and water surface area of ​​river segments and the scheduling flow and time), and required input parameters (mainly including water conservancy project flow or water level constraints, hydrodynamic parameters, and flow and water level data) and output variables (mainly hydrodynamic processes such as flow and water level, as well as characteristic values ​​such as water surface area and seepage) are defined as the characteristic attributes of the corresponding computational entities.

[0034] The computational entity is associated with the physical entity of the water network project through the "application" relationship, and the attribute fields of the computational entity are mapped to the attribute fields of the corresponding physical entity of the water network project. Finally, the computational entity and its attributes and relationships are stored in the NebulaGraph graph database.

[0035] Using the water network system knowledge network as a unified knowledge hub for data storage and model parameter retrieval, and with the physical entities of the water network project as the core, deep integration with the water network system state reasoning knowledge is achieved through attribute field mapping between related entities. The reasoning can proceed from upstream to downstream, using the water flow direction as the reasoning path and calling the underlying knowledge network and water network system state reasoning knowledge through the knowledge reasoning engine.

[0036] Step S104: Construct multi-objective optimization scheduling decision knowledge for water networks; the multi-objective optimization scheduling decision knowledge for water networks includes the objective function of water network optimization scheduling, constraints, multi-objective optimization solution algorithm, and multi-attribute decision algorithm.

[0037] First, the objectives of water network optimization and scheduling are clearly defined, which may include river water duration, maximum water surface area, outflow volume, and groundwater infiltration. Then, the specific functions of each water network optimization and scheduling objective are analyzed and determined, clarifying the functional relationship between each objective function and the constraints of the decision variables. All water network optimization and scheduling objective values ​​are derived through intelligent decision-making reasoning engines based on the water network system knowledge network and water network system state reasoning knowledge.

[0038] The constraints include flow rate, water level, flow velocity, gate capacity, water balance constraints, and non-negativity constraints. All upper and lower constraint boundaries are based on the water network system knowledge network and water network system state reasoning knowledge, and are derived through the intelligent decision reasoning engine for water network system optimization scheduling.

[0039] The multi-objective optimization algorithm employs Multi-Objective Particle Swarm Optimization (MOPSO). While MOPSO possesses strong global search capabilities in complex optimization problems, traditional MOPSO suffers from low search efficiency and overly rigid constraint handling in high-dimensional traffic optimization scenarios. To improve the algorithm's performance in traffic optimization tasks, this embodiment utilizes two core aspects of the improved MOPSO algorithm: particle initialization mechanism and constraint handling strategy. This combination enables the improved MOPSO to converge quickly to the optimal solution region while ensuring the feasibility and quality of the optimization results, significantly enhancing algorithm performance. Furthermore, the granularity dimension of the initialized particles is optimized using a dimensional decoupling strategy of "segmented day pre-storage + single-granularity traffic optimization." The day parameter is separated from the particle optimization dimension and pre-stored, allowing particles to optimize only single-granularity traffic parameters. During the fitness calculation phase, the pre-stored days are used to reconstruct the high-dimensional vector, achieving an efficient search mode of "low-dimensional optimization, high-dimensional adaptation." Meanwhile, a flexible constraint processing strategy based on violation degree and penalty function is adopted, abandoning the particle exclusion mechanism of rigid constraints. The flexible constraint processing strategy of "constraint violation degree calculation + penalty function punishment" is adopted. By quantifying the degree of constraint violation of particles, adaptive punishment is applied to particles that violate constraints. This retains potential high-quality search particles and guides particles to iterate in the direction of satisfying constraints through the punishment mechanism.

[0040] The multi-attribute decision-making algorithm determines the decision preferences by using subjective and objective weights in the weighting method. Subjective weights are determined based on different scheduling strategies of the decision-maker, while objective weights are determined using the entropy weighting method. At the same time, the non-dominated solution set data matrix obtained by the multi-objective optimization algorithm is normalized, and the weighted Euclidean distance between each decision objective and the optimal and worst solutions is calculated to determine the proximity and to rank the solutions.

[0041] Step S105: Based on the knowledge network of the water network system, the state reasoning knowledge of the water network system, and the multi-objective optimization scheduling decision knowledge of the water network, a multi-level knowledge fusion intelligent decision reasoning engine for water network system optimization scheduling is constructed to generate a water network system scheduling decision scheme.

[0042] This intelligent decision-making reasoning engine for water network system optimization and scheduling uses the water network system knowledge network as an index, integrating heterogeneous knowledge from multiple sources such as mechanistic formulas and empirical relationships. It performs decision-making reasoning by fusing knowledge from different layers through the reasoning engine. The engine's first layer is a "physical-rule-state" fusion layer based on the water network system knowledge network, enabling rapid organization and retrieval of the water network topology, physical parameters, rule constraints, and real-time states, providing a foundation for upper-level model calculations. The second layer is a prediction and deduction collaborative computing layer based on water network system state reasoning knowledge. Using the parameters, rules, and data provided by the first-layer knowledge network, it dynamically invokes prediction and deduction tasks issued by the top-level decision-making body and provides candidate solutions. The third layer is a strategy optimization and evolution layer based on dynamic weight adjustment constructed from water network multi-objective optimization scheduling decision-making knowledge. Based on the second-layer water network system state prediction and deduction, it dynamically adjusts the weight coefficients of multiple optimization objectives according to real-time scenarios and scheduling strategies, uses multi-objective optimization algorithms to solve the problem, generates a Pareto optimal solution set, and recommends and ranks decision solutions based on scheduling strategies, achieving a closed-loop decision-making process of "scenario-objective-strategy-solution."

[0043] This invention constructs a knowledge network for water network systems, state reasoning knowledge, and multi-objective optimization scheduling decision-making knowledge to form a multi-level knowledge fusion intelligent decision-making reasoning engine. It can effectively integrate multi-source heterogeneous knowledge and achieve unified organization and invocation, solving problems such as knowledge dispersion, poor model adaptability, and low decision-making efficiency in traditional scheduling. At the same time, relying on knowledge-driven methods, it realizes multi-model collaborative inference and multi-objective optimization, which can quickly respond to different scheduling scenarios and generate optimal scheduling schemes. It significantly improves the intelligence level, computational efficiency, and decision accuracy of water network system optimization scheduling decisions, and can better meet the complex scheduling needs of multi-source, multi-path, and multi-functional water network systems.

[0044] This embodiment takes the multi-objective optimization scheduling of water resources in the A watershed as an example, and provides a knowledge-driven water network system optimization scheduling decision-making method, which can be used on servers, terminals, mobile terminals, etc. The process includes the following steps: Step S201: Obtain knowledge about the water network system. For details, please refer to [link / reference needed]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0045] Step S202: Based on the knowledge of the water network system, construct a knowledge network for the water network system.

[0046] Specifically, step S202 includes: Step S2021: Extract the physical entities of the water network project, the entity relationships between the physical entities of the water network project, and the entity attributes of the physical entities of the water network project from the knowledge of the water network system.

[0047] Step S2022: Based on the physical entities and relationships of the water network project, construct a physical water network knowledge graph.

[0048] Step S2023: Embed entity attributes into the physical entities of water network projects in the physical water network knowledge graph; Step S2024: Encode the physical water network knowledge graph with embedded entity attributes to obtain the water network system knowledge network.

[0049] In this embodiment, based on the acquired knowledge of the water network system, core water network engineering physical entities are extracted through a combination of manual screening and machine recognition to construct a physical water network knowledge graph. Specifically, a data model can be built based on graph databases such as NebulaGraph, abstracting the physical entities and relationships within the watershed water network into a graph structure, expressed through "nodes" and "edges," and stored and managed using the NebulaGraph graph database. "Nodes" represent physical entities such as reservoirs, rivers, channels, sluices, and pumping stations, while "edges" represent the water flow direction relationships between entities. Then, using the constructed knowledge graph as an index, the engineering data, hydraulic parameters, scheduling rules, and operational data of the aforementioned water network engineering physical entities are further analyzed and embedded as entity attribute knowledge into the water network engineering physical entities. Entity attributes include engineering attributes, rule attributes, and state attributes. The constructed water network system knowledge network can be referenced... Figure 2 As shown.

[0050] Specifically, the engineering attributes of the physical nodes of the water network project include, for rivers or channels, cross-sectional shape, river length, slope coefficient, longitudinal slope, roughness coefficient, infiltration coefficient, and design flow capacity; for reservoirs, design water level, check water level, flood control limit water level, dead water level, normal storage water level, flood control capacity, dead storage capacity, outlet bottom elevation, and maximum outlet discharge; for gates, gate bottom elevation, gate top elevation, design water level, number of gate openings, gate size, design flow rate, single-gate design flow rate, check flow rate, normal water level, and head loss coefficient; and for pumping stations, installed flow rate, number of pumps, single pump flow rate, installed power, design head, and head loss coefficient.

[0051] The rule attributes for physical entity nodes in water network projects mainly include water level / head in rivers / channels and flow constraints (such as maximum overflow not exceeding...). m 3 / s, maximum water level not exceeding m), gate scheduling rules (e.g., the opening degree of each gate should not exceed m at any time). m, the water level in front of the sluice gate is lower than The gate is closed when m is reached, and the maximum overflow does not exceed [a certain value]. m 3 / s), pump station scheduling rules (e.g., pump station head not less than m, the water level upstream of the gate is not lower than *m, and the maximum flow rate does not exceed m 3 (s), etc.

[0052] The status attributes of physical entity nodes in water network projects mainly consist of historical, current, and simulated time series processes, including river / channel water level and flow, gate opening and flow rate, pumping station head and flow rate, and corresponding scheduling effects (such as water surface area and leakage).

[0053] Refer to Table 1 for an illustration of the settings for physical entity attribute fields in a water network project.

[0054] Table 1

[0055] Then, the encoding rule of "entity ID-attribute type-attribute value" is adopted to standardize all embedded attributes for use in the subsequent construction of water network system state reasoning knowledge and reasoning engine.

[0056] Numerical attributes (such as water level, flow rate, reservoir capacity, etc.) are retained to two decimal places and encoded using floating-point numbers (e.g., Res_001_Engineering Attribute_Design Water Level_109.00). Text-based attributes (such as cross-section type, scheduling rules, etc.): use string encoding (e.g., Can_001_project attribute_cross-section type_trapezoidal); Time series attributes (such as historical flow processes): The data is encoded using a "timestamp-value" array (e.g., Gate_001_Historical Traffic_2022 Spring Traffic_[[202206150800,320],[202206151000,650],...]). Complex structural attributes (such as cross-sectional morphology, water level-storage capacity curve, pump characteristic curve, and gate water level-opening-flow curve) are presented in JSON / array format.

[0057] In this embodiment, a physical water network knowledge graph is constructed by extracting physical entities, entity relationships, and attributes of water network projects. After embedding entity attributes into the graph and encoding them, a water network system knowledge network is formed. This enables unified representation and structured organization of multi-source heterogeneous information such as water network topology and physical parameters, effectively breaking down data silos and knowledge fragmentation problems. It provides standardized and efficient knowledge support for subsequent state reasoning, model calculation, and scheduling decisions, and improves the efficiency of knowledge retrieval in the water network system.

[0058] Step S203: Based on the knowledge network of the water network system, construct the state reasoning knowledge of the water network system.

[0059] Specifically, step S203 includes: Step S2031: Obtain the mechanism formulas, empirical relationships, input parameters and output variables on which the hydrodynamic process of the water network system is based.

[0060] Step S2032: Based on the mechanism formula, empirical relationship, input parameters and output variables, determine the computational entity and the characteristic attributes of the computational entity.

[0061] Based on the knowledge network of water network systems, the one-dimensional hydrodynamic process calculation method is created as a computational entity, mainly including two categories: hydrological mechanism formulas and empirical relations. The mechanism equations, empirical relation expressions, and their required input parameters (mainly including hydraulic engineering flow or water level constraints, hydrodynamic parameters, and flow and water level data, etc.) and output variables (mainly hydrodynamic processes such as flow and water level, as well as characteristic values ​​such as water surface area and leakage, etc.) are defined as the characteristic attributes of this computational entity and embedded.

[0062] The mechanism formula is the Saint-Venant equations, the fundamental governing equations of one-dimensional water flow, which include the continuity equation and the momentum equation:

[0063] In the formula, B represents the cross-sectional width of the water passage, in meters (m); Z represents the water level, in meters (m); t represents time, in seconds (s); and Q represents the flow rate, in cubic meters per second (m³). 3 / s; x represents the longitudinal distance of the canal along the main flow direction, in meters; q represents the lateral inflow, in meters. 3 / s·m; α represents the momentum correction factor; A represents the water flow area, in m². 2 g represents gravitational acceleration, m / s² 2 S f The friction ratio can be expressed by the following formula:

[0064] In the formula, n c R represents the Manning roughness coefficient of the water conveyance channel; R represents the hydraulic radius, in meters.

[0065] Based on empirical relationships derived from massive historical data, the following correlations are established between key characteristic indicators such as flow transmission time, flow loss, seepage, channel storage, and water surface area at upstream and downstream sections of a river and the scheduled flow and time: T i-j,t =f(O i,t ） ; Q 出i,t =f(O i入,t ） ; W渗漏i,t =f(O i入,t ） ; W 槽蓄i,t =f(O i入,t ） ; A i,t =f(O i,t ） ; In the formula, T i-j,t This represents the time required for water to travel from the i-th river section to the j-th river section when the flow rate is Q at time t, expressed in minutes. Q 出i,t This represents the outflow rate at the i-th river segment at time t, when the inflow rate at the i-th segment is Q, expressed in m³. 3 / s; W 渗漏i,t This represents the seepage rate of the river segment at time t when the inflow rate at the i-th cross-section is Q, expressed in cubic meters per second (m³). 3 ; W 槽蓄i,t This represents the channel storage capacity of the river segment when the inflow rate at the i-th segment is Q at time t, in cubic meters per second (m³). 3 / s; A i,t This represents the water surface area of ​​the river segment at time t when the inflow rate at the i-th cross-section is Q, in m². 3 / s.

[0066] Refer to Table 2 for an example of setting attribute fields for a calculated entity.

[0067] Table 2

[0068] Step S2033: Associate the computational entity with the physical entity of the water network project, and establish a mapping relationship between the characteristic attributes of the computational entity and the entity attributes of the physical entity of the water network project to obtain the state reasoning knowledge of the water network system.

[0069] In this embodiment, computational entities are associated with physical entities of water network engineering through "application" relationships, and a mapping is established between the attribute fields of computational entities and the attribute fields of corresponding physical entities of water network engineering. Finally, the computational entities, their feature attributes, relationships, and mapping tables are stored in the NebulaGraph graph database. The water network system knowledge network serves as a unified knowledge hub for storing data information and calling model parameters. With physical entities of water network engineering as the core, deep integration with water network system state reasoning knowledge is achieved through attribute field mapping between associated entities. Using the water flow direction as the reasoning path, the intelligent decision-making reasoning engine for water network system optimization scheduling calls the underlying knowledge network and water network system state reasoning knowledge to perform water network state reasoning from upstream to downstream.

[0070] Specifically, the construction of "application" relationships focuses on the point-edge storage mechanism based on the NebulaGraph graph database. A new "application (APPLIES_TO)" edge type is added to the existing physical water network knowledge graph. This establishes a corresponding association between computational entities and the constructed physical entities of the water network engineering project. The "application" relationship attribute (priority: 0 or 1) is set, using "0 or 1" to explicitly select the one-dimensional hydrodynamic process calculation method used for a specific node of the water network engineering physical entity. For example, if a river section needs to describe its flow dynamics through empirical relationships, the algorithm would be: Emp_TravelTime_001 - [APPLIES_TO] -> Riv_001.

[0071] For attribute field mapping, a two-way mapping rule of "input parameter - engineering entity attribute" and "output variable - engineering entity attribute" is established to ensure that the computational entity can automatically extract parameters and store calculation results from the water network engineering physical entity. An intermediate node for "attribute mapping" (coded Map_001) is created in the graph database, and the computational entity and the water network engineering physical entity are associated through the "mapping (MAPS_TO)" relationship. The mapping table node stores the structured data of the above mapping relationship and supports dynamic updates. For example, in the hydrological mechanism formula of the computational entity, the surface width B of the water-passing cross-section can be derived from the "cross-sectional form" attribute of the water network engineering physical entity (e.g., for a trapezoidal cross-section, it is calculated as "upper base width + 2 × water level × slope gradient"); the friction ratio S... f Manning roughness coefficient n mapped c And the hydraulic radius R according to formula S f =n c 2 Q 2 / (R) 4 / 3 A 2 The hydraulic radius R is calculated from the cross-sectional shape and water level (for example, if it is a trapezoidal cross-section, it is calculated according to...). Where b is the bottom width, h is the water depth, and m is the slope coefficient.

[0072] In this embodiment, by acquiring the mechanistic formulas, empirical relationships, and input / output parameters required for hydrodynamic process calculation, the computational entity and its characteristic attributes are determined. A mapping relationship is established between the computational entity and the physical entity of the water network project, and between the characteristic attributes of the computational entity and the attributes of the physical entity. This constructs water network system state reasoning knowledge, enabling flexible association between hydrodynamic calculation logic and the physical entity of the water network. Under conditions such as water network structure adjustment and simulation range changes, physical parameter matching and hydrodynamic calculation can be quickly achieved. Simultaneously, the core elements and associated logic of hydrodynamic calculation and decision optimization are standardized, allowing for flexible selection of computational objects, calculation methods, and adjustment of scheduling strategies. This solves the problems of deep binding between traditional hydrodynamic calculation and modeling methods and modeling range, and the separation between simulation pre-playing and decision optimization. It enables rapid modeling in multiple scenarios, providing accurate and efficient model support for water network state reasoning, scenario pre-playing, and scheduling decisions. While maintaining the accuracy of water network system state prediction and reasoning, it significantly improves computational efficiency, laying a solid model foundation for intelligent scheduling decisions.

[0073] Step S204: Construct multi-objective optimization scheduling decision knowledge for water networks; the multi-objective optimization scheduling decision knowledge for water networks includes the objective function of water network optimization scheduling, constraints, multi-objective optimization solution algorithm, and multi-attribute decision algorithm.

[0074] Specifically, step S204 includes: Step S2041: Determine the water network optimization and scheduling objectives. These objectives include the duration of water flow along the entire river channel, the maximum water surface area, the outflow volume, and the river channel leakage.

[0075] Step S2042: Determine the objective function for each water network optimization scheduling objective.

[0076]

[0077] In the formula, Indicates the time (in days) for the first full-line water flow through the Yongding River. This indicates the time (in days) for the Yongding River to be fully navigable. This represents the penalty function.

[0078]

[0079] In the formula, Indicates the first Duanheduan Water surface area at any given time (ha), Indicates when to start replenishing water (days). Indicates the end time of water replenishment (in days). This represents the penalty function.

[0080]

[0081] In the formula, This indicates the outflow rate (m³ / s) at the Yongding River's exit section. Indicates when water replenishment begins (days). Indicates the end time for water replenishment (in days). This represents the penalty function.

[0082]

[0083] In the formula, Indicates the first Duanheduan Infiltration flow rate at any given time (m³ / s), Indicates the start time (in days) for water replenishment. Indicates the end time of water replenishment (in days). This represents the penalty function.

[0084] All the above scheduling target values ​​are derived from the water network system knowledge network and the water network system state reasoning knowledge through the intelligent decision reasoning engine for water network system optimization scheduling.

[0085] Step S2043: Based on the knowledge network of the water network system and the state reasoning knowledge of the water network system, determine the constraints.

[0086] Constraints include flow rate, water level, flow velocity, gate capacity, and water balance constraints.

[0087] The specific constraints can be as follows:

[0088] In the formula, Indicates the lower boundary of river flow. Indicates the upper boundary of river flow. This indicates the scheduling of traffic.

[0089]

[0090] In the formula, Indicates the lower boundary of the river channel water level. Indicates the upper boundary of the river channel water level. Indicates the water level for scheduling.

[0091]

[0092] In the formula, Indicates the lower boundary of the river flow velocity. Indicates the upper boundary of the river flow velocity. This indicates the scheduling flow rate.

[0093]

[0094] In the formula, Indicates the amount of water flowing into the river channel. Indicates the inflow of water at the cross-section of the river channel. Indicates river evaporation, Indicates the amount of water seepage in the river channel Indicates the outflow volume of the lower section of the river channel. Indicates the amount of water taken from outside the river channel. This indicates the change in water storage in the river channel.

[0095]

[0096] In the formula, Indicates the gate station number. Indicates the number is The maximum discharge capacity of the sluice gate. Indicates the number is The actual discharge flow of the gate station.

[0097]

[0098] In the formula, Indicates the reservoir number, Indicates the number is The reservoir is at a normal high water level. Indicates the number is The dead water level of the reservoir.

[0099] Non-negativity constraint: All parameters satisfy the non-negativity constraint.

[0100] The above constraints are derived from the knowledge network of the water network system and the state reasoning knowledge of the water network system, through the intelligent decision reasoning engine for the optimization and scheduling of the water network system.

[0101] Step S2044: Determine the improved multi-objective particle swarm optimization algorithm and use it as the multi-objective optimization solution algorithm.

[0102] In some optional implementations, the improved multi-objective particle swarm optimization algorithm includes: employing a dimensional decoupling strategy of pre-storing segmented days and optimizing single-granularity flow, whereby the day parameter is pre-stored and decoupled from the particle optimization dimension, and the particles optimize only for the single-granularity flow dimension, and the high-dimensional vector is reconstructed by pre-storing the days during the fitness calculation stage; and a flexible constraint handling strategy based on violation degree and penalty function, which quantifies the degree of constraint violation by particles and applies adaptive penalties to particles that violate constraints.

[0103] In this embodiment, an improved multi-objective particle swarm optimization algorithm is used to solve the objective function and perform iterative optimization. The key is to improve the particle initialization mechanism and constraint handling strategy.

[0104] Specifically, the granularity dimension optimization of the initial particles adopts a dimension decoupling strategy of "segmented day pre-storage + single-granularity flow optimization". The day parameter is separated from the particle optimization dimension and pre-stored. The particles only optimize the single-granularity flow parameter. In the fitness calculation stage, the high-dimensional vector is reconstructed by pre-stored days, realizing an efficient search mode of "low-dimensional optimization and high-dimensional adaptation".

[0105] The specific implementation method is as follows: First, parameter splitting and pre-storage: In the early stages of algorithm initialization, the segmented day parameters in the optimization task are extracted and pre-stored separately, forming independent day vectors. Second, particle dimension simplification: In the particle initialization stage, only single-granularity flow parameters are encoded, simplifying the particle dimension from the traditional "flow dimension + day dimension" to a single flow dimension. Third, high-dimensional reconstruction in the fitness calculation stage: When evaluating particle fitness, the single-granularity flow vector is expanded in dimension based on the pre-stored day vector, reconstructing it into a high-dimensional flow vector that matches the actual optimization scenario. The specific reconstruction rule is: each flow value is repeatedly concatenated according to the corresponding segment's day number to form a continuous time-series flow vector.

[0106] For example, the initial encoding of a particle is a hybrid vector of flow rate and number of days, such as [2.1, 2.3, 6.1, 4.2, 2,3, 1, 4]. The first four elements are the flow rate values ​​for each segment, and the last four elements are the duration of the corresponding segment in days. The pre-stored number of days vector after splitting is [2,3,1,4]. This vector remains fixed throughout the optimization process and no longer participates in the iterative update of the particle. The improved initial encoding of the particle is a pure flow rate vector [2.1, 2.3, 6.1, 4.2]. The particle dimension is directly halved, significantly reducing the search space complexity. During the reconstruction process, the flow value 2.1 corresponds to day 2 and is repeated twice; the flow value 2.3 corresponds to day 3 and is repeated three times; the flow value 6.1 corresponds to day 1 and is repeated once; and the flow value 4.2 corresponds to day 4 and is repeated four times. Finally, a high-dimensional flow vector [2.1, 2.1, 2.3, 2.3, 2.3, 6.1, 4.2, 4.2, 4.2, 4.2] is obtained, and then the fitness is calculated based on this high-dimensional vector.

[0107] A flexible constraint handling strategy based on violation degree and penalty function: The flexible constraint handling strategy of "constraint violation degree calculation + penalty function punishment" is adopted. By quantifying the degree of constraint violation of particles, adaptive punishment is applied to particles that violate the constraints. This retains potential high-quality search particles and guides particles to iterate in the direction of satisfying the constraints through the punishment mechanism.

[0108] The specific implementation steps are as follows: Constraint Analysis and Violation Index Definition: First, clarify the various constraints in the flow optimization task (such as flow upper and lower limits, temporal continuity constraints, etc.), and define corresponding violation indexes for each type of constraint. Second, comprehensive calculation of particle constraint violation: For each iterative particle, traverse all constraints, calculate its violation under each constraint, and then obtain the particle's comprehensive violation by weighted summation. The weight coefficients can be set according to the importance of different constraints to ensure that important constraints have a greater impact on the particle's violation. Third, design and application of penalty function based on violation: Construct a penalty function positively correlated with the comprehensive violation, and integrate the penalty function value into the particle's fitness function according to the target optimization direction to achieve precise adaptive punishment for particles that violate constraints.

[0109] For example, the core constraint for traffic optimization is that the traffic value must be within the upper and lower boundaries [1, 10], meaning that the traffic value x in any segment must satisfy 1 ≤ x ≤ 10. The violation calculation formula is defined as follows: When x > upper boundary (10), the violation degree V1(x) = (x - upper boundary) / upper boundary; When x < lower boundary (1), the violation degree V2(x) = (lower boundary - x) / lower boundary; When 1≤x≤10, the degree of violation V(x)=0; If multiple constraints exist (such as flow fluctuation range constraints, temporal continuity constraints, etc.), the violation degree under each constraint needs to be calculated separately according to the above logic, and then the overall violation degree V of the particle is obtained by weighted summation. total (x), the formula is: V total (x) =ω1V1(x) +ω2V2(x) +... +ωV(x); Where ω1, ω2, ..., ω are the weight coefficients of each constraint, satisfying Σω i =1 (i=1,2,...,n), the weight can be set according to the importance of the constraint. For example, the weight of the core constraint (upper and lower boundaries of traffic) can be set to 0.6, and the weight of the secondary constraint (fluctuation range of traffic) can be set to 0.4.

[0110] The core design logic of the penalty function is "the higher the degree of violation, the stronger the penalty," while also needing to adapt the penalty method to the objective direction (maximization or minimization) of the multi-objective optimization. Let the k objective functions of the multi-objective optimization be f1(x), f2(x), ..., f(x), where some objectives are maximization objectives (e.g., groundwater recharge) and some are minimization objectives (e.g., outflow of water). The corresponding penalty function fusion formula is as follows: For minimizing the objective f i (x), the fitness value after penalty is: f i '(x) = f i(x)+λ·V total (x); For maximizing the objective f(x), the fitness value after penalty is: f'(x) = f(x) -λ·V total (x); Where λ is the penalty coefficient (λ>0), used to adjust the penalty intensity, which can be set according to the optimization scenario (e.g., initially set to 10, and adjusted to the optimal penalty effect through trial calculation); V total (x) represents the overall violation degree of the particle.

[0111] If the flow rate of a certain segment of a particle is 12, it obviously violates the constraint. The degree of violation is calculated as: V1(12) = (12-10) / 10 = 0.2, that is, the degree of violation of the particle under this constraint is 0.2. Assume that the original fitness value f of the minimization objective (flow loss) corresponding to the particle is... i (x)=5, overall violation degree V total With x = 0.2 and a penalty coefficient λ = 10, the fitness value f after penalty can be obtained. i The initial fitness value f'(x) = 5 + 10 × 0.2 = 7, which is worse than the original fitness value. If the initial fitness value f(x) of the particle corresponding to a certain maximization target (such as groundwater recharge) is 8, substituting it into formula 6, we get the penalized fitness value f'(x) = 8 - 10 × 0.2 = 6, which is also worse than the original fitness value. Through this penalty mechanism, the probability of particles that violate the constraints being selected in the population iteration is reduced, thereby guiding particles to evolve towards "satisfying constraints + good fitness". For particles that satisfy all constraints, their comprehensive violation degree V total When (x)=0, the penalty function value is 0, and the fitness value remains the original value, ensuring that high-quality feasible solutions are not penalized.

[0112] Step S2045: The multi-attribute decision algorithm is used to determine the weight of the decision target by subjective and objective weighting method, normalize the non-dominated solution set, and determine the proximity based on weighted Euclidean distance to realize the ranking of scheduling decision schemes.

[0113] The decision objective weights are determined by using a weighted method that employs both subjective and objective weights. Subjective weights are determined based on different scheduling strategies of the decision-makers, while objective weights are determined using the entropy weight method. Simultaneously, the non-dominated solution set data matrix obtained by the multi-objective optimization algorithm is normalized, and the weighted Euclidean distance between each decision objective and the optimal and worst solutions is calculated to determine the proximity, thus ranking the schemes.

[0114] The objective weights are calculated using the entropy weight method, and the calculation equation for any non-dominated solution set is as follows:

[0115] For calculating the overall weight, the subjective weight can be set as follows: Objective weight is The overall weight is The smaller the deviation caused by the differences in the three weight evaluations, the better the comprehensive weight reflects the information expressed by the subjective and objective weights. Therefore, a least squares comprehensive weight calculation method can be established:

[0116] In the formula, This represents the result of the objective function for the i-th scheme and the j-th objective function. Indicates the number of solutions. Indicates the number of scheduling targets. This represents the objective weight of the j-th objective. This represents the subjective weight of the j-th objective. This represents the overall weight of the j-th objective.

[0117] The non-dominated solution set data matrix obtained by the multi-objective optimization algorithm is normalized. The cosine method is then used to find the optimal and worst solutions among the finite options. The distances between each evaluated object and the optimal and worst solutions are then calculated to obtain the relative closeness of each evaluated object to the optimal solution, which serves as the basis for ranking the solutions. The calculation method for any non-dominated solution set is as follows:

[0118] In the formula, This represents the result of the objective function for the i-th scheme and the j-th objective function. This represents the subjective weight of the j-th objective. This indicates the relative similarity between the i-th solution and the optimal solution. A larger value indicates that the solution is closer to the optimal solution, and the solution is closer to the optimal solution that satisfies the scheduling strategy.

[0119] In this embodiment, by constructing multi-objective optimization scheduling decision knowledge for water networks, the scheduling objectives and corresponding objective functions are clarified. The constraints are accurately determined by relying on the knowledge network of water network systems and the state reasoning knowledge model. At the same time, an improved multi-objective particle swarm optimization algorithm is used as the solution algorithm, and a multi-attribute decision algorithm is used as the scheme ranking algorithm. This effectively solves the problems of single optimization objectives, insufficient multi-objective solution efficiency and insufficient global optimization capability in traditional water network scheduling.

[0120] Step S205: Based on the water network system knowledge network, water network system state reasoning knowledge, and water network multi-objective optimization scheduling decision knowledge, a multi-level knowledge fusion intelligent decision reasoning engine for water network system optimization scheduling is constructed to generate water network system scheduling decision schemes.

[0121] Specifically, step S205 includes: Step S2051: Construct a knowledge graph information fusion layer based on the water network system knowledge network to enable rapid organization and retrieval of water network topology, water network physical parameters, rule constraints and real-time status.

[0122] Step S2052: Construct a predictive inference collaborative computing layer based on the state reasoning knowledge of the water network system. Based on the data provided by the knowledge graph information fusion layer, dynamically call and respond to predictive inference tasks to provide candidate solutions for decision-making.

[0123] Step S2053: Construct a decision optimization reasoning layer based on the multi-objective optimization scheduling decision knowledge of the water network. Based on the water network system state prediction and deduction of the prediction and deduction collaborative computing layer, dynamically adjust the weight coefficients of multiple optimization objectives according to the real-time scenario and scheduling strategy, solve the problem using a multi-objective optimization algorithm, generate a Pareto optimal solution set, and recommend and rank the decision schemes based on the scheduling strategy. The water network system optimization scheduling decision reasoning engine includes a knowledge graph information fusion layer, a prediction and deduction collaborative computing layer, and a decision optimization reasoning layer.

[0124] Reference Figure 3 As shown, a multi-layered knowledge fusion-based intelligent decision-making reasoning engine for water network optimization scheduling is constructed. Using the underlying knowledge network of the water network system as an index, it integrates heterogeneous knowledge from multiple sources, including mechanism formulas and empirical relationships. The engine fuses knowledge from different layers for decision-making reasoning: The first layer is a knowledge graph-based "physical-rule-state" fusion layer, enabling rapid organization and invocation of the water network topology, physical parameters, rule constraints, and real-time states, providing a foundation for upper-layer model computation. The second layer is a multi-model coupling-based prediction and deduction collaborative computing layer, based on the parameters, rules, and data provided by the first-layer knowledge network, dynamically invoking prediction and deduction tasks issued by the top-level decision-making team and providing candidate solutions for the top-level decision-making team. The third layer is a dynamic weight adjustment-based strategy optimization and evolution layer, based on the second-layer water network system state prediction and deduction, dynamically adjusting the weight coefficients of multiple optimization objectives according to the real-time scenario and scheduling strategy, using a multi-objective optimization algorithm to solve the problem, generating a Pareto optimal solution set, and recommending and ranking decision solutions based on the scheduling strategy, achieving a closed-loop decision-making process of "scenario-objective-strategy-solution". A 4D graph of the Pareto solution set for a certain scheduling scenario is shown below. Figure 4 As shown, the optimal decision-making result for a certain scheduling scenario is referenced. Figure 5 As shown, the specific scheduling process of different strategies under a certain scheduling scenario is described in reference. Figure 6 As shown.

[0125] The main steps are as follows: The first layer is the construction of the water network "physical-rule-state" fusion inference model (knowledge graph information fusion layer): First, based on the NebulaGraph knowledge graph library, the Path Ranking Algorithm is used to realize the fast query and traversal of topology and physical parameters. At the same time, the water network scheduling rules are transformed into production rules (IF-THEN form). Furthermore, the attribute field mapping bidirectional mapping rules are used to dynamically update the water network status data (water level, flow, pressure, etc.) to the corresponding node attributes of the knowledge graph, realizing the dynamic fusion and real-time call of the water network "physical-rule-state" data.

[0126] The second layer is the construction of the future state inference model of the water network system (prediction and inference collaborative computing layer): based on the prediction and inference task type issued by the top-level decision and the output results of the water network "physical-rule-state" fusion layer inference model constructed in the first layer, the hydrological model is adaptively selected based on the Deep Q-Network (DQN) reinforcement learning algorithm to generate the future state inference results of the water network system (time series data such as flow, water level, pressure, and target).

[0127] The third layer involves constructing a decision optimization reasoning model (decision optimization reasoning layer). Based on the multi-objective particle swarm optimization algorithm, it uses the results generated from the future state reasoning model of the water network system constructed in the second layer as the initial population. Through iterative calculation, it generates a Pareto optimal solution set, which satisfies the characteristic that "no solution is superior to other solutions in all objectives." Simultaneously, using adjusted objective weights as constraints, it comprehensively evaluates the Pareto optimal solution set through an ideal solution ranking method, calculating the closeness of each solution to the ideal solution, and generating a recommended solution list from high to low closeness.

[0128] This embodiment constructs a multi-layered knowledge fusion and reasoning engine comprising a knowledge graph information fusion layer, a prediction and deduction collaborative computing layer, and a decision optimization and reasoning layer. This engine enables unified organization and efficient retrieval of multi-source heterogeneous knowledge within the water network. It can dynamically complete water network topology reasoning, state prediction and deduction, and scheduling scheme generation based on real-time scenarios. Furthermore, it dynamically adjusts target weights based on scheduling strategies to perform multi-objective optimization and scheme ranking, effectively improving the intelligence level and response speed of water network scheduling decisions, and enhancing the timeliness, accuracy, and practicality of decisions under changing conditions. It can perform flexible modeling, efficient simulation calculations, and multi-objective decision optimization based on a knowledge base, offering advantages such as flexible modeling, high computational efficiency, and accurate decision-making.

[0129] This invention proposes a method for unifying and representing water network physical parameters, operational data, simulation models, and expert experience into a unified knowledge base, and using this to drive real-time reasoning, multi-objective optimization, and adaptive decision-making for water network status. This method can deeply integrate multi-source knowledge and achieve rapid modeling, accurate simulation, effect evaluation, and multi-objective global optimization based on a knowledge base. It can effectively improve the comprehensive decision support capability of water network systems in coping with uncertain changes, dynamic adjustment in multiple scenarios, and multi-objective collaborative protection, and improve the accuracy of water network system status simulation and prediction, as well as the efficiency of multi-objective scheduling decision optimization.

[0130] This embodiment also provides a knowledge-driven water network system optimization scheduling decision-making 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 implements 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.

[0131] This embodiment provides a knowledge-driven water network system optimization scheduling decision-making device, such as... Figure 7 As shown, it includes: Module 301 is used to acquire knowledge about the water network system; The water network system knowledge network construction module 302 is used to construct a water network system knowledge network based on water network system knowledge. The water network system state reasoning knowledge construction module 303 is used to construct water network system state reasoning knowledge based on the water network system knowledge network. The water network multi-objective optimization scheduling decision knowledge construction module 304 is used to construct water network multi-objective optimization scheduling decision knowledge; the water network multi-objective optimization scheduling decision knowledge includes water network optimization scheduling objective function, constraints, multi-objective optimization solution algorithm, and multi-attribute decision algorithm; The intelligent decision-making reasoning engine construction module 305 for water network system optimization scheduling is used to build a multi-level knowledge fusion intelligent decision-making reasoning engine for water network system optimization scheduling based on water network system knowledge network, water network system state reasoning knowledge, and water network multi-objective optimization scheduling decision knowledge, and generate water network system scheduling decision schemes.

[0132] In some optional implementations, the water network system knowledge network construction module 302 is specifically used for: Extract the physical entities of water network projects, the entity relationships between physical entities of water network projects, and the entity attributes of physical entities of water network projects from the knowledge of water network systems; Based on the physical entities and relationships of water network engineering, a physical water network knowledge graph is constructed. Embed entity attributes into physical entities of water network engineering in the physical water network knowledge graph; Encode the physical water network knowledge graph with embedded entity attributes to obtain the water network system knowledge network.

[0133] In one optional implementation, the water network system state reasoning knowledge construction module 303 is specifically used for: To obtain the underlying mechanisms, empirical relationships, input parameters, and output variables for calculating the hydrodynamic processes of a water network system; Based on mechanistic formulas, empirical relationships, input parameters, and output variables, the computational entity and its characteristic attributes are determined. By associating computational entities with physical entities of water network engineering, and establishing a mapping relationship between the characteristic attributes of computational entities and the entity attributes of physical entities of water network engineering, we can obtain state reasoning knowledge of the water network system.

[0134] In one optional implementation, the water network multi-objective optimization scheduling decision knowledge construction module 304 is specifically used for: Define the objectives for optimizing water network scheduling; Determine the objective function for water network optimization scheduling for each water network optimization scheduling objective; Based on the knowledge network of water network system and the state reasoning knowledge of water network system, the constraints are determined. An improved multi-objective particle swarm optimization algorithm was determined and used as the multi-objective optimization solution algorithm. The multi-attribute decision algorithm is used to determine the weight of the decision target by subjective and objective weighting method, normalize the non-dominated solution set, and determine the proximity based on weighted Euclidean distance to realize the ranking of scheduling decision schemes.

[0135] In one alternative implementation, the improved multi-objective particle swarm optimization algorithm includes: A dimensional decoupling strategy of segmented day pre-storage and single-granularity flow optimization is adopted. The day parameter is decoupled from the particle optimization dimension and pre-stored. The particles only optimize the single-granularity flow dimension. The high-dimensional vector is reconstructed by pre-stored days in the fitness calculation stage. A flexible constraint handling strategy based on violation degree and penalty function is used to adaptively penalize particles that violate constraints by quantifying the degree of constraint violation of particles.

[0136] In one optional implementation, the intelligent decision-making reasoning engine construction module 305 for optimizing and scheduling water network systems is specifically used for: A knowledge graph information fusion layer is constructed based on the knowledge network of the water network system to enable rapid organization and retrieval of water network topology, physical parameters, rule constraints and real-time status; A predictive inference collaborative computing layer is constructed based on the state reasoning knowledge of the water network system. The predictive inference task is dynamically invoked and responded to based on the data provided by the knowledge graph information fusion layer, so as to provide candidate solutions for decision-making. Based on the knowledge of multi-objective optimization scheduling in water networks, a decision optimization reasoning layer is constructed. Based on the prediction and deduction of the water network system state in the prediction and deduction collaborative computing layer, the weight coefficients of multiple optimization objectives are dynamically adjusted according to the real-time scenario and scheduling strategy. The multi-objective optimization solution algorithm is used to solve the problem, generate a Pareto optimal solution set, and recommend and rank the decision solutions based on the scheduling strategy. The water network system optimization scheduling decision reasoning engine includes a knowledge graph information fusion layer, a prediction and inference collaborative computing layer, and a decision optimization reasoning layer.

[0137] The knowledge-driven water network system optimization scheduling decision-making device provided in this embodiment of the invention can execute the knowledge-driven water network system optimization scheduling decision-making method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0138] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0139] The following is a detailed reference. Figure 8 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 402 or a program loaded from memory 408 into random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the electronic device. The processor 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0140] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0141] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408, or installed from a ROM 402. When the computer program is executed by the processor 401, it performs the functions defined in the knowledge-driven water network system optimization scheduling decision-making method of the embodiments of the present invention.

[0142] Figure 8 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0143] 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. When the software or computer code is accessed and executed by the computer, processor, or hardware, the knowledge-driven water network system optimization scheduling decision-making method shown in the above embodiments is implemented.

[0144] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0145] 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 water network system optimization scheduling decision-making method, characterized in that, The method includes: Acquire knowledge of water network systems; Based on the knowledge of the water network system, a knowledge network of the water network system is constructed; Based on the aforementioned knowledge network of the water network system, construct state reasoning knowledge of the water network system; Constructing multi-objective optimization scheduling decision knowledge for water networks; the multi-objective optimization scheduling decision knowledge for water networks includes water network optimization scheduling objective function, constraints, multi-objective optimization solution algorithm, and multi-attribute decision algorithm; Based on the knowledge network of the water network system, the state reasoning knowledge of the water network system, and the multi-objective optimization scheduling decision knowledge of the water network, a multi-level knowledge fusion intelligent decision reasoning engine for water network system optimization scheduling is constructed to generate water network system scheduling decision schemes.

2. The method according to claim 1, characterized in that, The construction of a water network system knowledge network based on the knowledge of the water network system includes: Extract the physical entities of the water network project, the entity relationships between the physical entities of the water network project, and the entity attributes of the physical entities of the water network project from the knowledge of the water network system. Based on the physical entities of the water network project and the relationships between these entities, a physical water network knowledge graph is constructed. Embed the entity attributes into the physical entity of the water network project; The physical water network knowledge graph embedded with the entity attributes is encoded to obtain the water network system knowledge network.

3. The method according to claim 2, characterized in that, Based on the existing water network system knowledge network, the state reasoning knowledge of the water network system is constructed, including: To obtain the underlying mechanisms, empirical relationships, input parameters, and output variables for calculating the hydrodynamic processes of a water network system; Based on the aforementioned mechanism formula, the aforementioned empirical relationship, the aforementioned input parameters, and the aforementioned output variables, the computational entity and the characteristic attributes of the computational entity are determined. The computational entity is associated with the physical entity of the water network project, and a mapping relationship is established between the feature attributes of the computational entity and the entity attributes of the physical entity of the water network project to obtain the state reasoning knowledge of the water network system.

4. The method according to claim 1, characterized in that, The knowledge for constructing multi-objective optimization scheduling decisions for water networks includes: Define the objectives for optimizing water network scheduling; Determine the water network optimization scheduling objective function for each of the aforementioned water network optimization scheduling objectives; The constraints are determined based on the knowledge network of the water network system and the state reasoning knowledge of the water network system. An improved multi-objective particle swarm optimization algorithm is determined, and the improved multi-objective particle swarm optimization algorithm is used as the multi-objective optimization solution algorithm. The multi-attribute decision algorithm is used to determine the weight of the decision target by subjective and objective weighting method, normalize the non-dominated solution set, and determine the proximity based on weighted Euclidean distance to realize the ranking of scheduling decision schemes.

5. The method according to claim 4, characterized in that, The improved multi-objective particle swarm optimization algorithm includes: A dimensional decoupling strategy of segmented day pre-storage and single-granularity flow optimization is adopted. The day parameter is decoupled from the particle optimization dimension and pre-stored. The particles only optimize the single-granularity flow dimension. The high-dimensional vector is reconstructed by pre-stored days in the fitness calculation stage. A flexible constraint handling strategy based on violation degree and penalty function is used to adaptively penalize particles that violate constraints by quantifying the degree of constraint violation of particles.

6. The method according to claim 1, characterized in that, The construction of a multi-level knowledge fusion-based water network system optimization scheduling decision reasoning engine, based on the water network system knowledge network, the water network system state reasoning knowledge, and the water network multi-objective optimization scheduling decision knowledge, includes: A knowledge graph information fusion layer is constructed based on the knowledge network of the water network system to enable rapid organization and retrieval of water network topology, physical parameters, rule constraints and real-time status; Based on the state reasoning knowledge of the water network system, a predictive inference collaborative computing layer is constructed. Based on the data provided by the knowledge graph information fusion layer, the predictive inference task is dynamically invoked and responded to, so as to provide candidate solutions for decision-making. Based on the multi-objective optimization scheduling decision knowledge of the water network, a decision optimization reasoning layer is constructed. Based on the water network system state prediction and deduction of the prediction and deduction collaborative computing layer, the weight coefficients of multiple optimization objectives are dynamically adjusted according to the real-time scenario and scheduling strategy. The multi-objective optimization solution algorithm is used to solve the problem, generate a Pareto optimal solution set, and recommend and rank the decision schemes based on the scheduling strategy. The water network system optimization scheduling decision reasoning engine includes the knowledge graph information fusion layer, the prediction and inference collaborative computing layer, and the decision optimization reasoning layer.

7. A knowledge-driven water network system optimization scheduling decision-making device, characterized in that, The device includes: The acquisition module is used to acquire knowledge about the water network system. A water network system knowledge network construction module is used to construct a water network system knowledge network based on the water network system knowledge. The water network system state reasoning knowledge construction module is used to construct water network system state reasoning knowledge based on the water network system knowledge network. A knowledge construction module for multi-objective optimization scheduling of water networks is used to construct multi-objective optimization scheduling knowledge for water networks; the multi-objective optimization scheduling knowledge for water networks includes the objective function of water network optimization scheduling, constraints, multi-objective optimization solution algorithm, and multi-attribute decision algorithm. The intelligent decision-making reasoning engine construction module for water network system optimization scheduling is used to construct a multi-level knowledge fusion intelligent decision-making reasoning engine for water network system optimization scheduling based on the water network system knowledge network, the water network system state reasoning knowledge, and the water network multi-objective optimization scheduling decision knowledge, and generate water network system scheduling decision schemes.

8. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the knowledge-driven water network system optimization scheduling decision-making method according to 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 the computer to execute the knowledge-driven water network system optimization scheduling decision-making method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer instructions for causing a computer to execute the knowledge-driven water network system optimization scheduling decision-making method as described in any one of claims 1 to 6.

Images