Real-time scheduling visual dynamic modeling method and system for watershed complex water engineering system

Abstract

The invention belongs to the crossing field of hydrology and water resource scheduling and water conservancy informatization, and discloses a basin complex water engineering system real-time scheduling visualization dynamic modeling method and system, and the method comprises the steps: constructing a real-time scheduling topology visualization graph of a basin water engineering system; based on the real-time scheduling topology visualization graph, performing fusion configuration on water engineering node characteristic parameters and river section node characteristic parameters; constructing a composite flood routing model based on the node characteristic parameters of the fusion configuration; and constructing an intelligent scheduling model based on the composite flood routing model. According to the invention, through combination of a computer graphical interaction technology and a multi-model coupling algorithm, full-process digital management of water engineering scheduling is realized.

Description

Technical Field

[0001] The present invention belongs to the cross-field of hydrology and water resources scheduling and water conservancy informationization, and specifically relates to a visual dynamic modeling method and system for real-time scheduling of complex water engineering systems in a river basin. Background Art

[0002] Real-time scheduling of complex water engineering systems in a river basin is a core technical means to ensure flood control safety and optimize water resource allocation. With the intensification of global climate change and the frequent occurrence of extreme hydrological events, traditional scheduling systems have been unable to cope with the dynamic needs of multi-engineering coordination and multi-objective optimization. Especially in the case of major floods, how to quickly build a scheduling topology that includes a group of reservoirs, a dam network, and a linkage of flood storage areas, and achieve a plan deduction with a minute-level response capability, is directly related to the safety and effectiveness of the river basin flood control system. At present, the world's largest water engineering cluster has been built in my country's key river basins, but the timeliness and accuracy of scheduling decisions still face major technical challenges.

[0003] Most existing water project dispatching systems use predefined static topology architectures, and their modeling process relies on manual coding and offline configuration. When the topology needs to be temporarily adjusted due to the renovation and expansion of engineering facilities or sudden emergencies, the system must rewrite the underlying connection logic. Actual application data shows that the modeling cycle of such scenarios is generally measured in weeks or months, which seriously restricts emergency response capabilities. More importantly, the traditional visualization interface can only realize the planar display of topological relationships and lacks a dynamic coupling mechanism with the hydraulic calculation model, resulting in the formation of "data islands" in the formulation of dispatching plans and simulation verification.

[0004] In terms of model integration, existing technologies have not yet overcome the difficulty of collaborative computing of heterogeneous models. The reservoir operation model (based on the principle of water balance) and the river evolution model (requires solving the hydrodynamic equation) often cause conflicts in calculation results due to differences in algorithm principles and spatiotemporal scales. Especially when facing a group of hybrid projects, the system cannot autonomously identify the topological structure characteristics to adapt to the joint operation model, resulting in more than 30% of the operation plans requiring manual intervention and correction. This phenomenon is particularly prominent in cross-provincial basin operation, which seriously threatens the reliability and execution efficiency of the operation instructions.

[0005] The current system architecture generally has scalability bottlenecks. When adding new scheduling models or adjusting calculation processes, the core code needs to be reconstructed, and each iterative upgrade takes up to 2-3 months. The verification mechanism also remains at the level of single model parameter verification, lacking full-link verification of topological integrity, model coupling, and scheme feasibility. A 2022 actual combat exercise in a certain basin showed that the deduction error caused by the mismatch between the time step of the dam model and the river model was as high as 25%, exposing major defects in the traditional verification system. In addition, when the topological structure changes dynamically (such as the emergency activation of flood diversion facilities), the existing system takes more than 2 hours to re-define the calculation boundaries, which makes it difficult to meet the minute-level response requirements of the "four predictions" (forecast, warning, rehearsal, and plan) for flood control.

[0006] In view of the above problems, the present invention proposes a new visual dynamic modeling method and system for real-time scheduling of complex water engineering systems in a river basin. Summary of the invention

[0007] In order to solve the problems existing in the prior art, the present invention provides a visual dynamic modeling method and system for real-time scheduling of complex water engineering systems in a river basin, which combines computer graphical interaction technology with a multi-model coupling algorithm to realize full-process digital management of water engineering scheduling.

[0008] To achieve the above object, the present invention provides the following solutions:

[0009] A method for visualizing dynamic modeling of real-time scheduling of a complex water engineering system in a river basin, the method comprising:

[0010] Construct a real-time scheduling topology visualization diagram of the basin water engineering system;

[0011] Based on the real-time scheduling topology visualization diagram, the characteristic parameters of the water project nodes and the characteristic parameters of the river section nodes are integrated and configured;

[0012] Based on the node characteristic parameters of the fusion configuration, a composite flood evolution model is constructed;

[0013] Based on the composite flood evolution model, an intelligent scheduling model is constructed.

[0014] Preferably, the method for constructing a real-time scheduling topology visualization diagram of a watershed water engineering system includes:

[0015] Construct flood forecasting topological base layer based on river basin GIS data;

[0016] Based on the flood forecasting topological base layer, a multi-dimensional extensible graphic element library for basin water project scheduling is constructed;

[0017] Based on the multi-dimensional extensible graphic element library, dynamic topological linking is implemented to complete the construction of a real-time scheduling topological visualization diagram of the watershed water engineering system.

[0018] Preferably, based on the flood forecasting topological base layer, the method for constructing a multi-dimensional extensible primitive library includes:

[0019] Based on the physical characteristics and control behaviors of engineering entities, a composite parameter system including geometric parameters, hydraulic parameters, and control logic parameters is designed for each type of element;

[0020] At the same time, the dynamic injection of new engineering features is achieved through multi-layer open extension interfaces, and the parameter dimensions are customized according to the characteristics of the watershed, so as to establish a structured database that supports the topological reconstruction and intelligent calculation of complex water networks;

[0021] Among them, the methods of dynamically injecting new engineering features through multi-layer open extension interfaces, customizing parameter dimensions according to basin characteristics, and establishing a structured database that supports complex water network topology reconstruction and intelligent computing include: hierarchical interface protocol design, dynamic parameter binding and conflict detection mechanism, and intelligent parameter recommendation system construction;

[0022] The layered interface protocol design includes: core interface layer, extended interface layer and watershed adaptation layer; the core interface layer is used to forcibly inherit the hydraulic conductivity properties and spatial topological constraints of the basic graphics elements; the extended interface layer is used to declare the physical dimensions, value range and data verification rules of the newly added parameters based on the domain-specific language DSL, supporting non-programmers to define parameters through configuration files; the watershed adaptation layer has built-in terrain slope , soil permeability coefficient , river network density A watershed feature parameterization template for dynamically generating parameter legitimacy functions;

[0023] Dynamic parameter binding and conflict detection mechanisms include: Symbolic binding of hydraulic models: Automatic analysis of new parameters using symbolic calculation engine Hydraulic control equation The partial derivative relationship of OK The Jacobian matrix of the columns, Ensure that parameters are automatically integrated into the computing kernel after injection, without the need to reconstruct the model code; build a parameter dependency network based on a directed acyclic graph, detect contradictory combinations through topological sorting, trigger real-time alarms and lock the computing process, and implement conflict self-checking through mathematical symbol binding and graph theory algorithms;

[0024] For hydraulic structures, an intelligent parameter recommendation system based on engineering feature fingerprints and cross-engineering topology association learning is constructed, which includes: extracting core parameters for each type of hydraulic structure, and generating unique feature fingerprint codes by weighting them through the random forest algorithm. ,in, is the parameter importance weight, It is the core physical parameter value extracted from the project, which is used to quantify the structural or functional characteristics of different projects. Based on the graph convolutional network GCN, a topological map of hydraulic structures is constructed. The nodes are the project feature fingerprints, and the edges are the river connections between projects. The matching score between the new parameters and the target project is calculated by aggregating the neighborhood features, and the optimal parameter combination is dynamically recommended.

[0025] Preferably, based on the multi-dimensional extensible primitive library, the method for implementing topological dynamic linking includes:

[0026] Decompose the conductive graph elements of river and lake nodes into connector units with direction vectors, and construct a mathematical representation of the topological relationship of the unstructured water network through a set of directed line segments. ,in , are the coordinates of the connector start and end points, where , are the horizontal and vertical coordinates of the connector starting point, , are the horizontal and vertical coordinates of the connector end point, is the cross-sectional area function that changes with water level, where , are the functions of the cross-sectional area of ​​the starting section and the end section as the water level changes, is the dynamic conductivity matrix, which is composed of the water level gradient Drive, where and The conduction coefficient from the starting point to the end point and from the end point to the starting point respectively;

[0027] Each connector unit realizes asymmetric conduction characteristics through the direction vector weight function, and the dynamic correction formula of the conduction coefficient is:

[0028]

[0029]

[0030] in, is the base conductivity, is the connector direction vector, is an asymmetric regulatory factor, is the gradient response intensity coefficient, , are the starting water level and the ending water level of the connector.

[0031] Preferably, the method for constructing the composite flood evolution model includes: combining the improved Muskingum segmented calculation, the hydrodynamic model solved by the unstructured grid Godunov format, and the river-lake coupling model, comprehensively considering different water flow characteristics, and constructing the composite flood evolution model;

[0032] Among them, the improved Muskingum segmentation calculation model is used to divide the basin into several sub-river sections based on the terrain slope, roughness distribution, cross-section morphology, and hydrological station location attributes. The local Muskingum coefficient is introduced into each sub-river section. That is, the storage parameters and That is, the flow weight factor. The improved Muskingum method adopts a physically related initial parameter mapping mechanism. The initial value sets a nonlinear mapping function according to the sub-reach attributes. , ,in , , is a hyperparameter, The length of the river section, is the slope, For Manning's roughness, For the width of the river, For water depth; Sub-river sections, outflow ,in , , Depend on , Decide and satisfy , For time, is the time step; the inlet flow of the sub-river section and export flow The monitoring data is taken as input and the objective function is constructed , using recursive least squares RLS, real-time update and , and introduced the particle filter algorithm PF, which integrates historical flood event data with a Bayesian framework and periodically modifies hyperparameters , , During the calculation process, the basin is divided into multiple sub-sections, and the improved Muskingum model is applied to flood calculations according to the characteristics of the sub-sections. Then the results of each sub-section are integrated to obtain the flood evolution of the entire basin.

[0033] Preferably, the method for constructing an intelligent scheduling model includes:

[0034] Engineering connection relationship matrix analysis and hybrid structure eigenvalue decomposition based on graph theory;

[0035] Based on the results of analysis and decomposition, multi-mode scheduling of single projects and collaborative scheduling of project groups are carried out to complete the construction of an intelligent scheduling model.

[0036] Preferably, the method for performing multi-mode scheduling of a single project and coordinated scheduling of a group of projects includes: using an improved starfish optimization algorithm, the steps are:

[0037] Initialize population size The improved starfish optimization algorithm adopts a constraint satisfaction initialization method based on chaotic mapping when initializing population generation. That is, under the dynamic feasible domain that satisfies the reservoir outflow, the population is initialized by taking the chaotic mapping Cubic mapping strategy. The formula for taking the Cubic mapping to initialize the population is:

[0038] ;

[0039] in, is the population position matrix, , is the length of the scheduling period, is the upper boundary of the population position, which represents the upper boundary of the water level process in this problem. is the lower boundary of the population position, indicating the lower boundary of the water level process, , the final generated initialization population matrix is:

[0040] ;

[0041] Calculate the initial fitness value of each individual, that is, the objective function value. The minimum objective function value means the maximum peak reduction rate. The corresponding individual position, that is, the reservoir discharge process, is recorded as , Indicates Individual, The location of the dimension, Represents the current iteration number;

[0042] According to the position update formula of the improved starfish optimization algorithm, the position update of the population individuals is updated, that is, the process of updating the outflow of the reservoir. The position update of the improved starfish optimization algorithm includes the exploration phase and the development phase:

[0043] Exploration phase:

[0044] ;

[0045] ;

[0046] ;

[0047] in, is the updated position, is the position before the update, is the best position at the moment. yes are 5 randomly selected dimensions, is a random number between 0 and 1. is the current iteration number, is the maximum number of iterations, is a randomly generated angle parameter, is a dynamically adjusted angle parameter, ;

[0048] Development phase: Two position update strategies were designed based on the predation and regeneration behaviors of starfish:

[0049] ;

[0050] ;

[0051] ;

[0052] in, is the updated position, is the position before the update. and is a random number between 0 and 1. is the population size, is the distance between the best position and a randomly selected position in 5 dimensions, and is from Randomly selected from is the current optimal position, are the positions of 5 randomly selected starfish;

[0053] After the development stage, the tail mutation strategy is introduced, that is, a forced disturbance is added to the position of the population to avoid falling into the local optimal solution. As the number of iterations increases, the intensity of the disturbance gradually decreases to accelerate the convergence of the algorithm in the later stage. In the improved starfish optimization algorithm, the Levy flight strategy is selected to perturb the population position. is the Levy distribution;

[0054] ;

[0055] The objective function is calculated for each updated position, i.e., the reservoir discharge process. If the objective function value is better than the objective function value before the position update, the position is updated, otherwise the original position is retained. The cycle is iterated continuously until the loop termination condition is met or the iteration is completed, and the position of the optimal individual is the optimal discharge flow process.

[0056] The present invention also discloses a visual dynamic modeling system for real-time scheduling of a complex water engineering system in a river basin, the system is used to implement any one of the methods described, the system comprises: a first building module, a fusion configuration module, a second building module and a third building module;

[0057] The first construction module is used to construct a real-time scheduling topology visualization diagram of the watershed water engineering system;

[0058] The fusion configuration module is used to fuse and configure the characteristic parameters of the water project nodes and the characteristic parameters of the river section nodes based on the real-time scheduling topology visualization diagram;

[0059] The second construction module is used to construct a composite flood evolution model based on the node characteristic parameters of the fusion configuration;

[0060] The third building module is used to build an intelligent scheduling model based on the composite flood evolution model.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. Visual dynamic modeling: By constructing draggable basic graphic tools, a real-time scheduling topology diagram is constructed in a visual way, allowing users to intuitively understand and operate the scheduling relationship of complex water engineering systems, greatly improving the efficiency and accuracy of modeling, which is rare in the existing technology.

[0063] 2. Flexible configuration of multiple models: It covers a variety of flood evolution models and water project scheduling models, and supports a variety of solution algorithms. It can flexibly select appropriate models and algorithms according to the characteristics of different river basins and water projects to meet complex and changeable actual needs, reflecting stronger versatility and adaptability.

[0064] 3. Automatic verification and testing mechanism: Set up automatic verification of model and application configuration data integrity and scheduling plan calculation verification functions to ensure the accuracy of modeling data and the reliability of the model, effectively avoid model deviations caused by missing or incorrect data, and improve the practicality and application value of the model.

[0065] 4. Digital coding of hydraulic connections: By constructing a hydraulic relationship matrix (including adjacency matrix construction based on directed graphs and fuzzy membership assignment of weight matrices), the hydraulic relationship between nodes in the water engineering system can be described more accurately, taking into account factors such as flow transmission coefficient and hydraulic loss, providing the necessary conditions for accurate scheduling of computing tasks. The three-layer nested grid division of spatiotemporal discretization, the topological sorting algorithm of computing priority, and the event-driven dynamic step control in the scheduling of computing tasks can reasonably arrange computing tasks in space and time. It can not only accurately grasp the basin situation at the macro and micro levels, but also reasonably determine the calculation order and step size according to the topological structure and key events, which improves the computing efficiency and avoids the waste of computing resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0067] Figure 1 A schematic flow chart of a method for visualizing dynamic modeling of real-time scheduling of a complex water engineering system in a watershed according to an embodiment of the present invention;

[0068] Figure 2 The figure is a flow chart of the improved starfish optimization algorithm according to the embodiment of the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0071] The present invention "real-time scheduling visualization dynamic modeling method and system for complex water engineering system in a river basin" mainly solves the following technical problems:

[0072] 1. Improving the flexibility and real-time performance of scheduling topology construction

[0073] In view of the problems that the existing technology relies on fixed templates in scheduling topology construction, lacks real-time flood forecasting topology utilization and visualization construction tools, the present invention aims to create a method that can obtain flood forecasting topology in real time, use draggable basic graphic tools, classify graphics according to different water project types, and visually construct a real-time scheduling topology relationship diagram for complex water project systems, thereby improving the flexibility and real-time performance of scheduling topology construction, and making the constructed topology relationship diagram more accurately reflect the actual water project system structure.

[0074] 2. Enhance the adaptability and comprehensiveness of scheduling model configuration

[0075] Since the scheduling models in the prior art are not flexible enough when dealing with different types of engineering structures and the model solving algorithm is single, the present invention is committed to developing a scheduling model that can enable a single project to flexibly adapt to a variety of rainfall conditions and scheduling scenarios, support conventional scheduling and joint optimization scheduling of different structural engineering groups (parallel, series, mixed, etc.), and enable the model solving algorithm to comprehensively utilize the advantages of multiple algorithms (segmented trial algorithm, mathematical programming algorithm, heuristic intelligent algorithm, etc.), enhance the adaptability and comprehensiveness of the scheduling model configuration, and improve the effectiveness of the model in actual complex situations.

[0076] 3. Ensure the accuracy and efficiency of data verification and inspection

[0077] Taking into account the problems of the existing technologies in data verification and scheduling scheme calculation verification, such as the inability to comprehensively and accurately verify data integrity, the lack of rapid assembly, release and deployment capabilities, and insufficient verification depth and accuracy, the purpose of the present invention is to establish a complete model and application configuration data integrity automatic verification mechanism, which can timely and accurately verify the configuration information integrity of all water projects and section nodes, while improving the scheduling scheme calculation verification capabilities, achieving rapid assembly, one-click release, deployment and application, and using historical or real-time flood data to deeply and accurately verify the constructed model, ensuring the reliability and accuracy of the model in practical applications.

[0078] IV. Overall improvement of the real-time dispatching effect of complex water engineering systems in the basin

[0079] Overall, the present invention aims to solve the defects of the prior art in many aspects of real-time scheduling of complex water engineering systems in river basins. By solving the above-mentioned technical problems, the efficiency and accuracy of real-time scheduling of the entire river basin water engineering system can be improved, and the goals of rational allocation of water resources, flood prevention and disaster reduction, and ecological protection can be better achieved, thus meeting the needs of modern society for the efficient operation of river basin water engineering systems.

[0080] Embodiment 1

[0081] like Figure 1 As shown, the present invention provides a method and system for visual dynamic modeling of real-time scheduling of complex water engineering systems in a basin, including the following steps: S1. Real-time scheduling topology visualization construction; S2. Multi-source heterogeneous data fusion configuration; S3. Composite flood evolution model construction; S4. Intelligent scheduling model architecture; S5. Topological digital coding system; S6. Intelligent verification and validation system.

[0082] In this embodiment, S1. Real-time scheduling topology visualization construction:

[0083] S11. Constructing flood forecasting topological base layer based on river basin GIS data:

[0084] The basin geographic information system (GIS) data contains rich spatial information such as terrain and water system. First, collect high-precision GIS data in the basin, including terrain elevation, river network distribution, basin boundaries, etc. Through geospatial analysis technology, extract terrain features related to flood formation and propagation, such as valleys, depressions and other water-prone areas, as well as information such as the slope and width of the river. Based on this information, construct a network of possible flood flow paths as the basic framework of flood forecasting topology. For example, using digital elevation model (DEM) data, the natural flow direction of water from high-altitude areas to low-altitude areas is determined through the water flow direction algorithm to form a preliminary topological network, which provides a basic geospatial reference for the subsequent flood forecasting and scheduling topology construction.

[0085] Specifically, the data collection methods for terrain elevation, river network distribution, and watershed boundaries are as follows:

[0086] Terrain elevation data can be obtained in a variety of ways. Aerial photogrammetry is one of the common methods. Aircraft carry aerial cameras to photograph the watershed. After obtaining the aerial images, photogrammetry software is used to process them, such as stereo image pair matching technology, to obtain terrain elevation information. Satellite mapping is also feasible, such as using sensors on satellites such as TanDEM X to obtain surface reflection signals, and generating digital elevation models (DEMs) through radar interferometry (InSAR) or optical image stereo mapping technology. Ground measurement is to set control points in the watershed and use instruments such as total stations and levels for field measurements. The total station measures the three-dimensional coordinates of the control points, and the level measures the elevation difference. These field measurement data are integrated with data obtained by other methods to improve accuracy.

[0087] There are various ways to obtain river network distribution data. During the field survey, professionals are sent to investigate along the river, and GPS positioning equipment is used to record information such as the river direction, intersection, and river width, while observing characteristics such as water flow conditions and riverbank types. In terms of remote sensing image interpretation, with the help of high-resolution satellite remote sensing images (such as Landsat and high-resolution series satellite images) or aerial remote sensing images, the location of the river is identified by analyzing its spectral characteristics. For example, according to the reflectivity characteristics of the water body in different bands, a suitable water body index (such as the normalized difference water index NDWI) is used for extraction, and the river width can also be converted by the spatial resolution of the image. In addition, the existing river network distribution data can also be queried from the water conservancy department database. These data are often accumulated in various ways before, and have high accuracy and completeness.

[0088] For basin boundary data, watershed division based on terrain is an important method. Using terrain elevation data, through the hydrological analysis tools in GIS software, the flow direction and runoff accumulation are calculated based on DEM data, and the watershed location is determined to obtain the basin boundary. Administrative division determination is also a way. Refer to the local administrative division map and use the administrative boundary as part of the basin boundary. This method is more convenient when the basin spans multiple administrative regions. In addition, field surveys and visits are indispensable. For areas with unclear or disputed boundaries, professional personnel are organized to conduct field surveys and visit local residents to understand the geography and water flow ownership, assist in determining the final basin boundary and convert it into GIS vector data.

[0089] Specifically, the steps of extracting terrain features related to flood formation and propagation through geospatial analysis technology include:

[0090] Calculation of water flow direction: For a grid cell, it compares the elevation values ​​of the cell with its eight surrounding adjacent cells (up, down, left, right, upper left, upper right, lower left, and lower right). Starting from the central grid cell, the water will flow to the adjacent cell with the largest elevation drop. Assume that the elevation of the central grid cell is , and the elevations of its adjacent cells are ,calculate ,choose The direction of the adjacent unit corresponding to the maximum value in is taken as the flow direction. For example, if maximum, then the water flows from the central grid cell to its left adjacent cell.

[0091] Calculation of runoff accumulation: After determining the direction of water flow, calculate the runoff accumulation of each grid cell. The runoff accumulation indicates how many grid cells upstream of the grid cell have water flowing into this grid cell. Starting from the uppermost part of the watershed, initialize the runoff accumulation of each grid cell to 1. Then, according to the direction of water flow, add the runoff accumulation of the downstream grid cell to the runoff accumulation of the upstream grid cell. For example, if the water flow of grid cell (A) flows to grid cell (B), the runoff accumulation of grid cell (B) is equal to the original runoff accumulation of grid cell (B) plus the runoff accumulation of grid cell (A).

[0092] Specifically, the steps to construct a possible flood flow path network are as follows:

[0093] Construct flood flow paths based on terrain features: First, determine the waterlogging areas such as valleys and depressions based on terrain elevation data. These areas are where floods gather. At the same time, determine the river channels based on river network distribution data. Then, starting from valleys and depressions, connect valleys, depressions and rivers according to the natural flow direction of water (from high altitude to low altitude) to construct a flood flow path network.

[0094] Optimize flow paths using runoff accumulation: Areas with larger runoff accumulation indicate more water flows converge and are more likely to be the main flow paths of floods. For areas where runoff accumulation exceeds a certain threshold (T) (determined based on the actual situation of the basin, for example, T = 1000 grid cells), these areas are preferentially connected when constructing the flood flow path network.

[0095] Mathematical formula construction: Assume that (P) is a path in the flood flow path network, (H) is the elevation value of each point on the path, (L) is the length of the path, and (F) is the possibility function of flood flow. It can be defined as ( ), where (n) is the number of points on the path (P). The larger the value of (F), the greater the possibility that the flood will flow along the path. When constructing a flood flow path network, paths with larger (F) values ​​are preferred for connection.

[0096] S12. Construction of multi-dimensional extensible graphic element library:

[0097] This step constructs a multi-dimensional and extensible graphic element library for basin water project scheduling. Based on the engineering function attributes, the graphics elements are divided into three categories: regulation type, protection type and conduction type. Through parametric modeling and dynamic association rule definition, a structured database that supports complex water network topology reconstruction and intelligent computing is established. Specifically, based on the physical characteristics and regulation behavior of engineering entities, a composite parameter system including geometric parameters, hydraulic parameters and control logic parameters is designed for each type of graphic element. The geometric parameters are instantiated and positioned through spatial coordinate transformation. The hydraulic parameters are based on the Manning formula, Saint-Venant equations and other hydraulic principles to build a calculation model. The control logic parameters define the equipment operation rules and timing constraints. At the same time, unlike the traditional method that only supports a single static rule verification, the dynamic injection of new engineering features is realized through a multi-layer open extension interface, and the parameter dimensions can be customized according to the characteristics of the basin. Specifically, it includes the following technical features:

[0098] S121. Controlling elements (reservoirs, dams):

[0099] For the reservoir model in the regulation-type graph, the relationship between storage capacity and water level is represented by a seasonal function, specifically the storage capacity curve during the flood season (May-October) and dry season (November to April) , where the coefficient The least square method is used to fit historical hydrological data, and the fitting error is controlled within 3%. The flood discharge calculation model is defined as ,in is the flow coefficient ( ), is the width of the spillway, is the gate opening, It is the gate sill elevation.

[0100] In the dam element, the dynamic model of upstream and downstream water level difference is adopted , where is the riverbed roughness correction coefficient ( ), is the discharge duration (minutes), and the model quantifies the attenuation effect of long-term discharge on water levels.

[0101] S122. Protective elements (embankments, flood storage areas):

[0102] The calculation of flood storage area capacity of protective graphics introduces piecewise function: when the water level When, volume ;when When it expands to , where the terrain correction factor The value range is 0.2 to 0.5 and is calibrated by terrain data.

[0103] S123. Conductive elements (river sections, river-connected lake nodes):

[0104] The calculation of the roughness of the river section of the conductive element is based on the Manning formula , combined with the dynamic assignment of riverbed type: concrete lining ( ), gravel riverbed ( ), vegetation-covered riverbed ( ), where R is the hydraulic radius, S is the energy slope, and v is the average flow velocity of the section.

[0105] S124. Dynamic interface and basin adaptive parameter system:

[0106] Through the hierarchical extension interface architecture and the intelligent parameter adaptation technology driven by watershed characteristics, the dynamic injection of new engineering features and the flexible expansion of parameter dimensions are realized.

[0107] Layered interface protocol design:

[0108] Core interface layer: Forced inheritance of hydraulic conductivity properties and spatial topological constraints of basic graphics elements to ensure mathematical compatibility of new parameter dimensions with existing models. For example, when adding ecological flow parameters, the mass conservation term in the Saint-Venant equation is automatically associated to avoid manual modification of the control equation. Extended interface layer: Based on the domain-specific language (DSL), the physical dimensions, value ranges and data validation rules of the new parameters are declared, supporting non-programmers to define parameters through configuration files. Basin adaptation layer: Built-in terrain slope , soil permeability coefficient , river network density Parameterized templates of equal basin characteristics dynamically generate parameter legitimacy functions.

[0109] Dynamic parameter binding and conflict detection mechanism:

[0110] Symbolic binding of hydraulic models: using symbolic computing engines (such as SymPy) to automatically parse new parameters Hydraulic control equation The partial derivative relationship of OK The Jacobian matrix of the columns, Ensure that parameters are automatically integrated into the calculation kernel after injection, without the need to reconstruct the model code. Build a parameter dependency network based on a directed acyclic graph, detect contradictory combinations (such as the logical conflict between gate opening and ecological flow) through topological sorting, trigger real-time alarms and lock the calculation process. Implement conflict self-checking through mathematical symbol binding and graph theory algorithms to avoid manual debugging.

[0111] Intelligent parameter recommendation system:

[0112] For hydraulic structures such as reservoirs, dams, dikes, flood storage areas, river sections, and river-connected lake nodes, an intelligent parameter recommendation system based on engineering feature fingerprints and cross-engineering topological association learning is constructed, including:

[0113] Engineering feature fingerprint library: extract core parameters for each type of hydraulic structure (such as the reservoir capacity curve coefficient, the discharge capacity curve of the dam, and the roughness n of the river section), and generate a unique feature fingerprint code by weighting through the random forest algorithm ,in, is the parameter importance weight, It is the core physical parameter value extracted from the project, which is used to quantify the structural or functional characteristics of different projects. Based on the graph convolutional network (GCN), a topological map of hydraulic structures is constructed. The nodes are the project feature fingerprints, and the edges are the river connections between projects. The matching score between the newly added parameters and the target project is calculated through neighborhood feature aggregation, and the optimal parameter combination is dynamically recommended (for example, when recommending a flood discharge coefficient for a new reservoir, the historical parameters of similar storage capacity and dam elevation in the topological map are automatically matched).

[0114] S13. Implementing topology dynamic linking:

[0115] This step realizes the dynamic topology construction and self-consistency verification of the basin water engineering system through spatial topology modeling and hydraulic conductivity relationship calculation, and establishes a topological network with physical rationality and computational robustness based on the parameterized attributes of the primitives and the constraint rule library. The specific implementation methods include:

[0116] S131. Topology node generation and verification:

[0117] A graphic element library is pre-built, which contains a variety of graphics elements corresponding to the watershed water engineering system, such as reservoir graphics elements, river section graphics elements, etc. Each graphic element is designed to have a clear mapping relationship with the actual watershed water engineering, and this relationship is determined based on the type, function and other characteristics of the water engineering. For example, the reservoir graphic element corresponds to the actual reservoir engineering entity, and is intended to accurately represent the position and role of the reservoir in the watershed water engineering system in the topology construction. For each graphic element, a set of basic attribute information is defined, which will be retained when the graphic element is converted to a topological node. When the user drags the graphic element to the geographic coordinate system through the interactive interface, the system converts the graphic element into a topological node based on the parameterized attributes of the graphic element (such as the reservoir capacity-water level function , dam discharge control logic) automatically generates topological nodes, and point node data contains spatial coordinates (longitude ,latitude and elevation ), hydraulic parameters (design flow Roughness ) and dynamic control variables (such as gate opening ). Double verification is performed after the node is generated:

[0118] Spatial logic verification: Use pre-acquired topographic data, such as contour distribution data, to analyze the topographic characteristics of the catchment area in detail, including shape, terrain height distribution, etc. Make judgments based on the principle of natural convergence of water flow. Under natural conditions, water flows tend to converge in low-lying areas. By comparing the terrain corresponding to the coordinates of the reservoir location with the overall terrain of the catchment area, if the reservoir location is low-lying relative to the surrounding terrain and conforms to the overall water flow convergence direction of the catchment area, it is determined that the reservoir is located in a reasonable catchment area; otherwise, it is determined to be unreasonable and triggers a coordinate correction suggestion.

[0119] Parameter compatibility check: Build a parameter constraint library based on engineering specifications, such as the height of the embankment must meet m, gate opening With leakage flow The association must meet ,in It is the flow coefficient, which comprehensively considers the various resistance factors of water flow and the influence of the shape and roughness of the water-passing section on the flow. It is dimensionless. Represents the cross-sectional area of ​​the water flow, that is, the cross-sectional area through which the water flows, in m 2 If a parameter is detected to be out of limit, the normal value range and change trend of the parameter in the historical data are analyzed, and the reasonable correction direction and amplitude are determined in combination with the experience in engineering practice.

[0120] S132. Hydraulic conductivity relationship modeling:

[0121] Hydraulic conductivity relationship modeling: Different from the existing technology that uses homogenized grid units or fixed connection rules, this method decomposes conductive primitives such as river channels and lake nodes into connector units with direction vectors, and constructs a mathematical representation of the topological relationship of unstructured water networks through a set of directed line segments. ,in , are the coordinates of the connector start and end points, where , are the horizontal and vertical coordinates of the connector starting point, , are the horizontal and vertical coordinates of the connector end point, is the cross-sectional area function that changes with water level, where , are the functions of the cross-sectional area of ​​the starting section and the end section as the water level changes, is the dynamic conductivity matrix, which is composed of the water level gradient Drive, where and The conduction coefficient from the starting point to the end point and from the end point to the starting point respectively.

[0122] Each connector unit not only integrates geometric properties such as spatial coordinates and cross-sectional shapes, but also has a built-in dynamic conduction parameter matrix, which can correct the flow direction and conduction coefficient in real time according to the water level gradient, accurately depict the asymmetric conduction characteristics of complex bifurcated rivers and the two-way flow switching caused by water level fluctuations in natural rivers. The symmetry of the direction vector weight function is used to realize the asymmetric conduction characteristics. The dynamic correction formula of the conduction coefficient is as follows:

[0123] ;

[0124] ;

[0125] in, is the base conductivity, is the connector direction vector, is an asymmetric regulatory factor, is the gradient response intensity coefficient, , are the starting water level and the ending water level of the connector.

[0126] This technology abandons the static simplification assumption of the conduction path in traditional models, and while maintaining computational efficiency, it significantly improves the simulation accuracy of hydraulic coupling between mountain river channels and plain river networks, providing a new structural solution for high-fidelity simulation of watershed hydrodynamic systems. The hydraulic conduction relationship between nodes is defined through vectorized connectors, and conduction-type primitives such as river channels, river-connected lake nodes, etc. are abstracted into directed line segments. , whose properties include:

[0127] Hydraulic conductivity: Based on Manning's formula Dynamic calculation, where the hydraulic radius R is determined by the cross-section geometric parameters (bottom width B, slope coefficient m) and the real-time water depth OK, Nengpo By the elevation difference of adjacent nodes With connection length Calculated as ;

[0128] Dynamic flow control: Force the connection direction to be consistent with the water level gradient. When there is no pump station drive, the connection direction is automatically reversed and the head loss recalculation is triggered to ensure that the water flow direction complies with The laws of physics;

[0129] Head Loss Fusion Model: Total Head Loss Loss along the way (Darcy friction factor Iterative solution of the Colebrook-White equation) and local losses (Local loss coefficient , based on the connector type lookup table assignment) joint calculation to achieve high-precision quantification of conduction resistance.

[0130] S133. Real-time detection and correction of topology conflicts:

[0131] Hydraulic conflict detection: Real-time calculation of node water level and flow, if the downstream water level is detected , higher than upstream If there is no counter-flow equipment, it is marked as a water level deficit conflict;

[0132] Spatial conflict detection: Determine the method of spatial position comparison. For each element, obtain its position coordinate range. For example, for a reservoir element, obtain the coordinate values ​​of its four corner points or center coordinates and coverage range to determine its position range in geographic space.

[0133] Hierarchical correction strategy: automatic correction with limited thresholds for hydraulic conflicts, such as adjusting the connection direction and limiting the water level correction amplitude in water level deficit conflicts m; for spatial conflicts, the optimization model is used Solving for the minimum displacement vector , recommend users to press Adjust the position of the elements while keeping the topological connection relationship unchanged.

[0134] In this embodiment, S2. Multi-source heterogeneous data fusion configuration:

[0135] S21. Configuration of characteristic parameters of water project nodes:

[0136] S211. Construct three-dimensional parameter matrix (geometric characteristics, hydraulic characteristics, dispatching rules):

[0137] For each engineering node (such as reservoirs, dams, etc.), a three-dimensional parameter matrix is ​​constructed to fully describe its characteristics. In the geometric feature dimension, the spatial information of the project such as the shape and size is recorded, such as the dam length, dam height, and storage capacity shape (such as valley type, plain type, etc.) of the reservoir; in the hydraulic characteristic dimension, it includes parameters such as water level-flow relationship curve, seepage characteristics (for dam body), and water flow resistance; in the scheduling rule dimension, the scheduling strategies under different working conditions (such as flood season and dry season) are defined, such as the flood discharge rules of the reservoir (determining the flood discharge amount based on factors such as water level and inflow flow), the opening and closing conditions of the dam, etc. Through this three-dimensional parameter matrix, the multi-faceted characteristics of the engineering nodes are integrated to provide comprehensive data support for subsequent scheduling calculations.

[0138] S212. Dynamic interpolation expression of nonlinear storage capacity curve:

[0139] Considering that the actual reservoir capacity curve is often nonlinear, the traditional linear approximation method may lead to large errors. The dynamic interpolation method is used to accurately represent the reservoir capacity curve. The measured data of reservoir capacity under different water levels are collected, and then the continuous function relationship between reservoir capacity and water level is constructed using interpolation algorithms (such as cubic spline interpolation).

[0140] Establish cubic spline interpolation function, cubic spline interpolation function is a piecewise cubic function, in each subinterval superior, in It's the water level. is the coefficient to be determined. Determine the boundary conditions and assume that the second-order derivative is zero at the endpoints of the interval, that is, and , construct a linear system of equations, at the internal nodes At and its first-order derivative , second-order derivative Continuous. According to the continuity of the function value, we can get: According to the first-order derivative continuity, we can get: , according to the second-order derivative continuity:

[0141] Combined with the boundary conditions, we can construct a Total intervals Unknown number The linear system of equations.

[0142] Arrange the above conditions into a matrix form in is the coefficient matrix, Contains unknowns The vector of is a constant term vector. Use linear algebra solution methods (such as Gaussian elimination method, LU decomposition method, etc.) to solve the linear equations and get The value of .

[0143] In the reservoir operation calculation process, when a water level is given When The sub-interval , and then substitute the corresponding cubic spline function The corresponding storage capacity value is calculated .

[0144] In this way, during the scheduling calculation process, no matter what the water level is, the corresponding reservoir capacity value can be accurately obtained through the interpolation function, which improves the accuracy of the reservoir capacity calculation and provides a more reliable data basis for the reservoir scheduling decision.

[0145] S22. River section node feature configuration:

[0146] S221. B-spline parametric modeling of cross-sectional morphology:

[0147] In order to accurately describe the shape of the river section, B-spline curves are used for parametric modeling. First, the three-dimensional coordinates of several key control points on the cross-sectional shape are obtained by measuring or obtaining them from GIS data. These control points should include the turning point of the river bank, the deepest point, the location of the shoal, and other points that can reflect the characteristics of the cross-sectional shape. The B-spline function is used to fit these control points to obtain the mathematical expression of the cross-sectional shape.

[0148] For a given Control Points , B-spline curve The expression is:

[0149] ;

[0150] in , yes B-spline basis functions of order .

[0151] when hour

[0152] ;

[0153] when hour

[0154] ;

[0155] in is an element in the node vector, the node vector , the way the node vector is determined will affect the shape of the B-spline curve. Generally speaking, or It is common in practical applications, which can ensure the smoothness of the curve and have good flexibility. and the determined node vector Substitute the B-spline function expression and obtain the B-spline curve through calculation That is, the mathematical expression of the cross-sectional shape of the river.

[0156] This method can accurately represent various complex cross-sectional shapes, such as irregular natural river sections. In scheduling calculations, the B-spline parameterized model can accurately calculate the cross-sectional hydraulic parameters such as the water flow area and wetted perimeter, providing an accurate geometric basis for water flow calculations.

[0157] For a given water level , first determine the B-spline curve With water level Solve the equation Get the parameter value corresponding to the intersection point and .

[0158] Water flow area ;

[0159] Wet periphery ;

[0160] in and They are of and Coordinate components, yes right S222. Piecewise polynomial representation of the water level-discharge relationship:

[0161] Since the water level-flow relationship of the river section may show different characteristics in different water level sections, a piecewise polynomial is used to characterize this relationship. By analyzing historical water level and flow data, water level sections with different characteristics (such as low water level slow flow section, medium water level transition section, high water level rapid flow section) are determined. The low water level slow flow section usually shows that when the water level is low, the flow changes relatively slowly with the water level. The medium water level transition section is the area connecting the low water level slow flow section and the high water level rapid flow section. Its characteristic is that the water level-flow relationship begins to change significantly, and the flow rate gradually increases with the increase of water level. At high water levels, the flow changes very rapidly with the water level, showing a rapid flow state. For each water level section, a polynomial function is fitted to describe the water level-flow relationship. A quadratic polynomial may be used in the low water level slow flow section.

[0162] ;

[0163] in It's traffic. It's the water level. , , is the coefficient to be determined.

[0164] For low water level slow flow section Group water level-flow data points , , according to the principle of least squares, construct the error function.

[0165] ;

[0166] In order to minimize the error function, , , Find the partial derivative and set it equal to zero.

[0167] ;

[0168] ;

[0169] ;

[0170] Solve the linear equations to obtain the coefficients of the quadratic polynomial for the low water level slow flow section , , .

[0171] In the mid-water transition section, the high-water rapid section may require a higher-order polynomial such as The method for determining the coefficients of the polynomial is the same as the method for determining the coefficients of the quadratic polynomial described above.

[0172] This piecewise polynomial characterization method can more accurately reflect the actual changes in the water level-discharge relationship and improve the accuracy of flood evolution calculations.

[0173] In this embodiment, S3. Construction of composite flood evolution model:

[0174] S31. Multi-method model container architecture:

[0175] S311. Hydrological model (modified Muskingum segmented routing):

[0176] Muskingum segmentation model is a classic flood routing model, but it has certain limitations in practical applications. Existing segmentation methods are mostly based on fixed characteristic river length or empirical division. This paper adopts an improved Muskingum segmentation model to divide the basin into several sub-segments based on the attributes of terrain slope, roughness distribution, cross-section morphology, and hydrological station location. Each sub-segment introduces a local Muskingum coefficient. (storage parameter) and (discharge weighting factor), which is different from the traditional Muskingum model parameters that lack hydraulic interpretation when the initial values ​​of the parameters are given , Relying on empirical trial calculation or global optimization, the improved Muskingum method adopts a physically related initial parameter mapping mechanism, and the initial value sets a nonlinear mapping function according to the sub-reach attributes. , ,in , , is a hyperparameter, The length of the river section, is the slope, For Manning's roughness, For the width of the river, For the water depth. Sub-river sections, outflow ,in , , Depend on , Decide and satisfy , For time, The time step is the time step. The parameters of the model are optimized according to the specific characteristics of the basin. The Muskingum coefficient is dynamically adjusted considering the influence of river channel characteristics (such as slope, roughness, etc.) of different river sections on flood propagation. and export flow The monitoring data is taken as input and the objective function is constructed , using recursive least squares (RLS) method, real-time update and , and introduced the particle filter (PF) algorithm to fuse historical flood event data in a Bayesian framework and periodically correct hyperparameters , , , solve the local overfitting problem of RLS and improve the simulation accuracy of the model for the flood evolution process. In the calculation process, the basin is divided into multiple sub-river sections, and the improved Muskingum model is applied to flood calculation according to the characteristics of the sub-river sections. Then the results of each sub-river section are integrated to obtain the flood evolution of the entire basin.

[0177] S312. Hydrodynamic model (unstructured grid Godunov solution):

[0178] For complex basin terrain and water flow conditions, the hydrodynamic model is solved using the unstructured grid Godunov format. Unstructured grids can better adapt to irregular basin boundaries and complex terrain and improve the adaptability of the model. The Godunov format has the characteristics of high precision and stability, and can accurately solve the motion equations of water flow. In the model construction process, the Delaunay triangulation algorithm is first used to discretize the basin into unstructured grids, and each grid cell is assigned corresponding physical parameters (such as roughness, water depth, etc.) according to the terrain and water flow characteristics. The numerical algorithm based on the Godunov format solves the basic equations such as the mass conservation equation and momentum conservation equation of water flow.

[0179] The mass conservation equation: ;

[0180] Momentum conservation equation: ;

[0181] in For water depth, is the flow velocity vector, For time, is the acceleration due to gravity, is the water level, is Xie Cai coefficient, For other external forces.

[0182] At each time step For each grid cell boundary, the Riemann problem of the grid cell boundary is constructed based on the water depth and flow velocity of the left and right adjacent grid cells. The left state on a discontinuity surface is , the state on the right is , that is, the special initial conditions of the Riemann problem, the flux on the boundary is solved, and then the physical quantities inside the grid unit are updated according to the flux to simulate the movement of water flow in the basin, including the propagation of floods and changes in water levels.

[0183] S313. Tongjiang Lake Coupling Model (Characteristic Line - Implicit Alternating Solution):

[0184] There is a complex water flow exchange relationship between lakes and rivers connected to rivers. Lakes and rivers are regarded as an interrelated whole system. By considering multiple key factors, a coupled model of lakes connected to rivers is constructed to simulate the movement and change process of water flow in this system. The characteristic line-implicit alternating solution method is adopted to consider the two-way water flow exchange between lakes and rivers, the water flow movement driven by water level difference, and the regulation function of lakes themselves in the model.

[0185] Characteristic line method to solve river flow: divide the characteristic line grid along the river channel and solve the characteristic equation to obtain the flow and water level The flow and water level at the end section of the river are output as the boundary conditions of the lake entrance.

[0186] Implicit method for solving lake level-volume relationship: constructing a nonlinear system of equations based on the lake's continuity equation

[0187] ;

[0188] in is the lake area, , For inbound and outbound traffic, , is the precipitation and evaporation. The equations are discretized using implicit difference format and the lake level is updated by iterative solution.

[0189] The characteristic line method is used to deal with the propagation characteristics of water flow, and can accurately calculate the propagation time and flow change of water flow between lakes and rivers; the implicit method is used to solve the water level-volume relationship of lakes and rivers to improve the stability of calculation. Through this coupling model, the special evolution process of floods in the Tongjiang Lake area can be accurately simulated, providing more comprehensive support for the flood evolution calculation of the entire basin.

[0190] S32. Intelligent adaptation of model parameters:

[0191] S321. Model selection decision tree based on river characteristics:

[0192] A model selection decision tree is constructed based on different characteristics of the river channel (such as river channel length, slope, roughness, cross-sectional shape, etc.). First, various characteristic data of the river channel are collected, including river channel length, slope, roughness and cross-sectional shape, and the threshold or range of each feature is determined. Next, a feature is selected as the root node, such as the river channel length, and a judgment condition is set, such as "whether the river channel length is greater than a certain threshold". According to the judgment result, it is divided into two sub-nodes, representing "yes" and "no" respectively. In each sub-node, the next feature is selected for judgment to form a branch of the tree until the preset model selection standard or leaf node is reached. The leaf node corresponds to a specific flood evolution model (such as Muskingum model or hydrodynamic model). In this way, the decision tree can automatically guide the system to select the most suitable model according to different combinations of river channel characteristics, thereby improving the pertinence and computational efficiency of the model.

[0193] S322. Multi-objective optimization calibration of parameter sensitivity:

[0194] After the flood evolution model is determined, the parameters in the model are optimized. Multiple objective functions are considered, such as the accuracy of flood evolution simulation, computational efficiency, etc. The sensitivity of the parameters is determined by analyzing the impact of parameter changes on these objective functions. Then, optimization algorithms (such as genetic algorithms, particle swarm algorithms, etc.) are used to optimize the parameters.

[0195] Construct the objective function and use the root mean square error To measure accuracy, time Measures computational efficiency.

[0196] By initializing the position and direction of the particle swarm, the initial values ​​of the parameters to be optimized are randomly generated within the empirical range. , randomly initialize its position in the decision variable space ,in is the number of decision variables, i.e. the number of parameters to be optimized in the model to be optimized. The speed of randomly initialized particles .

[0197] Based on the individual fitness value in the initial population, i.e., the objective function value, the position of the individuals in the population is continuously updated iteratively through the particle swarm algorithm position update formula, i.e., the value of the parameter in the selected flood evolution model is updated to obtain the population after the individual position is updated until the accuracy requirement or the maximum number of iterations is met; specifically, the speed and position update formula of the particle swarm algorithm are as follows:

[0198] Speed ​​update formula: ;

[0199] Position update formula: ;

[0200] in is the inertia weight, which controls the degree to which the particle inherits the previous velocity, and is generally between 0.4-0.9; and are learning factors, which represent the particle's ability to learn from its own experience and group experience. about; and Is a random number between 0 and 1.

[0201] The parameter combination that makes the objective function the best during the iteration process is selected, that is, the model parameters with the highest flood evolution simulation accuracy and the fastest calculation efficiency. For example, in the improved Muskingum model, the Muskingum coefficient is optimized and calibrated so that it can accurately simulate the flood evolution process under different flood conditions and improve the calculation efficiency.

[0202] In this embodiment, S4. Intelligent scheduling model architecture:

[0203] S41. Engineering structure topology analysis:

[0204] S411. Engineering connection relationship matrix analysis based on graph theory:

[0205] The water projects in the basin are regarded as nodes in graph theory, and the connections between water projects are regarded as edges, and the project connection relationship matrix is ​​constructed. The elements in the matrix represent the connection relationship between nodes (such as 1 for connection and 0 for no connection) and the properties of the connection (such as flow transmission coefficient, hydraulic loss, etc.). By analyzing this matrix, the topological information of the project structure can be obtained, such as the degree of the node (the number of edges connected to the node), connectivity (whether there is a path from one node to another), etc. This topological information helps to gain a deep understanding of the characteristics of the project structure and provides a basis for the subsequent construction of the scheduling model.

[0206] S412. Eigenvalue Decomposition of Hybrid Structures:

[0207] For complex engineering structures such as parallel, series, and hybrid, the eigenvalue decomposition method is used to analyze their structural characteristics. By calculating the eigenvalues ​​and eigenvectors of the engineering connection relationship matrix, the key characteristic information of the structure is obtained. Constructing the engineering connection relationship matrix To describe the connection strength and direction between nodes, the element Representation Node For Node The interaction weights of (in is the diagonal matrix of eigenvalues, is the eigenvector matrix), which can quantitatively reveal the structural characteristics, the maximum eigenvalue Characterizes the overall stability of the system. When the system has the risk of instability, the eigenvector Amplitude of the middle element Reflection Node In the The participation degree in the first-order mode can be used to identify the key nodes. Based on these characteristic information, we can better understand the impact of complex engineering structures on scheduling decisions and provide a basis for formulating reasonable scheduling strategies.

[0208] S42. Multi-mode scheduling of single projects:

[0209] S421. Rule-based scheduling (water level / flow dual threshold control):

[0210] For single projects (such as single reservoirs or dams), task scheduling is performed according to pre-set rules and priorities, and dual thresholds for water level and flow are set. When the reservoir water level reaches the upper threshold, flood discharge is initiated to ensure that the reservoir water level is within a safe range; when the flow is lower than the lower threshold, the discharge of the dam is reduced to maintain a certain river ecological flow. During the scheduling process, water level and flow data are monitored in real time, and corresponding operations are performed according to the set thresholds to meet the safety and ecological requirements of project operation.

[0211] S422. Optimized scheduling:

[0212] In order to achieve the optimal scheduling of individual projects, the project scheduling problem is modeled and solved using the Starfish optimization algorithm. In reservoir scheduling, the optimization objectives are power generation benefits, flood control benefits, ecological benefits, etc. The Starfish optimization algorithm is used to solve the optimal flood discharge and water storage strategies at different times to maximize the comprehensive benefits of individual projects. For example, for flood control problems, the maximum peak reduction criterion is used as the objective function.

[0213] Objective function:

[0214] ;

[0215] in is the number of time periods in the scheduling period, h; for Reservoir outflow at the moment, m 3 / s.

[0216] Constraints:

[0217] Water balance constraints:

[0218] ;

[0219] in They are Reservoir storage, inflow and outflow at the beginning of the period; They are Reservoir storage, inflow, and outflow at the end of the period; For a long period of time.

[0220] Reservoir maximum water level constraint:

[0221] ;

[0222] in for Calculate the reservoir water level value at all times; for The maximum water level of the reservoir allowed at any time.

[0223] Discharge capacity constraints:

[0224] ;

[0225] in for Outbound flow at any given moment; for Time corresponds to water level The discharge capacity value includes the discharge bottom outlet, spillway and water flow capacity of the turbine.

[0226] Initial boundary condition constraints:

[0227] ;

[0228] in is the scheduling period; is the initial value of water level at the beginning of the dispatch period; It is the water level at the end of a given scheduling period. In the flood stage, this water level is reflected as the storage capacity reserved for subsequent rainfall. In the tail of the flood, it is reflected as the beneficial storage water level achieved by considering the utilization of rainwater resources.

[0229] Constraints on outbound traffic fluctuation:

[0230] ;

[0231] in It is the fluctuation range of outbound flow between adjacent time periods; It is the allowable value of the outbound flow rate variation between adjacent time periods.

[0232] When solving the problem using the improved starfish optimization algorithm, the penalty function method is adopted. It is necessary to select a suitable penalty coefficient and add a penalty term to the objective function to handle the above complex constraints. The forced repair rule uses appropriate rules to transform infeasible solutions into feasible solutions to ensure the effectiveness of the search. Taking the maximum water level constraint as an example, it is constructed as a penalty function and added to the objective function. The processing methods of other constraints are similar.

[0233] ;

[0234] ;

[0235] ;

[0236] in is the objective function with penalty function added; , are the penalty functions for the lowest water level and the highest water level respectively;

[0237] When using the improved starfish optimization algorithm to solve the reservoir flood control optimization scheduling problem, each optimization individual represents the outflow of the reservoir during the scheduling period. The search space of the population is the search range of the reservoir outflow, and the quality of the individual position is determined by the objective function. The higher the peak reduction rate of the reservoir, the better the individual position. The basic steps are as follows:

[0238] Initialize population size , which is different from the general intelligent algorithm that uses a pseudo-random number generation strategy to randomly generate initial solutions. The population generated by this method is not evenly distributed in the solution space, which can easily cause the algorithm to fall into a local optimum. The improved starfish optimization algorithm uses a constraint satisfaction initialization method based on chaotic mapping when initializing population generation. That is, under the dynamic feasible domain that satisfies the reservoir outflow, the chaotic mapping Cubic mapping strategy is used to initialize the population, providing a higher quality initial solution set for subsequent algorithm optimization iterations. The formula for initializing the population using Cubic mapping is as follows:

[0239] ;

[0240] in, is the population position matrix, , is the length of the scheduling period, is the upper boundary of the population position, which represents the upper boundary of the water level process in this problem. is the lower boundary of the population position, indicating the lower boundary of the water level process, , the final generated initialization population matrix is:

[0241] ;

[0242] Calculate the initial fitness value of each individual, that is, the objective function value. The minimum objective function value means the maximum peak-cutting rate. The corresponding individual position, that is, the reservoir discharge process, is recorded as , Indicates Individual, The location of the dimension, Represents the current iteration number.

[0243] According to the position update formula of the improved starfish optimization algorithm, the position update of the population individuals is carried out, that is, the out-of-reservoir process of the reservoir is updated. The position update of the improved starfish optimization algorithm includes an exploration phase and a development phase.

[0244] Exploration phase:

[0245] ;

[0246] ;

[0247] ;

[0248] in is the updated position, is the position before the update, is the best position at the moment. yes are 5 randomly selected dimensions, is a random number between 0 and 1. is the current iteration number, is the maximum number of iterations, is a randomly generated angle parameter, is a dynamically adjusted angle parameter, .

[0249] Development phase: Two position updating strategies were designed based on the predation and regeneration behaviors of sea stars.

[0250] ;

[0251] ;

[0252] ;

[0253] in, is the updated position, is the position before the update. and is a random number between 0 and 1. is the population size, is the distance between the best position and a randomly selected position in 5 dimensions, and is from Randomly selected from is the current optimal position, are the positions of 5 randomly selected starfish.

[0254] After the development stage, the tail mutation strategy is introduced, that is, a forced disturbance is added to the position of the population to avoid falling into the local optimal solution, and the intensity of the disturbance gradually decreases with the increase of the number of iterations, accelerating the convergence of the algorithm in the later stage. In the improved starfish optimization algorithm, the Levy flight strategy is selected to perturb the position of the population. is the Levy distribution.

[0255] ;

[0256] The objective function is calculated for each updated position, i.e., the reservoir discharge process. If the objective function value is better than the objective function value before the position update, the position is updated, otherwise the original position is retained. The cycle is repeated until the loop termination condition is met or the iteration is completed. At this time, the position of the optimal individual is the optimal discharge flow process. Figure 2 shown.

[0257] S43. Joint dispatch of water project groups:

[0258] S431. Feedforward-Feedback Coupling Control of Series Systems:

[0259] For the series water engineering system (such as multiple reservoirs connected in sequence), the feedforward-feedback coupling control strategy is adopted. Feedforward control adjusts the operating parameters of the downstream project in advance according to the operating status of the upstream project (such as flood discharge, water level, etc.). For example, when the flood discharge of the upstream reservoir increases, the downstream reservoir adjusts the opening of the water inlet gate in advance to cope with the upcoming flood. Feedback control adjusts the scheduling strategy of the upstream project according to the actual response of the downstream project (such as water level changes, flood discharge capacity, etc.) to ensure the stable operation of the entire series system. Through feedforward-feedback coupling control, the series system's ability to respond to floods and the efficiency of water resource allocation are improved.

[0260] S432. Nash equilibrium scheduling strategy for parallel systems:

[0261] In parallel water engineering systems (such as multiple dams operating in parallel), the Nash equilibrium scheduling strategy is adopted. While pursuing its own interests (such as minimizing its own flood control pressure and maximizing its power generation benefits, etc.), each dam considers the operation strategies of other dams and eventually reaches an equilibrium state. By establishing a game model, analyzing the strategy space and benefit function of each dam, and finding the Nash equilibrium point, that is, all dams are unwilling to change their strategies individually at this point. This scheduling strategy can achieve the optimal operation of the parallel system as a whole while ensuring the benefits of each dam.

[0262] In the game theory scheduling modeling of parallel water engineering systems, the realization of Nash equilibrium requires the following five steps:

[0263] Formal definition of game participants:

[0264] The system includes The set of participants is composed of parallel dams , each dam The strategy variable is the flood discharge , where the upper and lower bounds are determined by the physical constraints of the gate and safety regulations. The strategy space forms a Cartesian product .

[0265] Construct a multi-objective profit function:

[0266] The revenue function of each dam It is necessary to integrate multiple goals such as flood control, power generation, and ecology:

[0267] ;

[0268] in represents other dam strategy combinations, is the water level before the gate (according to the hydraulic equation Sure), is the risk sensitivity coefficient, For power generation efficiency, is the weight coefficient.

[0269] Modeling of constraint coupling mechanisms:

[0270] The system as a whole needs to meet the total flood discharge constraints of the basin:

[0271] ;

[0272] And the hydraulic coupling relationship between gates:

[0273] ;

[0274] in The coupling coefficient of the water level difference between the dams is determined by the river topography parameters.

[0275] Nash equilibrium solution algorithm:

[0276] Use backward induction to solve the equilibrium of non-cooperative games:

[0277] Construct a reaction function mapping, for each , solve the optimal response strategy:

[0278] ;

[0279] Subject to ,in is the global constraint function.

[0280] Iterate convergence judgment and update strategy combination , calculate the strategy change:

[0281] ;

[0282] when When the equilibrium point is reached .

[0283] In this embodiment, S5. Topological digital coding system:

[0284] S51. Generation of hydraulic relationship matrix:

[0285] S511. Construction of adjacency matrix based on directed graph:

[0286] The water conservancy system in the river basin is considered as a directed graph, where the water conservancy and cross-section nodes are nodes in the graph, and the water flow direction is a directed edge. An adjacency matrix is ​​constructed to represent the hydraulic relationship between nodes. If there is a direct hydraulic connection from node i to node j, the corresponding element in the adjacency matrix is ,otherwise In addition, the positive and negative values ​​of the matrix elements are determined according to the direction of water flow. The elements corresponding to the connection from the upstream node to the downstream node are positive, and vice versa. In this way, the adjacency matrix intuitively presents the hydraulic conduction relationship of the water engineering system, providing a basic relationship description for subsequent calculations.

[0287] S512. Fuzzy membership assignment of weight matrix:

[0288] Construct a weight matrix based on the adjacency matrix. Quantify the strength of the hydraulic connection, where Representation Node For Node The influence weight of .

[0289] ;

[0290] The fuzzy membership assignment method is used to determine the element values ​​of the weight matrix. The fuzzy membership function maps physical parameters to weight values. The strength of the weight matrix is ​​determined by the flow transmission coefficient, hydraulic loss coefficient, and time delay factor. The membership function is defined for different types of hydraulic connections based on actual hydraulic characteristics and engineering experience. For example, for the flow transmission coefficient, if the transmission efficiency is high within a certain range, its fuzzy membership can be set to a value close to 1; if the transmission efficiency is low, it can be set to a value close to 0. In this way, the weight matrix can more accurately describe the hydraulic relationship and provide support for subsequent precise calculations.

[0291] S52. Computational task scheduling:

[0292] S521. Three-layer nested grid division for space-time discretization:

[0293] For the computational tasks of the watershed, a three-layer nested grid division is performed for time and space discretization. In terms of space, the outermost layer is a large grid for the entire watershed, the middle layer is a medium grid divided according to different sub-basins or water project aggregation areas, and the innermost layer is a small grid around specific water projects or key sections. This three-layer nested grid division method can not only grasp the overall situation of the watershed from a macro perspective, but also accurately describe the details of the water flow near water projects and key sections at a micro level. In terms of time, according to the evolution speed of the flood and the response time requirements of the project scheduling, the calculation time is discretized into different time steps. For example, a shorter time step is used in the rapid evolution stage of the flood to ensure the calculation accuracy, and a longer time step is used in the relatively stable stage of the flood to improve the calculation efficiency.

[0294] S522. Topological sorting algorithm for calculating priority:

[0295] The priority of computing tasks is determined based on the topological structure. A topological sorting algorithm is used to sort the water project and section nodes according to their order in the topology. When mapping the water project and section nodes to the graph structure, each node can be regarded as a vertex in the graph, and the association between nodes based on the flow direction or the dependency of the computing results is represented by directed edges.

[0296] When using the topological sorting algorithm, the in-degree is calculated first and the nodes with in-degree zero are added to the queue. , its in-degree needs to be calculated , the in-degree represents the number of directed edges pointing to the node, reflecting the number of dependencies of the node on the calculation results of other nodes. Secondly, the nodes are iterated and the nodes are selected from the queue. Take a node from , this node has the highest computation priority because it has no unfinished predecessor dependencies. Each adjacent node of (That is, there are directed edges Node) perform the following operations: The in-degree of is reduced by 1, that is, ;like The in-degree of becomes 0, then Join the queue , indicating that the node All pre-dependencies have been processed and can enter the calculation process. Continue the iteration process to the queue According to the topological relationship, automatic sorting and extraction are performed according to the above method to provide necessary support for the calculation of the scheduling model.

[0297] S523. Event-driven dynamic step control:

[0298] An event-driven mechanism is introduced to implement dynamic step control and adaptive allocation of computing resources in time and space. During the calculation process, some key events are defined, such as the flood peak reaching a certain node and the scheduling operation of water projects (such as reservoir flood discharge, dam opening, etc.). When these events occur, the time step of the calculation is adjusted according to the nature of the event and the current calculation situation. For example, when the flood peak reaches a certain node, in order to more accurately capture the water flow changes near the peak, the exponential decay model is used. Dynamically reduce the time step, where is the initial time step, It is the attenuation strength coefficient, which controls the speed at which the step size decreases. The range is between 0.1 and 0.3. The larger the value, the faster the step size decreases. is the value of a certain node. is the design flow; when the water project scheduling operation is completed, the progressive stability recovery strategy is triggered according to the stability of the subsequent water flow. Increase the time step, where is the adjusted time step, is the current time step, is the step recovery coefficient, is the decay rate coefficient, is the time when the event that triggers the step size adjustment occurs, The maximum time step allowed by the system. This event-driven dynamic step control can improve computational efficiency while ensuring computational accuracy.

[0299] In this embodiment, S6. Intelligent verification and authentication system:

[0300] S61. Data integrity verification:

[0301] S611. Relevance verification based on knowledge graph:

[0302] Construct a knowledge graph about the water conservancy system in the river basin. Through multi-source heterogeneous data fusion technology, water conservancy entities (reservoirs, dams, sections), hydrological data (reservoir capacity curves, time series flow) and dispatching rules are abstracted into three-layer entities of "physical-data-logic". Use semantic parsing and entity relationship modeling technology to establish a multi-dimensional association network including spatial topological relationships (upstream and downstream connections), data attribution relationships (reservoir capacity curves belong to reservoirs) and rule binding relationships (dispatching rule applicability conditions), and use a spatiotemporal attribute graph model for storage.

[0303] Design a verification framework based on rule engine and graph traversal algorithm. Combine the rule engine to verify the physical constraints of reservoir capacity curve and flood discharge capacity (such as scheduling triggering water level over-limit alarm), and the time series reasoning engine to detect the time and space conflicts of upstream and downstream scheduling actions (such as time period overlap). ), and use graph neural networks to identify abnormal patterns in historical data; locate the root causes of conflicts through knowledge tracing and generate visual reports to achieve interpretable verification of data logic.

[0304] S612. Bayesian network completion of missing data:

[0305] For possible missing data, Bayesian network is used to complete it. First, a Bayesian network model is constructed, with known data as observed variables and missing data as variables to be inferred. The structure and parameters of the Bayesian network are determined according to the causal and probabilistic relationships in the water engineering system. For example, a Bayesian network is constructed based on the water level and flow data of the river section and the probabilistic relationship between them. When the flow data of a certain section is missing, the Bayesian network is used to infer the possible flow value based on the water level data and other relevant factors, thereby completing the missing data and improving the integrity of the data.

[0306] S62. Model Validation System:

[0307] S621. Verification of historical flood scenario reconstruction:

[0308] Use historical flood data to verify the scenario reconstruction. According to the hydrological data of historical flood events (such as flood flow, water level changes, etc.), input them into the constructed model and run the model to obtain the simulation results. Then compare the simulation results with the actual historical records, such as the key indicators such as the propagation time of the flood, the peak water level, and the inundation range. If the simulation results match the actual records within a certain error range, it means that the model is reliable in dealing with similar flood scenarios; if there is a large difference, the model needs to be adjusted and optimized.

[0309] S622. Rolling assimilation correction of real-time data:

[0310] Use real-time data for rolling assimilation correction. During the model operation, real-time hydrological data (such as water level, flow, etc.) are continuously obtained and compared with the model's prediction results. If there is a deviation, the model parameters are adjusted according to the size and nature of the deviation. For example, if the real-time water level is higher than the model's predicted water level, it may be necessary to adjust the flood evolution speed parameters in the model or the scheduling parameters of the water project. Through this rolling assimilation correction, the model can adapt to changes in actual conditions in a timely manner and improve the accuracy of the model.

[0311] Embodiment 2

[0312] The present invention also discloses a visual dynamic modeling system for real-time scheduling of a complex water engineering system in a river basin, the system is used to implement the method described in any one of the first embodiments, the system comprises: a first building module, a fusion configuration module, a second building module and a third building module;

[0313] The first building module is used to construct a real-time scheduling topology visualization diagram of the watershed water engineering system;

[0314] A fusion configuration module, used for fusion configuration of water project node characteristic parameters and river section node characteristic parameters based on the real-time scheduling topology visualization diagram;

[0315] The second building module is used to build a composite flood evolution model based on the node characteristic parameters of the fusion configuration;

[0316] The third building module is used to build an intelligent scheduling model based on the composite flood evolution model.

[0317] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A visual dynamic modeling method for real-time scheduling of complex water engineering systems in a river basin, characterized in that: The method comprises: Construct a real-time scheduling topology visualization diagram of the basin water engineering system; Based on the real-time scheduling topology visualization diagram, the characteristic parameters of the water project nodes and the characteristic parameters of the river section nodes are integrated and configured; Based on the node characteristic parameters of the fusion configuration, a composite flood evolution model is constructed; Based on the composite flood evolution model, an intelligent scheduling model is constructed.

2. The method according to claim 1, characterized in that The method of constructing a real-time scheduling topology visualization diagram of a watershed water engineering system includes: Construct flood forecasting topological base layer based on river basin GIS data; Based on the flood forecasting topological base layer, a multi-dimensional extensible graphic element library for basin water project scheduling is constructed; Based on the multi-dimensional extensible graphic element library, dynamic topological linking is implemented to complete the construction of a real-time scheduling topological visualization diagram of the watershed water engineering system.

3. The method according to claim 2, characterized in that Based on the flood forecasting topological base layer, the method of constructing a multi-dimensional extensible primitive library includes: Based on the physical characteristics and control behaviors of engineering entities, a composite parameter system including geometric parameters, hydraulic parameters, and control logic parameters is designed for each type of element; At the same time, the dynamic injection of new engineering features is achieved through multi-layer open extension interfaces, and the parameter dimensions are customized according to the characteristics of the watershed, so as to establish a structured database that supports the topological reconstruction and intelligent calculation of complex water networks; Among them, the methods of dynamically injecting new engineering features through multi-layer open extension interfaces, customizing parameter dimensions according to basin characteristics, and establishing a structured database that supports complex water network topology reconstruction and intelligent computing include: hierarchical interface protocol design, dynamic parameter binding and conflict detection mechanism, and intelligent parameter recommendation system construction; The layered interface protocol design includes: core interface layer, extended interface layer and watershed adaptation layer; the core interface layer is used to forcibly inherit the hydraulic conductivity properties and spatial topological constraints of the basic graphics elements; the extended interface layer is used to declare the physical dimensions, value range and data verification rules of the newly added parameters based on the domain-specific language DSL, supporting non-programmers to define parameters through configuration files; the watershed adaptation layer has built-in terrain slope , soil permeability coefficient , river network density A watershed feature parameterization template for dynamically generating parameter legitimacy functions; Dynamic parameter binding and conflict detection mechanisms include: Symbolic binding of hydraulic models: Automatic analysis of new parameters using symbolic calculation engine Hydraulic control equation The partial derivative relationship of OK The Jacobian matrix of the columns, Ensure that parameters are automatically integrated into the computing kernel after injection, without the need to reconstruct the model code; build a parameter dependency network based on a directed acyclic graph, detect contradictory combinations through topological sorting, trigger real-time alarms and lock the computing process, and implement conflict self-checking through mathematical symbol binding and graph theory algorithms; For hydraulic structures, an intelligent parameter recommendation system based on engineering feature fingerprints and cross-engineering topology association learning is constructed, which includes: extracting core parameters for each type of hydraulic structure, and generating unique feature fingerprint codes by weighting them through the random forest algorithm. ,in, is the parameter importance weight, It is the core physical parameter value extracted from the project, which is used to quantify the structural or functional characteristics of different projects. Based on the graph convolutional network GCN, a topological map of hydraulic structures is constructed. The nodes are the project feature fingerprints, and the edges are the river connections between projects. The matching score between the new parameters and the target project is calculated by aggregating the neighborhood features, and the optimal parameter combination is dynamically recommended.

4. The method according to claim 2, characterized in that: Based on the multi-dimensional extensible primitive library, the method for implementing topological dynamic linking includes: Decompose the conductive graph elements of river and lake nodes into connector units with direction vectors, and construct a mathematical representation of the topological relationship of the unstructured water network through a set of directed line segments. ,in , are the coordinates of the connector start and end points, where , are the horizontal and vertical coordinates of the connector starting point, , are the horizontal and vertical coordinates of the connector end point, is the cross-sectional area function that changes with water level, where , are the functions of the cross-sectional area of ​​the starting section and the end section as the water level changes, is the dynamic conductivity matrix, which is composed of the water level gradient Drive, where and The conduction coefficient from the starting point to the end point and from the end point to the starting point respectively; Each connector unit realizes asymmetric conduction characteristics through the direction vector weight function, and the dynamic correction formula of the conduction coefficient is: in, is the base conductivity, is the connector direction vector, is an asymmetric regulatory factor, is the gradient response intensity coefficient, , are the starting water level and the ending water level of the connector.

5. The method according to claim 1, characterized in that The method of constructing a composite flood evolution model includes: combining the improved Muskingum segmented algorithm, the hydrodynamic model solved by the unstructured grid Godunov format, and the river-lake coupling model, comprehensively considering different water flow characteristics, and constructing a composite flood evolution model; Among them, the improved Muskingum segmentation calculation model is used to divide the basin into several sub-river sections based on the terrain slope, roughness distribution, cross-section morphology, and hydrological station location attributes. The local Muskingum coefficient is introduced into each sub-river section. That is, the storage parameters and That is, the flow weight factor. The improved Muskingum method adopts a physically related initial parameter mapping mechanism. The initial value sets a nonlinear mapping function according to the sub-reach attributes. , ,in , , is a hyperparameter, The length of the river section, is the slope, For Manning's roughness, For the width of the river, For water depth; Sub-river sections, outflow ,in , , Depend on , Decide and satisfy , For time, is the time step; the inlet flow of the sub-river section and export flow The monitoring data is taken as input and the objective function is constructed , using recursive least squares RLS, real-time update and , and introduced the particle filter algorithm PF, which integrates historical flood event data with a Bayesian framework and periodically modifies hyperparameters , , During the calculation process, the basin is divided into multiple sub-sections, and the improved Muskingum model is applied to flood calculations according to the characteristics of the sub-sections. Then the results of each sub-section are integrated to obtain the flood evolution of the entire basin.

6. The method according to claim 1, characterized in that Methods for building intelligent scheduling models include: Engineering connection relationship matrix analysis and hybrid structure eigenvalue decomposition based on graph theory; Based on the results of analysis and decomposition, multi-mode scheduling of single projects and collaborative scheduling of project groups are carried out to complete the construction of an intelligent scheduling model.

7. The method according to claim 6, characterized in that The method for performing multi-mode scheduling of a single project and coordinated scheduling of a group of projects includes: using an improved starfish optimization algorithm, the steps are: Initialize population size The improved starfish optimization algorithm adopts a constraint satisfaction initialization method based on chaotic mapping when initializing population generation. That is, under the dynamic feasible domain that satisfies the reservoir outflow, the population is initialized by taking the chaotic mapping Cubic mapping strategy. The formula for taking the Cubic mapping to initialize the population is: ； in, is the population position matrix, , is the length of the scheduling period, is the upper boundary of the population position, which represents the upper boundary of the water level process in this problem. is the lower boundary of the population position, indicating the lower boundary of the water level process, , the final generated initialization population matrix is: ； Calculate the initial fitness value of each individual, that is, the objective function value. The minimum objective function value means the maximum peak reduction rate. The corresponding individual position, that is, the reservoir discharge process, is recorded as , Indicates Individual, The location of the dimension, Represents the current iteration number; According to the position update formula of the improved starfish optimization algorithm, the position update of the population individuals is updated, that is, the process of updating the outflow of the reservoir. The position update of the improved starfish optimization algorithm includes the exploration phase and the development phase: Exploration phase: ； ； ； in, is the updated position, is the position before the update, is the best position at the moment. yes are 5 randomly selected dimensions, is a random number between 0 and 1. is the current iteration number, is the maximum number of iterations, is a randomly generated angle parameter, is a dynamically adjusted angle parameter, ; Development phase: Two position update strategies were designed based on the predation and regeneration behaviors of starfish: ； ； ； in, is the updated position, is the position before the update. and is a random number between 0 and 1. is the population size, is the distance between the best position and a randomly selected position in 5 dimensions, and is from Randomly selected from is the current optimal position, are the positions of 5 randomly selected starfish; After the development stage, the tail mutation strategy is introduced, that is, a forced disturbance is added to the position of the population to avoid falling into the local optimal solution. As the number of iterations increases, the intensity of the disturbance gradually decreases to accelerate the convergence of the algorithm in the later stage. In the improved starfish optimization algorithm, the Levy flight strategy is selected to perturb the population position. is the Levy distribution; ； The objective function is calculated for each updated position, i.e., the reservoir discharge process. If the objective function value is better than the objective function value before the position update, the position is updated, otherwise the original position is retained. The cycle is iterated continuously until the loop termination condition is met or the iteration is completed, and the position of the optimal individual is the optimal discharge flow process.

8. A visual dynamic modeling system for real-time scheduling of complex water engineering systems in a river basin, the system being used to implement the method described in any one of claims 1 to 7, characterized in that: The system comprises: a first building module, a fusion configuration module, a second building module and a third building module; The first construction module is used to construct a real-time scheduling topology visualization diagram of the watershed water engineering system; The fusion configuration module is used to fuse and configure the characteristic parameters of the water project nodes and the characteristic parameters of the river section nodes based on the real-time scheduling topology visualization diagram; The second construction module is used to construct a composite flood evolution model based on the node characteristic parameters of the fusion configuration; The third building module is used to build an intelligent scheduling model based on the composite flood evolution model.

Images