A method and system for intelligent irrigation decision-making in irrigation districts

Abstract

This invention provides a method and system for intelligent irrigation decision-making in irrigation districts, belonging to the field of agricultural irrigation technology. The method includes: constructing a lossy directed network structure model representing dynamic water conveyance losses based on the irrigation district's water conveyance system; calculating the gross water demand load of each demand node from the end point by combining water demand characteristics and water conveyance loss parameters; establishing a dynamic water allocation model embedded with node flow balance constraints based on water conveyance loss reduction, with the goal of minimizing the total water supply; solving the model to generate an optimized flow allocation sequence, and generating control commands for the regulating actuator accordingly. This invention achieves systematic dynamic compensation and optimized allocation of water conveyance losses across the entire irrigation district, improving the spatiotemporal matching accuracy and overall utilization efficiency of water resources, and effectively overcoming the water supply-demand mismatch problem caused by neglecting network losses.

Description

Technical Field

[0001] This invention belongs to the field of agricultural irrigation technology, specifically relating to a method and system for intelligent irrigation decision-making in irrigation districts. Background Technology

[0002] Irrigation district water resource allocation is a core element in ensuring agricultural production safety and improving water resource utilization efficiency. Its core task is to fairly, promptly, and adequately deliver limited water resources to various dispersed water-using units to meet the dynamic needs of different crops at different growth stages. Currently, with the development of information technology and modern control theory, irrigation district allocation has gradually shifted from relying entirely on manual experience to using mathematical models for decision-making assistance. Common methods include water allocation based on linear programming and scheduling control based on simple rules.

[0003] However, existing methods still face significant technical bottlenecks when dealing with large-scale, multi-level, and structurally complex real-world irrigation systems. Most optimization models focus on the water balance between water sources and users, simplifying the complex water conveyance network into one or more fixed transport efficiency coefficients. This simplification fails to accurately characterize the inevitable water losses, such as leakage and evaporation, that occur as water flows through long main canals, branch canals, and distribution canals. Since these losses accumulate along the flow path and are dynamically related to flow rate and canal conditions, simply multiplying by a global efficiency coefficient cannot reflect the spatial transmission effect and flow-dependent characteristics of these losses within the network. This leads to a significant systematic error between the theoretical water supply calculated by the model and the actual water received in the fields.

[0004] In actual operation, the aforementioned technical deficiencies often manifest in the following ways: to ensure downstream water supply meets standards, it is necessary to increase upstream water release, leading to over-irrigation and water waste in upstream areas; or, due to underestimating cumulative losses along the course, downstream areas may not receive sufficient water during critical water demand periods, affecting crop yield and quality, resulting in an unreasonable situation of upstream flooding and downstream drought. Furthermore, the flow balance constraints of existing methods often only consider the simple conservation of inflow and outflow at nodes, failing to reflect the differentiated water transport losses of each directional path at the constraint level. This causes the optimization results to deviate from actual hydraulic behavior, making it difficult to truly achieve precise water allocation.

[0005] Meanwhile, existing methods, when modeling the demand side, typically use the net irrigation water requirement of crops directly as the flow load of demand nodes, ignoring the multi-stage water transport losses that occur from upstream network nodes to field water units. This forward allocation approach fails to calculate the gross water demand that compensates for losses from the end-user demand, resulting in the actual effective water reaching the field being lower than the crop's demand, thus reducing the irrigation guarantee rate. On the other hand, existing methods often struggle to simultaneously coordinate the complex coupling relationships among multiple water sources, crops, time periods, and constraints (such as channel flow capacity, gate operation restrictions, and channel water storage safety), lacking globality and dynamic adaptability in decision-making, and lacking effective bottleneck identification and load adjustment mechanisms in the face of emergency situations with insufficient water supply capacity. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of the aforementioned background technology and provide a method and system for intelligent irrigation decision-making in irrigation districts. This enables systematic dynamic compensation and optimized allocation of water conveyance losses across the entire irrigation district, thereby improving the accuracy of water resource spatiotemporal matching and overall utilization efficiency, and effectively overcoming the problem of water supply and demand mismatch caused by neglecting network losses.

[0007] The technical solution adopted in this invention is: a method for intelligent irrigation decision-making in irrigation districts, comprising the following steps: Based on the spatial connection relationship and engineering parameters of the irrigation district water conveyance system, a lossy directed network structure model is constructed to characterize the dynamic water conveyance loss generated by water flow in a directed path; the network nodes in the lossy directed network structure model include water source nodes and demand nodes. Based on the water demand characteristics of each water-using unit, the net irrigation demand in the future scheduling cycle is determined, and combined with the water conveyance loss parameters in the lossy directed network structure model, the gross water demand load at each demand node is calculated from the end demand. With minimizing the total water supply as the global optimization objective, a dynamic water allocation model is established based on the topological connections of the lossy directed network structure model, the capacity upper bound constraints of each directed path, and the gross water demand load of each demand node. This model incorporates spatiotemporal coupling characteristics across multiple time periods. The dynamic water allocation model embeds node flow balance constraints based on the water conveyance loss parameters. These constraints require that the net inflow of each directed path into a node, after reduction by its corresponding water conveyance loss parameters, remains conserved with the total outflow from that node. The dynamic water allocation model is solved to generate the optimized flow allocation sequence of each directed path in each time period within the scheduling cycle; Based on the optimized flow distribution sequence, a control command sequence for regulating the actuator is generated.

[0008] In the above technical solution, the construction of a lossy directed network structure model representing the dynamic water transport loss generated by water flow in a directed path includes: By analyzing the geospatial data of the irrigation area and the layout data of water conservancy projects, all diversion points, confluence points and water transmission terminals are extracted as network nodes. The water intake is defined as the water source node, corresponding demand nodes are created for each water-using unit, and the physical water transmission channels connecting adjacent nodes are defined as directed paths with a single flow direction attribute. For each directed path, based on the geometric parameters of the cross-section of the corresponding physical water conveyance channel, the physical roughness of the inner wall, and the gravity gradient characteristics along the path, the maximum safe transport flux of the directed path under steady-state flow is calculated using a fluid dynamics resistance model, and this maximum safe transport flux is set as the upper limit constraint of the capacity of the directed path.

[0009] In the above technical solution, the water conveyance loss parameters of each directed path in the lossy directed network structure model are determined in the following way: Obtain permeability test data or historical operation monitoring data for each water conveyance channel and establish a mapping relationship between water conveyance loss and flow rate; Based on the mapping relationship, the water loss parameter is configured for each directed path. The water loss parameter represents the proportion of flow loss during the process of fluid being transported from the starting point to the end point of the directed path, and is dynamically adjusted according to the mapping relationship as the flow rate changes through the directed path.

[0010] In the above technical solution, the step of calculating the gross water flow load at each demand node from the end-user demand includes: Identify the directed path connecting each demand node and obtain the water conveyance loss parameters corresponding to that directed path; A local water use factor is introduced to characterize the process from the water conveyance network to the field water use unit, which covers leakage, evaporation and management losses in the water distribution process; By using the water loss parameters and the local water use coefficient, the net irrigation demand is converted into a gross water flow load that must be extracted from the upstream network nodes, ensuring that the effective water volume reaching the end after deducting the multi-level cumulative losses during transmission meets the net irrigation demand.

[0011] In the above technical solution, establishing a dynamic water allocation model that includes multi-time-period spatiotemporal coupling characteristics includes: The complete scheduling cycle is discretized into several consecutive time periods, and the flow value of each directed path in each time period is defined as the decision variable to be solved. Construct a global objective function that requires the total outflow from the water source node to reach its minimum value across all time periods; The node flow balance constraint is constructed such that, for any network node other than the water source node, at any time period, the sum of the net values ​​of all upstream directed path flows pointing to that node after deducting the losses determined by their respective water conveyance loss parameters is equal to the sum of all downstream directed path flows pointing from that node. Construct demand response constraints so that the effective flow into each demand node in each time period is not less than the gross water demand load corresponding to that demand node. Construct directed path capacity constraints such that, within each time period, the flow value of each directed path is between zero and the upper bound constraint of the directed path's capacity. Establish water source capacity constraints so that the total flow rate out of the water source node does not exceed the maximum available water volume of the current water source in each time period.

[0012] In the above technical solution, the dynamic water allocation model also includes the following constraints: Gate operation constraints are used to limit the flow change of each directed path between adjacent time periods to not exceeding the maximum regulation rate of the corresponding regulating actuator; Time period connection constraints are used to limit the change in water storage volume of the physical water conveyance channel corresponding to each directed path between adjacent time periods to meet the inflow-outflow balance and keep the water storage volume within a safe range.

[0013] In the above technical solution, based on the crop type, crop planting area, crop growth stage and forecast meteorological data corresponding to each water use unit, the crop water balance model is used to calculate the net irrigation demand of each water use unit in each period of the future scheduling cycle.

[0014] In the above technical solution, the solution process of the dynamic water allocation model includes: The global optimization problem spanning multiple time periods is decomposed into a series of sub-problems of single-time period traffic allocation arranged in chronological order; Construct a standard flow network model with gain attributes that is topologically equivalent to the lossy directed network structure model, wherein the water conveyance loss parameter is transformed into the gain factor attribute of the directed path in the standard flow network model, and the gain factor characterizes the retention ratio of unit flow after transmission through the directed path. The gross water demand load of each demand node is converted into the lower bound constraint of the flow of the corresponding directed path in the standard flow network model. On the standard flow network model, a maximum flow optimization algorithm with lower bound constraints on flow and gain factor attributes is run to search for a feasible flow distribution scheme that minimizes the outflow of water source while satisfying the lower bounds of flow at all demand nodes.

[0015] In the above technical solution, the solution process also includes bottleneck diagnosis and load adjustment steps: If there is no feasible network flow that satisfies the lower bound of the flow of all demand nodes in the single-time flow allocation subproblem during a specific time period, then the bottleneck directed path or bottleneck node set that restricts the global water supply capacity can be identified based on the maximum flow minimum cut theorem. Based on the preset irrigation priority rules, the gross water demand load of the demand nodes in the bottleneck area is dynamically reduced proportionally. The maximum flow optimization algorithm is rerun based on the reduced load until a feasible flow distribution scheme for the current time period is obtained.

[0016] In the above technical solution, the generation of the control command sequence for the regulating actuator includes: Obtain the flow characteristic curves of each regulating gate in the irrigation area, and establish a mapping model between the flow through the gate, the water level difference before and after the gate, and the gate opening degree. Based on the optimized flow distribution sequence of each directed path, combined with the water level status data of real-time monitoring or simulation calculation, the target opening degree of each regulating gate in each time period is calculated in reverse using the mapping model. The target opening degree is encapsulated as a control command with a timestamp, and its rate of change is smoothed to conform to the mechanical operation specifications of the adjustment mechanism.

[0017] In the above technical solution, the control command sequence is sent to the irrigation district control terminal to execute irrigation scheduling.

[0018] This invention provides a system for intelligent irrigation decision-making in irrigation districts, the system being used to implement the above-mentioned method, comprising: The canal system topology modeling module is used to construct a lossy directed network structure model that represents the dynamic water conveyance loss generated by water flow in a directed path, based on the spatial connection relationship and engineering parameters of the irrigation area water conveyance system; the network nodes in the lossy directed network structure model include water source nodes and demand nodes. The irrigation demand quantification module is used to determine the net irrigation demand in the future scheduling cycle based on the water demand characteristics of each water-using unit, and to calculate the gross water demand load at each demand node from the end demand by combining the water conveyance loss parameters in the lossy directed network structure model. The optimization decision modeling module is used to establish a dynamic water allocation model with multi-period spatiotemporal coupling characteristics, based on the topological connections of the lossy directed network structure model, the capacity upper bound constraints of each directed path, and the gross water demand load of each demand node, with the global optimization objective of minimizing the total water supply. The dynamic water allocation model embeds node flow balance constraints, which require that the net inflow of each directed path into a node, after being reduced by its corresponding water loss parameters, remains conserved with the total outflow from that node. The flow optimization solution module is used to solve the dynamic water allocation model and generate the optimized flow allocation sequence of each directed path in each time period within the scheduling cycle. The intelligent irrigation execution module is used to generate a control command sequence for adjusting the execution mechanism based on the optimized flow distribution sequence.

[0019] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the method described above.

[0020] The present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0021] The beneficial effects of this invention are as follows: The intelligent irrigation decision-making method, system, equipment, and medium provided by this invention construct a lossy directed network structure model that characterizes the dynamic water conveyance loss generated by water flow in a directed path. Based on this model, dynamic water demand characteristic data of each water-using unit are integrated, and the gross water demand load of each demand node, taking into account transmission loss, is calculated backward from the end demand. Furthermore, a dynamic water allocation model is established with minimizing the total water supply as the global optimization objective and embedding water conveyance loss parameters in the node flow balance constraints for dynamic reduction and balance. Accurate optimized flow allocation sequences for each directed path are obtained through time-period decomposition and network flow optimization with gain attributes. Finally, this sequence is converted into specific control commands for the regulating actuator and issued for execution, thereby realizing systematic dynamic compensation and optimized allocation of water conveyance loss throughout the entire irrigation area. This invention achieves the technical effect of significantly improving the spatiotemporal matching accuracy and overall utilization efficiency of water resources, effectively overcoming the water supply and demand mismatch problem caused by ignoring network losses, and solving the unreasonable situation of "upstream flooding and downstream drought" in traditional methods. It provides complete technical support for the precise scheduling and intelligent management of water resources in irrigation areas.

[0022] Furthermore, the intelligent irrigation decision-making method for irrigation districts proposed in this invention abstracts the water conveyance system into a lossy directed network structure model and embeds node flow balance constraints based on water conveyance loss parameters into the dynamic water allocation model. This constraint requires that the net inflow of each directed path flow into a node, after being reduced by its corresponding water conveyance loss parameter, remains conserved with the total outflow from that node. This design fundamentally differs from the traditional constraint mode in existing technologies that only considers simple flow conservation, accurately characterizing the path loss characteristics and spatial differences of water flow during network transmission. By using a technical approach of reverse calculation of gross water demand from end-point demand, systematic compensation for multi-level cumulative water conveyance losses is achieved, ensuring that the effective water volume reaching the end-point after deducting losses at each level during transmission can meet the net irrigation demand. With minimizing the total water supply as the global optimization objective, water resource consumption is reduced to the minimum while ensuring irrigation demand, significantly improving water resource utilization efficiency. This method ultimately generates a control command sequence for the regulating actuator, realizing a complete closed loop from decision optimization to execution control, demonstrating significant advantages of strong theoretical innovation and high practical value.

[0023] Furthermore, this invention proposes a method for constructing a lossy directed network structure model. By analyzing the geospatial data of the irrigation district and the layout data of water conservancy projects, all branching points, confluence points, and water delivery terminals are accurately extracted as network nodes, and the physical water delivery channels connecting adjacent nodes are defined as directed paths with a single flow direction attribute. This method uses a fluid dynamics resistance model to calculate the maximum safe transport flux of each directed path under steady-state flow and sets it as an upper capacity constraint, ensuring the physical feasibility and engineering safety of the model from an engineering hydraulics perspective. This refined modeling approach based on actual engineering parameters enables the network structure model to truly reflect the topological connections and hydraulic characteristics of the irrigation district's water delivery system, providing a reliable physical basis for subsequent optimization decisions and avoiding systematic biases caused by simplified modeling.

[0024] Furthermore, this invention clarifies a method for determining water conveyance loss parameters. By acquiring permeability test data or historical operational monitoring data from each physical water conveyance channel, a mapping relationship between water conveyance loss and flow rate is established. Based on this mapping relationship, water conveyance loss parameters that are dynamically adjusted according to flow rate changes are configured for each directed path. This technical feature overcomes the limitations of using fixed loss coefficients in existing technologies, accurately characterizing the actual dynamic changes in water conveyance loss with flow rate conditions. This enables the optimization model to accurately calculate the loss under different flow rate conditions, significantly improving the model's accuracy in representing actual hydraulic behavior and laying a parameter foundation for achieving precise water allocation and loss compensation.

[0025] Furthermore, this invention describes a method for reverse-engineering gross water demand load from end-user demand. By identifying the directed paths connecting demand nodes and obtaining their water transport loss parameters, a local water use coefficient covering leakage, evaporation, and administrative losses is introduced. This accurately converts net irrigation demand into gross water demand load that must be extracted from upstream network nodes. This reverse-engineering approach is one of the core innovations of this invention, fundamentally different from the traditional forward allocation approach. It ensures that after deducting the cumulative losses during transmission, the effective water volume reaching the end-user can fully meet the net irrigation needs of crops, effectively solving the downstream water shortage problem caused by ignoring the cumulative effect of losses, and significantly improving irrigation security and water supply accuracy.

[0026] Furthermore, this invention elucidates a method for constructing a dynamic water allocation model, including discretizing the scheduling cycle into continuous time periods, defining the flow of each directed path in each time period as decision variables, constructing a global objective function, and establishing multiple constraints such as node flow balance constraints, demand response constraints, directed path capacity constraints, and water source capacity constraints. Among these, the design of the node flow balance constraint is particularly crucial. This constraint requires that for any network node other than the water source node, the sum of the net values ​​of the upstream directed path flow after deducting the losses determined by their respective water conveyance loss parameters equals the sum of the downstream directed path flow. This innovative constraint form embeds differentiated water conveyance losses into the flow conservation equation, achieving an organic unity of loss compensation and flow optimization. The demand response constraint ensures that the effective flow obtained by each demand node is not less than its gross water demand load, guaranteeing the reliable satisfaction of irrigation needs. The synergistic effect of multiple constraints enables the optimization model to minimize water supply while meeting engineering constraints, demonstrating the technical advantages of global optimization and multi-constraint coordination.

[0027] Furthermore, this invention adds two important types of constraints: gate operation constraints and time period connection constraints. Gate operation constraints limit the flow variation of each directed path between adjacent time periods to no more than the maximum regulation rate of the corresponding regulating actuator, avoiding water flow shocks and mechanical damage caused by sudden flow changes, and ensuring the engineering feasibility of the scheduling scheme. Time period connection constraints ensure that the water storage capacity of the physical water conveyance channel corresponding to each directed path changes between adjacent time periods while satisfying inflow-outflow balance and remaining within a safe range, maintaining the hydraulic safety of channel operation. The introduction of these two types of constraints makes the optimization model closer to actual engineering operating conditions, improving the feasibility and safety of the scheduling scheme.

[0028] Furthermore, this invention clarifies the calculation method for net irrigation demand. Based on the crop type, crop planting area, crop growth stage, and forecast meteorological data corresponding to each water-using unit, a crop water balance model is used for accurate calculation. This method comprehensively considers multiple factors such as crop water requirement characteristics, changes in meteorological conditions, and soil moisture status, and can accurately predict the actual water demand of each water-using unit in each period of the future scheduling cycle. This provides scientific and reliable demand data support for subsequent gross water demand load calculation and water allocation optimization, reflecting the technical characteristics of demand-driven scheduling decision-making.

[0029] Furthermore, this invention proposes an efficient model-solving method that deconstructs the cross-time-period global optimization problem into a series of time-sequential single-time-period flow allocation subproblems, and constructs a standard flow network model with gain attributes. Specifically, the water transfer loss parameter is transformed into a gain factor attribute of the directed path, characterizing the retention ratio of a unit flow after transmission; the gross water demand load of the demand nodes is transformed into a flow lower bound constraint for the corresponding directed path. Running a maximum flow optimization algorithm with flow lower bound constraints and gain factor attributes on this standard flow network model can efficiently search for feasible flow distribution schemes that satisfy all demand constraints and minimize water outflow. This solution strategy transforms the complex nonlinear multi-time-period coupled problem into an efficiently solvable standard network flow problem, significantly reducing computational complexity and improving solution efficiency, making real-time optimization scheduling of large-scale irrigation districts possible.

[0030] Furthermore, this invention proposes a bottleneck diagnosis and load adjustment mechanism. When there is no feasible network flow that satisfies the lower bound of the flow of all demand nodes during a specific time period, the directed path or set of bottleneck nodes that restricts the global water supply capacity is identified based on the maximum flow minimum cut theorem. Then, according to a preset irrigation priority rule, the gross water demand load of the demand nodes in the bottleneck-affected area is dynamically reduced proportionally. This mechanism effectively solves the scheduling decision-making problem under insufficient water supply capacity. Through scientific bottleneck identification and priority reduction strategies, it maximizes irrigation benefits under resource-constrained conditions, demonstrating the robustness and adaptability of this invention in responding to emergencies.

[0031] Furthermore, this invention describes a method for generating control command sequences. By acquiring the flow characteristic curves of each regulating gate and establishing a mapping model between the flow rate through the gate, the water level difference before and after the gate, and the gate opening degree, the target opening degree of each regulating gate at each time period is calculated in reverse using the optimized flow allocation sequence and real-time water level data. Finally, this is encapsulated into timestamped control commands and subjected to rate-of-change smoothing. This method transforms abstract optimized flow rates into executable gate operation commands, achieving a precise mapping from the decision-making level to the execution level. The smoothing process ensures that the commands conform to the mechanical operation specifications of the regulating mechanism, guaranteeing the safety and stability of the control process.

[0032] Furthermore, this invention clarifies the execution method of the control command sequence, sending the commands to the irrigation district control terminal to execute irrigation scheduling. This technical feature achieves seamless integration of optimization decision-making and actual execution, enabling the technical solution of this invention to be directly applied to the automated control system of the irrigation district, forming a complete decision-making-execution closed loop, improving the level of intelligent management of the irrigation district, and possessing significant engineering application value.

[0033] Furthermore, this invention provides a system for intelligent irrigation decision-making in irrigation districts, comprising five functional modules: topology modeling, demand quantification, model building, optimization solution, and execution control. Each module has a clear division of labor and works collaboratively to complete core functions such as network structure modeling, demand load calculation, optimization model building, flow sequence solution, and control command generation, forming a complete intelligent irrigation decision-making system architecture. This system architecture is modular and scalable, facilitating engineering implementation and functional upgrades, and providing a system-level technical solution for the intelligent transformation of irrigation districts.

[0034] Furthermore, the present invention provides a computer device and a computer-readable storage medium, both of which are capable of implementing the methods described above. By solidifying the technical solution of the present invention into a computer program and deploying it in a corresponding hardware device or storage medium, the intelligent irrigation decision-making method for irrigation districts can be promoted and applied in the form of a software product. This facilitates integration and connection with existing irrigation district information systems, lowers the threshold for technology promotion, expands the application scope of the present invention, and has good industrialization prospects. Attached Figure Description

[0035] Figure 1 A flowchart illustrating a method for intelligent irrigation decision-making in irrigation districts provided by the present invention; Figure 2 This is a schematic diagram illustrating the process of constructing a lossy directed channel network topology in one optional embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of a smart irrigation decision-making system for irrigation districts provided by the present invention. Detailed Implementation

[0036] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments to facilitate a clear understanding of the present invention, but these descriptions do not constitute a limitation on the present invention.

[0037] refer to Figure 1The document presents a flowchart illustrating a method for intelligent irrigation decision-making in irrigation districts provided in this application. In the description of this embodiment, it should be noted that, unless otherwise explicitly stated or limited, the term "lossy directed network structure model" refers to a network topology model capable of characterizing the dynamic water loss generated by water flow in a directed transmission path. In this embodiment, it is specifically implemented as a lossy directed canal network topology diagram. "Directed path" refers to a connection channel in the network with a single flow direction attribute. In this embodiment, it is specifically represented as a directed edge pointing from an upstream node to a downstream node. "Water source node" refers to the water supply starting point in the network, corresponding to water intake points such as reservoir outlets and pump station outlets in this embodiment. "Demand node" refers to a node representing a water-using terminal in the network, corresponding to the water inlet of each irrigation area in this embodiment. "Water loss parameter" refers to a parameter characterizing the proportion of flow loss during the directed path transmission process, specifically manifested in this embodiment as a water loss coefficient, and this parameter changes dynamically with the flow rate. "Regulation actuator" refers to a regulating device controlling the flow rate of a directed path, specifically various regulating gates in this embodiment. The specific meanings of the above terms should be understood in conjunction with the overall technical solution of this application.

[0038] Specifically, the method includes the following steps: S1. Based on the spatial connectivity and engineering parameters of the irrigation district's channels, a lossy directed canal network topology diagram representing the water flow transmission and distribution process is constructed. In this embodiment, the lossy directed network structure model is specifically implemented as a lossy directed canal network topology diagram.

[0039] Specifically, the technical principle of this step is based on graph theory and canal hydraulics, abstracting the canal system of the physical irrigation area into a quantifiable and computable network model. By characterizing the two core features of directional water flow transmission and dynamic loss along the canal, it provides a realistic physical basis for subsequent irrigation decisions. First, basic data collection and standardization are completed. The acquisition of spatial connections between canals relies on geographic information system (GIS) technology. Topographic data of the irrigation area is obtained through UAV aerial surveys or ground-based lidar scanning, extracting the centerline coordinates of canals at all levels, including main canals, branch canals, distribution canals, and farm canals, clarifying the locations of intersections, water distribution points, and control points between canals, forming a spatial topological framework. The collection of engineering parameters requires combining on-site surveys and historical engineering archives. Core parameters include canal cross-sectional parameters, hydraulic parameters, geometric parameters, and leakage parameters. Canal cross-sectional parameters cover cross-sectional type, bottom width, slope coefficient, and maximum allowable water depth; hydraulic parameters mainly include roughness, determined by the lining material; geometric parameters include canal length and the elevation difference between the starting and ending points; and leakage parameters are soil permeability coefficients, determined based on experimental data of soil types along the canal.

[0040] When constructing the topology graph based on the above data, the definitions and attribute assignments of nodes and edges are clearly defined. Nodes include water source nodes, control nodes, and demand nodes. Water source nodes are the starting points for water supply, such as reservoirs and pumping stations; control nodes are the locations of sluice gates and diversion gates; and demand nodes are the water inlets of irrigation areas. Each node needs to be labeled with a unique number, geographical coordinates, node type, and associated edge information. Edges correspond to actual channel segments, and directed paths are used to represent the direction of water flow from upstream nodes to downstream nodes. Each edge is associated with core attribute parameters, including the channel segment number, the corresponding set of engineering parameters, the water conveyance loss coefficient function, the maximum flow capacity, and the minimum flow capacity. Among these, the construction of the water conveyance loss coefficient function is the core of loss-featured data. The traditional fixed coefficient method is abandoned, and a dynamic loss model is derived based on hydraulic formulas. The water conveyance loss coefficient is the specific implementation form of the water conveyance loss parameter. The Manning formula is used to calculate the flow capacity and head loss along the channel segment. The Manning formula expression is as follows: ; In the above formula, Q represents the flow rate of the channel section, n represents the channel roughness, A represents the cross-sectional area of ​​the channel, R represents the hydraulic radius, and S represents the hydraulic gradient. For a trapezoidal cross-section, the calculation of the cross-sectional area A follows the corresponding geometric relationship; the hydraulic radius R is the ratio of the cross-sectional area to the wetted perimeter, which is also calculated according to the cross-sectional type following the corresponding geometric relationship; the hydraulic gradient S is the ratio of the elevation difference between the start and end points of the channel section to the length of the channel section. Combined with the calculation of friction loss, the leakage loss flow rate is derived using Darcy's law, and the derived formula is as follows: ; In the above formula, Let represent the leakage loss flow rate, k represent the soil permeability coefficient, L represent the channel length, b represent the channel bottom width, and h represent the actual water depth. Therefore, the total water conveyance loss flow rate of the channel section is the sum of the flow loss corresponding to the friction head loss and the leakage loss flow rate. The flow loss corresponding to the friction head loss is obtained by inversely calculating the Manning formula. The expression for the total water conveyance loss flow rate is as follows: ; In the above formula, Represents the total water loss during transport. This represents the flow loss corresponding to the head loss along the pipeline. This represents the leakage loss flow rate. The final dynamic function of the water conveyance loss coefficient is expressed as follows: ; In the above formula, The coefficient represents the water loss, and Q represents the flow rate of the channel section. This represents the total water loss flow rate. This function is directly related to the flow rate Q, enabling a precise characterization of the loss coefficient's dynamic changes with flow rate.

[0041] The topology graph is stored using an adjacency list structure, facilitating efficient computer retrieval and calculation. Each node in the adjacency list corresponds to a linked list, and the linked list elements store core information such as the target node number of the adjacent edge, the water conveyance loss coefficient function, and the maximum flow capacity. After construction, validation was performed. By selecting typical flow conditions from historical scheduling data, the total water conveyance loss of a certain path in the topology graph was calculated and compared with the loss data from actual monitoring to ensure that the loss calculation error was controlled within a reasonable range. At the same time, the consistency between the node connection relationship and the actual canal system was verified to avoid topology breaks or incorrect connections, ensuring that the model can truly reflect the transmission and loss process of water flow in the canal system, providing a reliable basic model support for subsequent water demand calculation and optimization decisions.

[0042] S2. Based on the crop type, crop planting area, crop growth stage and forecast meteorological data corresponding to each irrigation area, calculate the net irrigation water demand of each demand node in the future multiple scheduling periods; based on the net irrigation water demand and the water conveyance loss coefficient in the lossy directed canal network topology diagram, calculate the gross water demand load of each demand node in each period.

[0043] Specifically, the calculation of net irrigation water requirement is based on the principle of crop water balance, which states that the net irrigation water requirement of a crop during a certain growth period is equal to the difference between the crop's water requirement and the effective water supply. The core formula is as follows:

[0044] In the above formula, Represents net irrigation water requirement. Represents crop water requirements. Represents effective precipitation. G represents the initial soil moisture content, and G represents the groundwater recharge. The calculation of each parameter requires the combination of multi-source data and standardized models: crop type is determined by combining remote sensing image interpretation and field surveys to clarify the dominant crop and mixed crop ratio in each irrigation area; crop planting area is extracted by geographic information system technology to extract the boundary of irrigation area and calculated by combining vegetation index inversion from remote sensing imagery; crop growth stage is determined based on the accumulated temperature model, by statistically analyzing the average daily temperature of the region over the years to calculate the cumulative accumulated temperature of the crop from sowing to the present, and comparing it with the accumulated temperature demand curve of the target crop to determine the growth stage corresponding to each scheduling period.

[0045] The forecast meteorological data were obtained from the output of a regional numerical weather prediction model, after adjusting the spatiotemporal resolution. Key data included daily average temperature, daily precipitation, sunshine duration, wind speed at a specific altitude, and relative humidity. Crop water requirements were calculated using the Penman-Monteith formula recommended by the FAO (Food and Agriculture Organization of the United Nations). This formula comprehensively considers meteorological factors and crop physiological characteristics, offering high accuracy. The formula consists of two parts: first, the evapotranspiration of a reference crop is calculated; then, the crop water requirement is obtained after correction using a crop coefficient. The specific formula is as follows: ; ; In the above formula, Represents reference crop evapotranspiration; The slope of the saturated vapor pressure curve is calculated from the daily average temperature. G represents net radiation on crop surface, derived from the sunshine duration and atmospheric radiation transfer model; G represents soil heat flux, which can be approximated as 0 in the short-term scheduling. Represents the wet / dry constant. Represents wind speed at a specific altitude; This represents the saturated vapor pressure, calculated from the daily average temperature. This represents the actual water vapor pressure, which is derived from relative humidity and saturated water vapor pressure. Represents crop water requirements. The crop coefficient is determined based on crop type and growth stage. Effective precipitation is calculated using a modified CN method. The CN value is determined based on soil type, and surface runoff and infiltration are calculated in conjunction with precipitation. The portion of infiltration that can be absorbed by crop roots is considered effective precipitation. Initial soil moisture content is monitored using soil moisture sensors at the main root distribution layer of the crop, and the data is the average of multiple monitoring points. Groundwater recharge is calculated based on regional hydrogeological data. When the groundwater depth is shallow, the corresponding recharge coefficient formula is used; when the depth is large, it can be ignored.

[0046] The core of calculating gross water demand load is the reverse derivation considering water conveyance losses in the canal system. That is, the net irrigation water demand of a demand node is the actual usable water volume after transmission losses through multiple upstream canals. Therefore, the gross water demand must still guarantee the net water demand after these transmission losses. First, based on the topology of a lossy directed canal network, Dijkstra's algorithm is used to search for the optimal water conveyance path from the water source node to each demand node, extracting the water conveyance loss coefficient function for all canal segments along the path. Let the net irrigation water demand of a certain demand node during scheduling period t be... The scheduling period length is Then, the net water demand of this node is the ratio of the net irrigation water demand to the length of the scheduling period, expressed as follows: ; In the above formula, This represents the net water demand of the demand node during the scheduling period t. This represents the net irrigation water demand of the demand node during the scheduling period t. This represents the length of the scheduling period. Since water flows from the source node to the demand node through multiple channel segments, the input and output flows of each channel segment satisfy the following condition: the output flow equals the product of the input flow and the water loss coefficient of that channel segment, and the output flow of the preceding channel segment equals the input flow of the following channel segment. Ultimately, the net water demand of the demand node equals the product of the water loss coefficients of all channel segments along the path from the source node to the demand node and the initial input flow. Because the water loss coefficient is a function of the flow rate, the Newton-Raphson iterative method is used to solve for the initial input flow, which is the gross water demand load of the demand node. The initial iteration value is set as the ratio of the net water demand to the product of the loss coefficients at the rated flow of each channel segment. After each iteration, the deviation between the actual net flow and the target net water demand is calculated. The iteration stops when the deviation is less than a set threshold. The initial input flow obtained at this point is the gross water demand load of the demand node for that period. The gross water demand load for each scheduling period needs to be calculated independently, as meteorological data and crop growth stages change over time, ensuring the time-varying nature and accuracy of the load.

[0047] S3. Based on the structure of the lossy directed canal network topology, the channel capacity attributes, and the gross water demand load of each demand node in each time period, a time-varying network flow optimization decision model is established with the goal of minimizing the total water supply. The time-varying network flow optimization decision model includes node flow balance constraints with embedded water conveyance loss coefficients. The node flow balance constraints dynamically reduce the flow of each edge flowing into any network node according to the water conveyance loss coefficient corresponding to each edge, so that the reduced net inflow is equal to the total outflow from any network node.

[0048] Specifically, the core technology of this step is to construct a mathematical model that balances global optimization and dynamic adaptation. Based on time-varying network flow theory, the irrigation decision-making process in the irrigation district is transformed into an optimization problem under multiple constraints. The goal is to minimize the total water supply while meeting the water demand of all nodes, thereby achieving efficient water resource utilization. The model construction requires clearly defining three key elements: decision variables, objective function, and constraints. Among these, the node flow balance constraint with embedded water conveyance loss coefficients is a key innovation that distinguishes it from traditional models.

[0049] The decision variable is defined as the flow rate of all channel edges in a lossy directed channel network during each scheduling period, i.e., the flow rate value of the channel edge in different scheduling periods. The range of values ​​for the decision variable must satisfy the flow capacity constraint of the channel edge, that is, the value of the decision variable must be between the minimum safe flow rate and the maximum flow capacity of the channel edge. The minimum safe flow rate is used to avoid siltation caused by channel drying, and the maximum flow capacity is calculated by Manning's formula combined with the maximum allowable water depth of the channel.

[0050] The objective function is designed based on the principle of efficient water resource utilization, with the goal of minimizing the total water supply. The total water supply is the sum of the output flow of all water source nodes during all scheduling periods, and its expression is as follows: ; In the above formula, Z represents the total water supply, T represents the total number of future scheduling periods, and S represents the set of water source nodes. This represents the output flow rate of water source node s within the scheduling period t. The physical meaning of this objective function is to optimize the flow distribution at each channel edge, while ensuring the gross water demand load of all demand nodes, to minimize the total water supply from the water source, reduce water waste, and simultaneously reduce pumping energy consumption.

[0051] Constraints are crucial for ensuring the feasibility and rationality of the model. They primarily include four categories: node flow balance constraints, channel flow capacity constraints, gate operation constraints, and time-period coordination constraints. Among these, node flow balance constraints are the core innovation. The technical principle of node flow balance constraints is that for any node in the network, the flow into that node from all channel sides must be reduced by their respective water loss coefficients to maintain balance with the flow out of that node from all channel sides. That is, the net inflow after reduction equals the total outflow, ensuring the conservation of water flow mass at the node. The mathematical expression is as follows: ; In the above formula, v represents any node in the network. The set of channel edges representing the inflow node v. The set of channel edges representing the outflow node v. This represents the traffic on channel edge e during scheduling period t. The representative channel e in traffic The water conveyance loss coefficient is calculated as follows. For demand nodes, the reduced net inflow must equal the gross water demand load of that node, as expressed mathematically:

[0052] In the above formula, Represents demand nodes, Represents the node of inflow demand The channel edge collection, This represents the traffic on channel edge e during scheduling period t. The representative channel e in traffic The water conveyance loss coefficient is below. Representative demand node The gross water demand load during scheduling period t. For a water source node, the outflow is equal to the water supply of that node, as expressed mathematically below: ; In the above formula, Representing water source nodes, Represents the outflow water source node The channel edge collection, This represents the traffic on channel edge e during scheduling period t. Represents the water source nodes within the scheduling period t. The constraint, by dynamically reducing the inflow flow, accurately characterizes the impact of water conveyance losses on node balancing, avoiding the flow allocation bias caused by neglecting losses in traditional models.

[0053] The channel flow capacity constraint ensures that water does not overflow the channel and meets the minimum safe flow rate, expressed as follows: ; In the above formula, The minimum safe flow rate for channel edge e. This represents the traffic on channel edge e during scheduling period t. t represents the maximum flow capacity of channel edge e, where e represents the channel edge number and t represents the scheduling period number. The maximum flow capacity is calculated using the Manning formula combined with the maximum allowable water depth of the channel, while the minimum safe flow rate is determined based on channel maintenance requirements.

[0054] The gate operation constraints take into account the mechanical adjustment limitations of the gate. The gate opening degree of each control node corresponds to a specific flow range, and the opening adjustment rate cannot exceed the mechanically permissible value to avoid water flow impact causing damage to the channel system. The expression is as follows: ; In the above formula, This represents the traffic on channel edge e during scheduling period t. This represents the traffic on channel edge e during scheduling period t-1. The maximum flow regulation rate represents the channel edge number e, where e represents the channel edge number and t represents the scheduling period number. The maximum flow regulation rate is determined by the gate type and mechanical performance.

[0055] The time-period connection constraint considers the channel's water storage capacity to ensure that changes in flow rate between adjacent time periods do not cause sudden changes in the channel's water level. The expression is as follows: ; ; In the above formula, This represents the water storage volume at the end of time period t along the channel edge e. This represents the water storage volume at the end of time period t-1 along channel e. Represents the length of the scheduling period. This represents the input flow of channel edge e during time period t. This represents the output flow of channel edge e during time period t. This represents the minimum water storage capacity at channel edge e. The maximum water storage capacity represents the channel edge number (e), where e represents the channel edge number and t represents the scheduling period number. The minimum and maximum water storage capacities are determined by the channel cross-sectional dimensions and the safe water level.

[0056] The model is constructed using a nonlinear programming framework. Since the water loss coefficient is a nonlinear function of the flow rate, the node flow balance constraints are also nonlinear. To improve solution efficiency, a piecewise linearization method can be used to transform the water loss coefficient function into a multi-segment linear function, converting the model into a mixed-integer linear programming model, which is easier to solve using a mature solver. The model's input data includes structural parameters of the lossy directed canal network topology, gross water demand load of each demand node at different time periods, canal flow capacity parameters, gate regulation rates, and canal storage constraints. All data must be standardized into matrix or vector form to ensure that the solver can recognize and utilize it.

[0057] S4. Based on the time-varying network flow optimization decision model, the independent network flow sub-problem is decomposed for each scheduling period and solved to obtain the optimized flow sequence of each channel edge in the lossy directed channel network topology graph for multiple future scheduling periods.

[0058] Specifically, the core of this step is to reduce the complexity of model solving by adopting a time-period decomposition strategy while ensuring the global optimization effect. This is achieved by decomposing the time-varying network flow problem across multiple time periods into independent sub-problems for each time period, and using an efficient solution algorithm to obtain the optimized flow sequence for each channel edge. The theoretical basis for this decomposition is that in irrigation scheduling in irrigation districts, the water flow transmission process in each scheduling period is mainly affected by the gross water demand and channel attribute constraints of that period. The correlation between time periods is mainly achieved through the connection of channel water storage. Therefore, by setting water storage connection conditions, the global problem can be decomposed into multiple single-time-period sub-problems, and the solution result of each sub-problem satisfies the intra-time period constraints and the inter-time period connection constraints.

[0059] The decomposition process first requires clarifying the boundary conditions of each sub-problem. For each scheduling period, the input data for the sub-problem includes the static structural parameters of the lossy directed canal network topology, the gross water demand load of each demand node in that period, the canal storage capacity of the previous period, canal flow capacity constraints, and gate regulation rate constraints. The canal storage capacity for the initial scheduling period is the initial storage capacity, obtained from measured data. The objective function for each sub-problem is to minimize the water supply volume for that period, expressed as follows: ; In the above formula, S represents the water supply volume during the scheduling period t, and S represents the set of water source nodes. This represents the output flow rate of water source node s within the scheduling period t.

[0060] The constraints of the subproblem include the node flow balance constraint, channel flow capacity constraint, gate operation constraint, and water storage capacity connection constraint for this time period. The expression for the water storage capacity connection constraint is as follows: ; ; In the above formula, This represents the water storage volume at the end of time period t along the channel edge e. This represents the water storage volume at the end of time period t-1 along channel e. Represents the length of the scheduling period. This represents the set of upstream channel edges corresponding to the inflow channel edge e. This represents the flow of the upstream channel edge e' during time period t. Representing the upstream channel edge e' in traffic The water conveyance loss coefficient is below. This represents the set of downstream channel edges corresponding to outflow channel edge e. This represents the flow rate of the downstream channel edge e'' during time period t. This represents the minimum water storage capacity at channel edge e. This represents the maximum water storage capacity of channel edge e, where e represents the channel edge number and t represents the scheduling period number.

[0061] The algorithm for solving subproblems needs to be selected based on the model type. If piecewise linearization is used to process the water loss coefficient function, the subproblem is a mixed-integer linear programming problem, which can be solved using the simplex method. If nonlinear constraints are retained, the interior-point method or an improved genetic algorithm can be used. Taking the mixed-integer linear programming problem as an example, the solution process is divided into three stages: preprocessing, iterative solution, and feasibility verification. In the preprocessing stage, the constraints of the subproblem need to be transformed into standard form, that is, the constraints need to be transformed into matrix equations and upper and lower bounds of variables. At the same time, the piecewise linear function of the water loss coefficient needs to be transformed into a combination of binary variables and linear constraints. The iterative solution phase employs a two-stage simplex method. The first stage seeks a feasible solution by introducing artificial variables to construct an auxiliary objective function. The goal is to minimize the sum of these artificial variables. If the auxiliary objective function value is 0, a feasible solution exists; otherwise, the constraints need adjustment. The second stage iteratively optimizes the objective function within the feasible region. A test number is calculated to determine if the current solution is optimal. If a negative test number exists, the variable with the smallest test number is selected as the input variable. A ratio test is used to determine the basic variables, and the basis matrix and solution vector are updated until all test numbers are non-negative, yielding the optimal solution. The feasibility verification phase verifies whether the optimal solution satisfies all constraints. Key checks include flow balance at demand nodes, whether channel flow is within its capacity, and whether water storage meets connection requirements. If constraints are violated, the number of segments in the piecewise linearization or constraint parameters need to be adjusted, and the solution recalculated.

[0062] After solving each subproblem, the optimized flow rate for all channel edges within that time period is obtained. The optimized flow rates for all time periods are then arranged chronologically to form the optimized flow rate sequence for each channel edge. During the solution process, iteration termination conditions must be set, including the difference in the objective function between two adjacent iterations being less than a set threshold or the number of iterations reaching a set upper limit, to ensure a balance between solution efficiency and accuracy. For complex canal networks with multiple water conveyance paths, the main path must be specified through constraints during the solution process to avoid increased water conveyance losses due to dispersed flow distribution.

[0063] Verification of the optimized flow sequence is a crucial step, requiring analysis from three dimensions: First, water balance verification, calculating the ratio of the total gross water demand load of all demand nodes to the total water supply from the source. This ratio should be close to the average water conveyance efficiency of the canal system, ensuring no significant water imbalance. Second, constraint fulfillment verification, checking whether the optimized flow rate at each canal is within its capacity, whether the gate regulation rate meets the limits, and whether the water storage is within a safe range. Third, practical feasibility verification, combining historical scheduling data to determine whether the optimized flow sequence conforms to the actual operating patterns of the irrigation district. If problems are found during verification, the constraints or solution parameters of the sub-problems need to be adjusted retrospectively until the optimized flow sequence meets all requirements. Finally, the optimized flow sequence is stored in a standardized data format for easy generation of subsequent gate control commands.

[0064] S5. Based on the optimized flow sequence and gate flow characteristics, generate an opening control command sequence for each gate in the time dimension, and execute the opening control command sequence to achieve intelligent irrigation of the irrigation area.

[0065] Specifically, this step transforms the abstract optimized flow rate into executable gate operation commands, establishes a mapping relationship between flow rate and opening degree based on gate flow characteristics, and achieves precise execution of the irrigation process through automatic control technology. The core components include gate flow characteristic modeling, opening degree calculation, control command generation, and execution feedback.

[0066] Modeling the gate flow characteristics is fundamental to achieving accurate flow-opening mapping. This requires obtaining the flow characteristic curve for each gate through field experiments, i.e., the functional relationship between flow rate, opening, and the difference in water head between the upstream and downstream sides of the gate. The experimental method employs a stepped opening method, selecting multiple typical opening values ​​covering the entire range from 0 to the maximum opening. After stabilizing at each opening for a period, the upstream and downstream water head difference is monitored using a water level sensor, and the actual flow rate is monitored using an electromagnetic flowmeter. Multiple sets of opening, water head difference, and flow rate data are recorded. The least squares method is used to fit the data. The fitting model is selected according to the gate type: a linear model is used for flat gates, and a nonlinear model is used for curved gates. The coefficient of determination must meet the set requirements to ensure the accuracy of the mapping relationship. The fitted flow characteristic function is stored in the gate control database, associated with the corresponding gate number and channel side number for easy retrieval. The linear model expression is as follows: ; In the above formula, Represents the flow rate through the gate. Represents the gate opening degree. This represents the difference in water head upstream and downstream of the gate. , , These are all fitting coefficients. The expression for the nonlinear model is as follows: ; In the above formula, Represents the flow rate through the gate. Represents the gate opening degree. This represents the difference in water head upstream and downstream of the gate. , , All are fitting coefficients.

[0067] The core of gate opening calculation is to inversely deduce the target opening based on optimized flow rate and head difference. For each scheduling period, for each gate's corresponding channel side, the optimized flow rate for that period is first extracted, and the head difference between the upstream and downstream of the gate is monitored in real time using a water level sensor. The optimized flow rate and head difference are then substituted into the gate's flow characteristic function to inversely solve for the target opening. For linear models, the target opening is directly solved through algebraic operations, as shown in the following formula: ; In the above formula, Represents the target opening degree. This represents optimizing traffic. This represents the difference in water head upstream and downstream of the gate. , , All are fitting coefficients. For nonlinear models, Newton's iteration method is used for solution. The initial values ​​of the iteration are estimated based on the optimized flow rate and the fitting coefficients. The iteration formula is the current opening value minus the difference between the flow characteristic function value at the current opening value and the optimized flow rate, then divided by the derivative of the flow characteristic function at the current opening value, until the deviation between the calculated flow rate and the optimized flow rate is less than a set threshold. The opening value obtained at this point is the target opening. The iteration formula is as follows: ; In the above formula, This represents the opening value in the (k+1)th iteration. This represents the opening value in the k-th iteration. The flow characteristic function at the opening value The function value below, Q represents optimized traffic. The flow characteristic function at the opening value The derivative value below.

[0068] The gate's operational constraints must be considered during the opening calculation process to avoid water flow impact and mechanical damage caused by sudden changes in opening. First, there is the opening limit constraint: the target opening must be between 0 and the maximum opening. If the calculated target opening exceeds the limit, the boundary value is taken and simultaneously fed back to the optimization model to fine-tune the channel flow for that period to ensure demand is met. Second, there is the regulation rate constraint: the change in opening between adjacent periods must be less than or equal to the maximum regulation rate. If this limit is exceeded, the target opening is adjusted to the opening value of the previous period plus or minus the maximum regulation rate, and the corresponding actual flow is recalculated. If the flow deviation exceeds a set threshold, the channel flow allocation for that period needs to be re-optimized. The maximum regulation rate is determined by the gate's mechanical performance.

[0069] The generation of control command sequences requires a standardized command format. Each command includes information such as command ID, gate number, execution time, target opening degree, regulation rate, and safety threshold. The execution time corresponds to the start time of the scheduling period, the regulation rate is the ratio of the opening degree change between adjacent periods to the length of the scheduling period, and the safety threshold includes the maximum allowable head difference. The command sequence is arranged chronologically, stored in a format recognizable by the programmable logic controller (PLC), and transmitted to the gate control system via industrial Ethernet.

[0070] The execution process employs a closed-loop automatic control mode. After receiving control commands, the programmable logic controller (PLC) drives the gate actuator (in this embodiment, the gate) to adjust its opening degree according to a set adjustment rate. An opening degree sensor provides real-time feedback on the actual opening degree, and a flow sensor provides real-time feedback on the actual flow rate. The actual value is compared with the target value, and a proportional-integral-derivative (PID) control algorithm is used for real-time correction. The parameters of the PID controller are determined through on-site debugging to ensure fast system response, small overshoot, and small steady-state error. When the actual flow rate is less than the optimized flow rate, the PID output is increased, driving the gate to increase its opening degree; when the actual flow rate is greater than the optimized flow rate, the PID output is decreased, driving the gate to decrease its opening degree, until the deviation between the actual flow rate and the optimized flow rate is less than a set threshold.

[0071] Emergency protection mechanisms are implemented during execution to ensure the safe operation of the canal system. These include: 1) Water level protection: When the canal water level exceeds the warning level, an alarm signal is automatically issued, and the gates are opened wider to discharge floodwater; when the water level is below the danger level, the gate opening is reduced to prevent the canal from drying out. 2) Fault protection: When a sensor or actuator malfunctions, the system automatically switches to manual control mode and sends a fault alarm to the monitoring center. 3) Extreme weather protection: Upon receiving an extreme weather warning, regular control commands are suspended, and a pre-set emergency dispatch plan is executed.

[0072] After irrigation is completed, data such as actual opening degree, actual flow rate, and execution time are stored in a historical database. This data serves as input for the next round of optimization decisions, updating model parameters such as the water conveyance loss coefficient function and gate flow characteristic curve. This continuously improves the accuracy of decision-making and execution, forming a closed-loop intelligent irrigation system of decision-making-execution-feedback-optimization. Simultaneously, an irrigation execution report is generated, including indicators such as water supply volume for each time period, actual water consumption at each demand node, and water resource utilization efficiency, providing data support for irrigation district management.

[0073] The aforementioned method for intelligent irrigation decision-making in irrigation districts constructs a canal network topology map that includes water conveyance loss coefficients. Based on this, dynamic crop water demand data is integrated to calculate the gross water demand load of each area, taking into account transmission losses. A decision model is then established with the goal of minimizing total water supply and dynamically reducing and balancing node flow using the aforementioned loss coefficients. A precise canal flow allocation sequence is obtained through time-period decomposition and network flow optimization. Finally, the sequence is converted into specific gate control commands and executed. This achieves systematic dynamic compensation and optimized allocation of water conveyance losses across the entire irrigation district, improving the accuracy of water resource spatiotemporal matching and overall utilization efficiency, and effectively overcoming the water supply-demand mismatch problem caused by neglecting network losses.

[0074] refer to Figure 2 In one optional embodiment, based on the spatial connectivity and engineering parameters of the irrigation district channels, a lossy directed canal network topology diagram characterizing the water flow transmission and distribution process is constructed, including the following steps: S11. Based on the geographic mapping data and engineering survey data of the irrigation area, extract the channel connection nodes and flow directions of all channels; based on the channel connection nodes and flow directions, define each water distribution point, confluence point and channel endpoint as a network node; based on the channel connection nodes and flow directions, define each actual channel segment as a directed path from the upstream node to the downstream node; construct the initial network structure according to the network nodes and directed paths.

[0075] Specifically, the geographic surveying data utilizes high-precision topographic surveying results, including 1:500 scale topographic vector maps and orthophoto maps of the irrigation area. Geometric information such as the centerline and boundary lines of the channels are extracted using the spatial analysis module of the geographic information system software. Engineering survey data encompasses design drawings and construction records of the canal system, used to confirm the actual connection relationships and water flow direction of the channels. The extraction of channel connection nodes is achieved by overlaying geographic surveying data and engineering surveying data. A line element intersection analysis algorithm is used to identify and extract the water-dividing points (i.e., nodes where one channel branches into multiple channels), confluence points (i.e., nodes where multiple channels converge into one channel), and the starting and ending points of the channels. Each extracted connection node must record its unique spatial coordinates (using the Gauss-Kruger projection coordinate system). The flow direction is determined based on the channel elevation data and the engineering design flow direction, verified through topographic slope analysis to ensure that the direction of the directed path is consistent with the actual flow direction of water from high potential energy to low potential energy. Each actual channel segment and its corresponding directed path need to be associated with its corresponding channel entity information, including channel name and level (main canal, branch canal, etc.). The initial network structure is stored in a graph data format, clarifying the relationship between network nodes and directed paths, forming a node-edge mapping table.

[0076] S12. Based on the channel's cross-sectional area, bottom slope, and Manning roughness coefficient, calculate the maximum safe flow capacity of the channel represented by each directed path using the Manning formula. Assign the maximum safe flow capacity as the upper bound constraint of the corresponding directed path's capacity. The expression for the Manning formula is: ; in, Indicates the maximum safe overcurrent capacity. This indicates the cross-sectional area of ​​the channel for water passage. This represents the Manning roughness coefficient. Indicates the hydraulic radius. This indicates the slope of the canal bottom.

[0077] Specifically, the cross-sectional area A is calculated using measured data of the channel cross-section. For regular cross-sections (such as rectangular or trapezoidal), it is calculated according to the corresponding geometric formula based on the cross-sectional dimensions. For irregular cross-sections, it is calculated using the coordinate data of the measured points on the cross-section through numerical integration. The channel bottom slope S is obtained by dividing the difference in elevation data between the channel's starting and ending points by the channel length. The elevation data is taken from elevation control points in the geographic survey data to ensure the accuracy of the slope calculation. The Manning roughness coefficient n is determined based on the channel's lining material, the smoothness of the inner wall, and the vegetation growth within the channel. It is determined by combining engineering survey records and relevant specifications, and the value needs to be corrected by referring to measured roughness data from similar irrigation areas during the selection process. The Manning formula expression used is: ,in The maximum safe flow capacity represents the maximum allowable flow rate of the channel under the premise of ensuring safe operation (without overflow, scouring, etc.); A represents the cross-sectional area of ​​the channel; n represents the Manning roughness coefficient, reflecting the resistance of the channel's inner wall to the water flow; R represents the hydraulic radius, which is the ratio of the cross-sectional area A to the wetted perimeter, the perimeter of the water flow in contact with the channel wall in the cross-section, calculated according to the corresponding geometric relationship based on the cross-sectional type; S represents the channel bottom slope, reflecting the degree of inclination of the channel bottom. During the calculation process, the channel segment corresponding to each directed path is calculated separately, and the results are then displayed. As one of the core attributes of this directed path, it is assigned a capacity upper bound constraint to constrain the maximum flow rate of this channel segment in subsequent optimization decisions.

[0078] By analyzing geospatial data and water conservancy project layout data to extract network nodes, and using a fluid dynamics resistance model to calculate the upper limit constraint of capacity, the network structure model can truly reflect the physical topology and hydraulic characteristics of the irrigation area water conveyance system, providing a reliable engineering basis for optimization decisions and avoiding systematic biases caused by simplified modeling.

[0079] S13. Based on channel leakage test data or historical calibration data, assign a flow-related water loss coefficient to each directed path; wherein, the water loss coefficient represents the proportion of the flow that is lost in the transmission process and cannot reach the destination from the starting point of the corresponding directed path.

[0080] Specifically, the water conveyance loss coefficient represents the proportion of the flow that is lost during the transmission process and cannot reach the destination from the starting point of the corresponding directed path. Channel leakage test data is obtained through in-situ testing using a constant flow method in a closed channel section. This involves setting control gates at both ends of a channel section, closing the downstream gate, and injecting a constant flow into the upstream section. After the water level stabilizes, the leakage flow at the downstream gate is measured. A model relating leakage flow to input flow is established using leakage test data under different injection flow rates. Historical calibration data is taken from the irrigation district's annual scheduling and operation records, including input and output flow monitoring data for the channel section. After removing outliers through statistical analysis, the water conveyance loss ratio under different flow conditions is calculated and used as calibration data. The correlation between the water conveyance loss coefficient and flow rate is established through a fitting function. Based on the test or calibration data, the least squares method is used to fit the functional relationship η of the water conveyance loss coefficient η with respect to flow rate Q, η=f(Q). Common fitting models include linear models and exponential models; the specific model type is determined based on the goodness of fit of the data. For channel sections lacking actual measured data, an analogy method can be used, referencing the water conveyance loss coefficient function of channel sections of the same type, scale, and soil conditions for assignment, and reserving an interface for subsequent calibration and correction. During the assignment process, each directed path is associated with a corresponding η=f(Q) function to ensure that the water conveyance loss coefficient changes dynamically with the flow rate, accurately reflecting the transmission loss characteristics under different flow conditions.

[0081] By establishing a mapping relationship between water conveyance loss and flow rate, the water conveyance loss parameter is dynamically adjusted according to the flow rate conditions, overcoming the limitations of the traditional fixed loss coefficient and significantly improving the model's accuracy in representing actual hydraulic behavior.

[0082] S14. Based on the initial network structure, the water intake is defined as the water source node, and corresponding demand nodes are created for all irrigation areas that need irrigation. Each demand node is connected to the channel connection node leading to the water source node through a directed path to obtain the directed canal network topology.

[0083] Specifically, the definition of water source nodes needs to be combined with engineering survey data to clarify the location of actual water intakes such as reservoir outlets and pump station outlets. Their corresponding spatial coordinates should be matched with nodes in the initial network structure. If no existing node is matched, a new node should be added at the corresponding location and defined as a water source node, labeling its water source type (e.g., surface water source, groundwater source) and other attributes. Demand nodes are created for each independent irrigation area, with the irrigation area's inlet as the core location. Each demand node needs to be associated with its corresponding irrigation area information, including irrigation area and crop type. The connection between demand nodes and canal connection nodes must follow the principle of water flow accessibility. Using the network analysis function of the geographic information system, the nearest canal connection node to the demand node that leads to the water source node is found, and a directed path is created from this canal connection node to the demand node. This directed path represents the connection section between the irrigation area's inlet and the main canal system, and its attribute parameters (e.g., flow capacity, water loss coefficient) are determined based on the engineering design parameters of the inlet. By adding nodes and connecting edges as described above, the network connection between the water source and the demand side is improved based on the initial network structure, forming a complete directed canal network topology.

[0084] S15. Based on the directed canal network topology, water conveyance loss coefficient, and capacity upper bound constraint, construct a lossy directed canal network topology.

[0085] Specifically, the upper bound constraints of the capacity of each directed path calculated in S12 (maximum safe flow capacity) and the water loss coefficient function assigned to each directed path in S13 are batch-associated as attribute parameters to the directed paths corresponding to the directed canal network topology, forming a complete network model containing node information, edge information, and edge attribute parameters. Simultaneously, to ensure the integrity and accuracy of the network model, a lower bound of the flow capacity (minimum safe flow) for each directed path can be added. This value is determined according to the canal maintenance requirements and is used to avoid problems such as siltation and vegetation growth caused by insufficient flow. Furthermore, the integrity of the constructed lossy directed canal network topology is verified, checking for anomalies such as isolated nodes (excluding water source nodes and demand nodes) and missing attribute edges. Network connectivity analysis verifies the path reachability from each water source node to each demand node. Attribute consistency verification is also performed to ensure that the attribute parameters (water loss coefficient, upper and lower bounds of flow capacity) of each directed path match the corresponding canal entity data. After verification, the lossy directed canal network topology diagram is stored in a standardized network data format to facilitate subsequent water demand calculation and optimization decision-making model calls. This format needs to support fast node and edge retrieval, efficient reading of attribute parameters, and fast analysis of network paths.

[0086] In one optional embodiment, the gross water demand load of each demand node in each time period is calculated based on the net irrigation water demand and the water conveyance loss coefficient in the lossy directed canal network topology diagram, including the following steps: S21. Based on the topology of the lossy directed canal network, obtain the water conveyance loss coefficient of the unique directed path connecting the demand nodes.

[0087] Specifically, in the lossy directed canal network topology, each demand node is associated with a canal connection node via a dedicated directed path created in step S14. This directed path is the only path connecting the demand node to the main canal system, ensuring the uniqueness of the water flow transmission path. The acquisition process is achieved through node-edge association retrieval in the topology graph: first, the node number and attribute information of the target demand node are located; based on the adjacency list structure of the topology graph, the inflow edge set of the demand node is queried. Since the demand node is an irrigation terminal node, its inflow edge set only contains the dedicated directed path created in S14; the corresponding water loss coefficient is extracted from the attribute parameters of this dedicated directed path. This coefficient represents the dynamic loss parameter of the directed path under the current scheduling conditions, and its flow-related characteristics have already been assigned in step S13. If multiple inflow edges are found for a demand node during the retrieval process, the topology construction process in step S14 needs to be traced back to verify the connection relationship between the demand node and the channel connection node, ensuring the uniqueness of the dedicated directed path and avoiding deviations in obtaining the water conveyance loss coefficient caused by multiple paths.

[0088] S22. Based on the water conveyance loss coefficient and the local water utilization coefficient from the canal to the irrigation area, the net irrigation water demand is converted into the gross water demand load that needs to be obtained from the canal connection nodes; wherein, the calculation formula for the gross water demand load is: ; in, Indicates the demand node During the scheduling period The gross water demand load, Indicates the demand node During the scheduling period Net irrigation water requirement within the area Indicates the duration of the scheduling period; This represents the local water use coefficient from the canal connection node to the irrigation area outlet; the local water use coefficient characterizes the water conveyance loss, water distribution loss, and management loss in the process from the canal connection node to the irrigation area. Indicates the connection requirement node The water loss coefficient of the directed path.

[0089] Specifically, the local water use coefficient is a key parameter characterizing the overall water resource utilization efficiency from the canal connection node to the irrigation district outlet. It encompasses water transport losses (such as leakage and evaporation losses in the connecting canal), water distribution losses (such as local head losses in the water allocation process), and management losses (such as water waste caused by scheduling delays) during the transmission process. Its value is determined based on field tests and historical data calibration. The ratio of the output flow rate at the canal connection node to the actual inflow rate at the irrigation district outlet is calculated as the local water use coefficient. The calibration process needs to cover different flow conditions to ensure the applicability of the coefficients. The formula for calculating gross water demand is: ,in This represents the gross water demand load of demand node v during scheduling period k, which is the minimum flow that the channel connection node needs to output to this demand node to ensure that the net irrigation demand of the irrigation district can still be met after deducting various losses. This represents the net irrigation water demand of demand node v during scheduling period k, calculated using the crop water balance principle mentioned above. The duration of the scheduling period should be consistent with the calculation period of net irrigation water demand to ensure the uniformity of flow units; This represents the local water utilization coefficient from the canal connection node to the irrigation area outlet; This represents the water transport loss coefficient of the directed path connecting the demand node v, i.e., the proportion of lost flow to input flow during the transmission process of this directed path. This represents the flow transmission efficiency of the directed path. During the calculation, it's crucial to ensure unit consistency for all parameters: net irrigation water demand is expressed in volume units, and scheduling period duration in time units. The ratio of these two units is converted to flow units, and then subtracted by multiplying the local water use coefficient by the directed path transmission efficiency to obtain the gross water demand load. Each demand node must independently calculate its gross water demand load for each scheduling period. After calculation, this load is used as a core parameter and associated with the corresponding demand node and scheduling period, providing input data for the subsequent optimization decision model and ensuring the model accurately matches the actual water supply needs of each demand node.

[0090] By introducing a local water use coefficient and extrapolating the gross water demand load from the end-user demand, a systematic compensation for the loss throughout the entire process from the water conveyance network to the field water use unit is achieved, ensuring that the effective water volume reaching the end-user can meet the net irrigation needs of crops and significantly improving the irrigation guarantee rate.

[0091] In one optional embodiment, based on the structure of the lossy directed canal network topology, the canal capacity attributes, and the gross water demand load of each demand node at each time period, a time-varying network flow optimization decision model is established with the objective of minimizing the total water supply, including the following steps: S31. Discretize the scheduling period into K consecutive time periods, and define the decision variable as the flow of each directed path in the directed channel network topology graph with loss for each scheduling time period k.

[0092] Specifically, the determination of the scheduling cycle needs to be based on the actual irrigation needs of the irrigation district and the crop growth cycle. A complete scheduling cycle can be selected as an irrigation season or a month to ensure coverage of the crop's critical water-demand stages. The discrete process needs to determine the time period length according to the scheduling accuracy requirements. For short-term, precise scheduling, a one-day period can be selected; for medium- to long-term planning and scheduling, a three- or seven-day period can be selected to ensure... The scheduling cycle is fully covered by a continuous time period, meaning the total duration of the scheduling cycle is [missing information]. The definition of decision variables must be unique and identifiable, and a two-dimensional identifier system is used for representation: Let the set of all directed paths in the network be... For any directed path Define decision variables Let be the flow rate of directed path e during scheduling period k. This decision variable directly reflects the water flow distribution status of the canal network in each period and is the core carrier for subsequent objective function optimization and constraint construction. Its value needs to be determined through optimization and must meet the constraints of canal engineering characteristics and irrigation demand.

[0093] S32. Based on the structure of the lossy directed canal network topology, the channel capacity attributes, and the gross water demand load of each demand node in each time period, construct an objective function with the goal of minimizing the total flow out of the water source node in all time periods.

[0094] Specifically, the core of constructing the objective function is to balance efficient water resource utilization with guaranteed irrigation demand, achieving an optimal balance between water resource consumption and irrigation benefits by minimizing the total water supply. First, the set of water source nodes S (defined in step S14) is defined, and the total water supply is the sum of the outflow from all water source nodes across all scheduling periods. Combining the relationship between decision variables and the flow of water source nodes—that is, the outflow from a water source node equals the sum of the flows of all directed paths originating from that source node—the objective function is constructed as follows: ; In the above formula, Z represents the total outflow from the water source node across all time periods (i.e., the total water supply), K represents the total number of time periods after the scheduling cycle is discretized, s represents the water source node, and S represents the set of water source nodes. This represents a directed path originating from the water source node s. Let represent the set of all directed paths originating from the water source node s. Represents the directed path within scheduling period k. The objective function is to optimize the flow distribution of each directed path at different time periods, thereby minimizing the total water supply and reducing water waste and pumping energy consumption while meeting all irrigation needs and engineering constraints.

[0095] S33. In the topology of a directed channel network with losses, construct a node flow balance constraint for each network node except for the water source node, such that for any node i in any scheduling period k, the net value of all directed path flows into node i after deducting losses is equal to the sum of all directed path flows out of node i; where the expression for the node flow balance constraint is: ; in, This represents a directed path ending at node i. Let represent the set of all directed paths pointing to node i. Represents a directed path The water conveyance loss coefficient, This represents a directed path during scheduling period k. Traffic; This represents a directed path starting from node i. Let represent the set of all directed paths originating from node i. This represents a directed path during scheduling period k. Traffic; This represents the set of all channel connection nodes.

[0096] Specifically, the core purpose of this constraint is to ensure the mass conservation of water flow at the nodes of the canal network, accurately characterize the impact of water conveyance losses on the flow balance at the nodes, and avoid flow distribution deviations caused by neglecting losses in traditional models. In the expression for the node flow balance constraint, This represents a directed path ending at node i. Let represent the set of all directed paths pointing to node i. Represents a directed path The water conveyance loss coefficient, assigned by step S13, characterizes the directed path. The proportion of data lost during transmission relative to the input data. This refers to the traffic transmission efficiency of the directed path; This represents a directed path during scheduling period k. Traffic; This represents a directed path starting from node i. Let represent the set of all directed paths originating from node i. This represents a directed path during scheduling period k. Traffic; This represents the set of all channel connection nodes, defined in step S11, encompassing water distribution points, confluence points, and channel endpoints; k represents the scheduling period number. During constraint construction, constraints are assigned to each channel connection node for each scheduling period to ensure that all nodes meet the flow balance requirements, providing a basic framework for subsequent flow optimization and allocation.

[0097] S34. Construct demand satisfaction constraints for each demand node. Demand satisfaction constraints are used to require that the flow into the demand node in each scheduling period k is not less than the corresponding gross water demand load.

[0098] Specifically, the demand node is the terminal point for irrigation water. This constraint is the core prerequisite for ensuring irrigation demand, ensuring that the water demand of the irrigation area corresponding to each demand node can be met at all times. The inflow flow of the demand node is the net value of the dedicated directed path connecting the demand node (created in step S14) after deducting water conveyance losses, combined with the gross water demand load calculated in step S22. The constraint expression for fulfilling the requirements is as follows: ; In the above formula, This represents the water conveyance loss coefficient of the dedicated directed path connecting the demand node v. This represents the traffic on the dedicated directed path within the scheduling period k. This refers to the effective flow of traffic flowing into demand node v; This represents the gross water demand load of demand node v during scheduling period k; This represents the set of all demand nodes; k represents the scheduling period number. The constraint uses "not less than" instead of "equal to" to handle sudden fluctuations in crop water demand, reserving a certain flow redundancy to ensure reliable fulfillment of irrigation needs. During the construction process, the gross water demand load of each demand node must be associated with its corresponding dedicated directed path to ensure accurate constraint matching.

[0099] S35. Construct flow capacity constraints for each directed path. Flow capacity constraints are used to require that the flow of a directed path in each scheduling period k is between zero and the corresponding upper capacity constraint.

[0100] Specifically, this constraint is constructed based on the physical characteristics of the canal system to prevent safety accidents such as overflow and scouring caused by excessive flow, while also limiting the flow rate to a non-negative value (water cannot flow in reverse). Capacity upper bound constraint The maximum safe flow capacity of the channel segment corresponding to the directed path is calculated using the Manning formula in step S12. The flow capacity constraint expression is as follows: ; In the above formula, This represents the flow of the directed path e within the scheduling period k; Let represent the upper bound constraint of the capacity of directed path 'e'; E represents the set of all directed paths; and k represents the scheduling period number. The lower bound in the constraint is set to zero because water flows from upstream to downstream along directed paths in the canal system, and there is no reverse flow, which conforms to the actual water flow transmission pattern. During the construction process, the upper bound constraint of the capacity of each directed path is precisely associated with the decision variables to ensure that the flow of each directed path is always within the safe operating range.

[0101] S36. Construct water source supply capacity constraints. Water source supply capacity constraints are used to require that the total flow from the water source node in each scheduling period k does not exceed the maximum water supply capacity of the water source in scheduling period k.

[0102] Specifically, the water supply capacity of a water source is limited by factors such as the type of water source and the conditions of engineering facilities. For example, the water source of a reservoir is limited by its capacity and the capacity of its sluice gates, while the water source of a pumping station is limited by the rated flow rate of its pump units. This constraint ensures that the water supply from the source remains within its carrying capacity, preventing excessive water supply that could lead to water source depletion or damage to the project. To determine the maximum water supply capacity of water source node s during scheduling period k, the constraint expression for water supply capacity is as follows: ; In the above formula, This represents a directed path originating from the water source node s. Let represent the set of all directed paths originating from the water source node s. Represents the directed path within scheduling period k. The sum of the two is the total flow rate from water source node s during scheduling period k; S represents the maximum water supply capacity of water source node s during scheduling period k; S represents the set of water source nodes; k represents the scheduling period number. The value needs to be determined by combining the water source type and real-time monitoring data. For example, the water source for a reservoir can be dynamically adjusted based on data such as the current reservoir capacity and water inflow forecast for the basin, while the water source for a pumping station can be determined based on the operating status of the pump units. During the construction process, the maximum water supply capacity of the water source for each time period is updated in real time to ensure the timeliness and feasibility of the constraints.

[0103] S37. Based on decision variables and objective functions, and combined with node flow balance constraints, demand satisfaction constraints, flow capacity constraints, and water source supply capacity constraints, a time-varying network flow optimization decision model is constructed.

[0104] Specifically, the core of model construction is to integrate the aforementioned decision variables, objective function, and various constraints into a unified mathematical programming framework, clarifying the relationships between each element. First, the correspondence between decision variables and constraints is clarified: Decision variables... It is necessary to simultaneously satisfy the node flow balance constraint, demand satisfaction constraint, flow capacity constraint, and water source supply capacity constraint; the objective function minimizes the total water source supply by optimizing the values ​​of decision variables.

[0105] Since the node flow balance constraints include a flow-related water loss coefficient α, and α is a function of flow (determined by step S13), the constructed time-varying network flow optimization decision model is a nonlinear programming model. To improve the model's solution efficiency, a piecewise linearization method can be used to process the water loss coefficient function α=f(q), transforming the nonlinear function into a multi-segment linear function, and thus converting the model into a mixed-integer linear programming model, which can be solved using a mature mixed-integer linear programming solver. The model's input data needs to be standardized, including: structural parameters of the lossy directed canal network topology (node ​​set, edge set, and relationships), upper capacity constraints for each directed path, water loss coefficient functions for each directed path, gross water demand load for each demand node at each time period, and maximum water supply capacity for each water source node at each time period. All data should be uniformly converted into matrix or vector form to ensure that the solver can efficiently identify and call upon it. After the model is built, an integrity check is performed to examine the logical consistency of the decision variables, objective function, and constraints, avoiding constraint conflicts or the disconnect between the objective function and constraints, and ensuring that the model can accurately depict the actual needs and engineering constraints of irrigation scheduling in the irrigation district.

[0106] In one optional embodiment, based on a time-varying network flow optimization decision model, the independent network flow sub-problem for each scheduling period is decomposed and solved to obtain the optimized flow sequence for each channel edge in the lossy directed channel network topology graph over multiple future scheduling periods, including the following steps: S41. Decompose the time-varying network flow optimization decision model into K independent single-period network flow sub-problems according to the scheduling period; wherein, the input of each single-period network flow sub-problem is the lossy directed channel network topology and the gross water demand load of each demand node in the current period; the output of each single-period network flow sub-problem is the flow allocation value of each directed path in the lossy directed channel network topology in the corresponding period.

[0107] Specifically, the core basis of the decomposition is the time-independent characteristic of time-varying network flows. That is, the water flow transmission and allocation in each scheduling period are mainly constrained by the current operating conditions (gross water demand and water source supply capacity). Inter-period connections are made through channel water storage, but this does not affect the flow optimization logic within a single period. This decomposition strategy can significantly reduce the solution complexity of multi-period coupled models. The input parameters for each single-period network flow sub-problem include: static structural parameters of the lossy directed channel network topology (node ​​set, edge set, and association relationships), upper capacity constraints for each directed path in the current period, water conveyance loss coefficients for each directed path in the current period, gross water demand load for each demand node in the current period, and maximum water supply capacity for each water source node in the current period. The output of each single-period network flow sub-problem is the flow allocation value for all directed paths within that period. The output must satisfy the current node flow balance constraints, demand satisfaction constraints, flow capacity constraints, and water source supply capacity constraints. During the decomposition process, a unique time period identifier must be assigned to each sub-problem to ensure accurate matching of input parameters with the time period. The total number of time periods after discretizing the scheduling period is determined by step S31. Value = (Total scheduling cycle duration) / (Duration of a single scheduling period) ); Duration of a single scheduling period It needs to be set according to the scheduling accuracy requirements. For short-term precise scheduling, 1 day can be used, while for medium- and long-term planned scheduling, 3 days or 7 days can be used.

[0108] S42. For the current scheduling period k to be determined, the gross water demand load of each demand node is transformed into a lower bound constraint of the flow of a directed path from the network node corresponding to the demand node to a sink point.

[0109] Specifically, the standard network flow algorithm uses "flow collection at a sink point" as its core logic. It constructs a closed-loop network by adding new sink points to achieve the standardized transformation of demand constraints. The specific transformation process is as follows: A new global sink point t is added, which is used for each demand node. Create a directed path from v to t. The lower bound constraint of the flow of this directed path is defined as the gross water demand load of the current time period v. That is, the constraint formula is: ; in, For scheduling period Inner directed path (Demand Node) Pointing to the confluence point ) traffic; For demand nodes During the scheduling period The gross water demand load is calculated by step S22 and represents the demand node. The minimum effective flow rate that needs to be guaranteed; the upper bound of the flow rate in the constraint is set to infinity (or a reasonable value much greater than the maximum gross water demand load, such as 10 times the maximum gross water demand load of all demand nodes), to avoid artificially limiting the upper limit of demand satisfaction. The physical meaning of this transformation is: demand nodes The effective incoming flow must at least meet the following requirements. And all of them need to be transported to the junction point. This transforms "water demand guarantee at demand nodes" into "directed path" "Flow lower bound constraint", adapting to the constraint expression form of the standard flow algorithm. Newly added sink point. With directed path It is only used for solving subproblems and does not change the physical structure of the original directed channel network with loss.

[0110] S43. Convert the water conveyance loss coefficient of each directed path in the lossy directed canal network topology into a gain factor, and combine the gain factors to construct a standard flow network with edge gain attributes that is equivalent to the lossy directed canal network topology.

[0111] Specifically, due to water loss in the original network, the flow rate at the inflow edge is not equal to the flow rate at the outflow edge, making it impossible to directly apply standard network flow algorithms. Therefore, an equivalent transformation of the loss characteristics is achieved through a gain factor. The core transformation formula for the gain factor is: ; in, For directed paths The gain factor characterizes the directed path. The efficiency of traffic transmission, i.e., the flow into the directed path The proportion of traffic that actually reaches downstream nodes after transmission (the traffic received by the downstream node = the inflow traffic × the outflow traffic). ); For directed paths The water conveyance loss coefficient, assigned by step S13, characterizes the directed path. The proportion of lost traffic to input traffic during transmission (lost traffic = inflow traffic × ... Its value range is [0,1). The closer the value is to 1, the greater the water loss during transport. This indicates that there was no water loss during transport.

[0112] The rules for constructing a standard flow network based on the gain factor are as follows: (1) All nodes in the original network (including water source nodes, channel connection nodes, and demand nodes) and newly added confluence points are retained. (2) Retain all directed paths in the original network, and set each directed path as follows: The attribute is from "water loss coefficient" "Updated to "gain factor" “;(3)New directed paths (Demand Node) Pointing to the confluence point The gain factor of the edge is set to 1 (no transmission loss) because this edge is only a virtual edge for constraint transformation and does not correspond to the actual water conveyance process. In the constructed standard flow network, the outflow of any directed path is equal to the product of the inflow and the gain factor, realizing the equivalent mapping from the lossy network to the standard gain network, ensuring that the solution results of subsequent algorithms can be reversed to restore the actual flow of the original network.

[0113] S44. On the standard flow network, run the maximum flow algorithm with flow lower bound constraints and gain attributes to solve the single-period network flow subproblem and find feasible flows that satisfy the flow lower bound constraints of all demand nodes. If a feasible flow exists, find the feasible flow that minimizes the outflow of the water source nodes as the optimal flow allocation scheme for the current period. If no feasible flow satisfies all flow lower bound constraints, identify the bottleneck demand nodes in the standard flow network by analyzing the maximum flow minimum cut theorem, reduce the gross water demand load of a preset percentage of bottleneck demand nodes according to the preset irrigation priority rules, and repeat step S44 until the optimal flow allocation scheme for the current period is obtained.

[0114] By converting water loss parameters into gain factors to construct a standard flow network model and using a maximum flow optimization algorithm with gain attributes to solve it, the complex nonlinear multi-period coupling problem is transformed into a standard network flow problem that can be solved efficiently, which significantly reduces computational complexity and makes real-time optimization scheduling of large-scale irrigation areas possible.

[0115] Specifically, all water source nodes are taken as originating and confluence points. To find the destination point, while satisfying the upper and lower bound constraints of the directed path flow and the gain transmission characteristics, we need to find the path from the set of originating points to the destination point. The maximum flow. If the solved maximum flow satisfies all newly added directed paths. lower bound constraint on flow (i.e.) For all If the condition is met, then the flow distribution at this time is a feasible flow. Further, the flow distribution that minimizes the total outflow of all water source nodes is selected from the feasible flow set, which is the optimized flow allocation scheme for the current period. The selection logic can be achieved by introducing objective function weights (the weight of water source outflow is set to 1, and the weights of other flows are set to 0) into the maximum flow algorithm to ensure that the water supply is minimized while meeting the demand.

[0116] If no feasible flow exists, the specific optimization process is as follows: (1) Based on the maximum flow minimum cut theorem, identify the minimum cut set in the standard flow network. The demand nodes in the "unreachable area" separated by the cut set are the bottleneck demand nodes (the demand of such nodes cannot be guaranteed by the existing network capacity); (2) According to the preset irrigation priority rules (such as food crops taking priority over cash crops, critical growth period taking priority over non-critical growth period, large irrigation area taking priority over small irrigation area, etc.), calculate the gross water demand load of the bottleneck demand nodes. (2) Reduce by proportion (the preset percentage can be 5%-20%, which can be adjusted according to the irrigation district management requirements); (3) Update the directed path corresponding to the bottleneck demand node. To address the lower bound constraint on flow, the maximum flow algorithm in step S44 is re-executed until a feasible flow that satisfies all adjusted lower bound constraints is obtained, and the optimal flow allocation scheme for the current time period is determined. During load reduction, the nodes being reduced, the reduction ratio, and the reduction amount must be recorded as important content in the irrigation scheduling report.

[0117] By employing a bottleneck diagnosis mechanism based on the maximum flow minimum cut theorem and a load reduction strategy based on irrigation priority, the scheduling decision-making problem under limited water supply capacity is effectively solved, maximizing irrigation benefits under resource constraints and enhancing the robustness and adaptability of the system.

[0118] S45. Perform steps S42 to S44 sequentially for all K scheduling periods to obtain the optimized traffic allocation scheme for each scheduling period; based on the optimized traffic allocation scheme for all scheduling periods, generate the optimized traffic sequence for each channel edge.

[0119] Specifically, the optimized flow allocation scheme of the previous period provides the basis for calculating the current channel water storage, and provides boundary conditions (such as channel water storage connection constraints) for solving the sub-problems of the next period, ensuring the continuity of water flow between periods. After the solution is completed, for each directed path... According to the scheduling time period ( to Extract the optimized traffic allocation values ​​for each time period and construct the optimized traffic sequence for the directed path. .

[0120] After optimizing the traffic sequence generation, cross-time period consistency verification is performed. The core verification metrics and parameters are explained below: (1) Gate regulation rate constraint: ; Explanation of formula parameters: For directed paths During the scheduling period Optimize traffic, For directed paths During the scheduling period The optimized traffic flow, and the absolute value of the difference between the two values, represents the magnitude of traffic change in adjacent time periods; For directed paths The maximum allowable flow rate regulation rate of the corresponding gate is determined by the gate's mechanical performance to avoid gate impact damage caused by sudden flow changes. If this constraint is not met, the optimized flow rate for the next time period (within the feasible region) is fine-tuned to ensure the physical feasibility of gate operation.

[0121] (2) Safety constraints on channel water storage capacity: ; Explanation of formula parameters: For directed paths The corresponding channel segment during the scheduling period The water storage capacity is determined by the inflow rate during that period. Outflow Time period duration calculation results in ( ); This is the minimum safe water storage capacity for a section of the canal; water levels below this value are prone to siltation and vegetation growth. This is the maximum safe water storage capacity for a channel section; exceeding this value increases the risk of overflow. Both are determined by the channel cross-sectional dimensions and engineering design standards. If the water storage capacity exceeds the safe range, the flow distribution plan for the corresponding time period will be adjusted retrospectively.

[0122] (3) Global target rationality verification: Summarize the total water supply of all time periods. This verifies whether it is the minimum value within the feasible region, ensuring that the optimization result meets the core objective of "minimizing the total water supply". In this formula, Z represents the total flow rate (i.e., the total water supply) from the water source nodes across all time periods, K represents the total number of time periods after the scheduling cycle is discretized, s represents the water source nodes, and S represents the set of water source nodes. This represents a directed path originating from the water source node s. Let represent the set of all directed paths originating from the water source node s. Represents the directed path within scheduling period k. The traffic flow is calculated. After verification, the optimized traffic sequences of all directed paths are stored in a standardized data format (such as a matrix format, where rows represent directed paths and columns represent time periods) for easy use in subsequent gate opening calculations.

[0123] The aforementioned method for intelligent irrigation decision-making in irrigation districts constructs a canal network topology map that includes a water conveyance loss coefficient. Based on this, dynamic crop water demand data is integrated to calculate the gross water demand load of each area, taking into account transmission losses. A decision model is then established with the goal of minimizing total water supply and dynamically reducing and balancing node flow using the aforementioned loss coefficient. A precise canal flow allocation sequence is obtained through time-period decomposition and network flow optimization. Finally, the sequence is converted into specific gate control commands and executed. This achieves systematic dynamic compensation and optimized allocation of water conveyance losses across the entire irrigation district, improving the accuracy of water resource spatiotemporal matching and overall utilization efficiency, and effectively overcoming the water supply-demand mismatch problem caused by neglecting network losses.

[0124] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0125] Based on the same inventive concept, this application also provides a system for implementing the above-described method for intelligent irrigation decision-making in irrigation districts. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations of one or more system embodiments for intelligent irrigation decision-making in irrigation districts provided below can be found in the limitations of the method for intelligent irrigation decision-making in irrigation districts described above, and will not be repeated here.

[0126] In one exemplary embodiment, such as Figure 3 As shown, a system 30 for intelligent irrigation decision-making in irrigation districts is provided to implement the methods in the above-described method embodiments. The system includes: The canal system topology modeling module 31 is used to construct a lossy directed canal system network topology diagram that represents the water flow transmission and distribution process based on the spatial connection relationship and engineering parameters of the irrigation area.

[0127] The irrigation demand quantification module 32 is used to calculate the net irrigation water demand of each demand node in multiple scheduling periods in the future, based on the crop type, crop planting area, crop growth stage and forecast meteorological data corresponding to each irrigation area; and to calculate the gross water demand load of each demand node in each period based on the net irrigation water demand and the water conveyance loss coefficient in the lossy directed canal network topology diagram.

[0128] The optimization decision modeling module 33 is used to establish a time-varying network flow optimization decision model with the objective of minimizing the total water supply, based on the structure of the lossy directed canal network topology, the channel capacity attributes, and the gross water demand load of each demand node in each time period. The time-varying network flow optimization decision model includes node flow balance constraints with embedded water conveyance loss coefficients. The node flow balance constraints dynamically reduce the flow of each edge flowing into any network node according to the water conveyance loss coefficient corresponding to each edge, so that the reduced net inflow is equal to the total outflow from any network node.

[0129] The flow optimization solution module 34 is used to solve the independent network flow sub-problem for each scheduling period based on the time-varying network flow optimization decision model, and obtain the optimized flow sequence of each channel edge in the lossy directed channel network topology graph in multiple future scheduling periods.

[0130] The intelligent irrigation execution module 35 is used to generate an opening control command sequence for each gate in the time dimension based on the optimized flow sequence and gate flow characteristics, and execute the opening control command sequence to realize intelligent irrigation of the irrigation area.

[0131] Embodiments of this application also provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the aforementioned method embodiments.

[0132] Embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0133] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0134] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for intelligent irrigation decision-making in irrigation districts, characterized in that, The method includes the following steps: Based on the spatial connection relationship and engineering parameters of the irrigation district water conveyance system, a lossy directed network structure model is constructed to characterize the dynamic water conveyance loss generated by water flow in a directed path; the network nodes in the lossy directed network structure model include water source nodes and demand nodes. Based on the water demand characteristics of each water-using unit, the net irrigation demand in the future scheduling cycle is determined, and combined with the water conveyance loss parameters in the lossy directed network structure model, the gross water demand load at each demand node is calculated from the end demand. With minimizing the total water supply as the global optimization objective, a dynamic water allocation model is established based on the topological connections of the lossy directed network structure model, the capacity upper bound constraints of each directed path, and the gross water demand load of each demand node. This model incorporates spatiotemporal coupling characteristics across multiple time periods. The dynamic water allocation model embeds node flow balance constraints based on the water conveyance loss parameters. These constraints require that the net inflow of each directed path into a node, after reduction by its corresponding water conveyance loss parameters, remains conserved with the total outflow from that node. The dynamic water allocation model is solved to generate the optimized flow allocation sequence of each directed path in each time period within the scheduling cycle; Based on the optimized flow distribution sequence, a control command sequence for regulating the actuator is generated.

2. The method according to claim 1, characterized in that, The construction of the lossy directed network structure model representing the dynamic water transport loss generated by water flow in a directed path includes: By analyzing the geospatial data of the irrigation area and the layout data of water conservancy projects, all diversion points, confluence points and water transmission terminals are extracted as network nodes. The water intake is defined as the water source node, corresponding demand nodes are created for each water-using unit, and the physical water transmission channels connecting adjacent nodes are defined as directed paths with a single flow direction attribute. For each directed path, based on the geometric parameters of the cross-section of the corresponding physical water conveyance channel, the physical roughness of the inner wall, and the gravity gradient characteristics along the path, the maximum safe transport flux of the directed path under steady-state flow is calculated using a fluid dynamics resistance model, and this maximum safe transport flux is set as the upper limit constraint of the capacity of the directed path.

3. The method according to claim 1, characterized in that, The water conveyance loss parameters of each directed path in the lossy directed network structure model are determined in the following way: Obtain permeability test data or historical operation monitoring data for each water conveyance channel and establish a mapping relationship between water conveyance loss and flow rate; Based on the mapping relationship, the water loss parameter is configured for each directed path. The water loss parameter represents the proportion of flow loss during the process of fluid being transported from the starting point to the end point of the directed path, and is dynamically adjusted according to the mapping relationship as the flow rate changes through the directed path.

4. The method according to claim 1, characterized in that, The method of calculating the gross water flow load at each demand node from the end-user demand includes: Identify the directed path connecting each demand node and obtain the water conveyance loss parameters corresponding to that directed path; A local water use factor is introduced to characterize the process from the water conveyance network to the field water use unit, which covers leakage, evaporation and management losses in the water distribution process; By using the water loss parameters and the local water use coefficient, the net irrigation demand is converted into a gross water flow load that must be extracted from the upstream network nodes, ensuring that the effective water volume reaching the end after deducting the multi-level cumulative losses during transmission meets the net irrigation demand.

5. The method according to claim 1, characterized in that, The establishment of a dynamic water allocation model incorporating multi-time-period spatiotemporal coupling characteristics includes: The complete scheduling cycle is discretized into several consecutive time periods, and the flow value of each directed path in each time period is defined as the decision variable to be solved. Construct a global objective function that requires the total outflow from the water source node to reach its minimum value across all time periods; The node flow balance constraint is constructed such that, for any network node other than the water source node, at any time period, the sum of the net values ​​of all upstream directed path flows pointing to that node after deducting the losses determined by their respective water conveyance loss parameters is equal to the sum of all downstream directed path flows pointing from that node. Construct demand response constraints so that the effective flow into each demand node in each time period is not less than the gross water demand load corresponding to that demand node. Construct directed path capacity constraints such that, within each time period, the flow value of each directed path is between zero and the upper bound constraint of the directed path's capacity. Establish water source capacity constraints so that the total flow rate out of the water source node does not exceed the maximum available water volume of the current water source in each time period.

6. The method according to claim 5, characterized in that, The dynamic water allocation model also includes the following constraints: Gate operation constraints are used to limit the flow change of each directed path between adjacent time periods to not exceeding the maximum regulation rate of the corresponding regulating actuator; Time period connection constraints are used to limit the change in water storage volume of the physical water conveyance channel corresponding to each directed path between adjacent time periods to meet the inflow-outflow balance and keep the water storage volume within a safe range.

7. The method according to claim 1, characterized in that, Based on the crop type, crop planting area, crop growth stage, and forecast meteorological data corresponding to each water use unit, the net irrigation demand of each water use unit in each period of the future scheduling cycle is calculated using a crop water balance model.

8. The method according to claim 1, characterized in that, The solution process for the dynamic water allocation model includes: The global optimization problem spanning multiple time periods is decomposed into a series of sub-problems of single-time period traffic allocation arranged in chronological order; Construct a standard flow network model with gain attributes that is topologically equivalent to the lossy directed network structure model, wherein the water conveyance loss parameter is transformed into the gain factor attribute of the directed path in the standard flow network model, and the gain factor characterizes the retention ratio of unit flow after transmission through the directed path. The gross water demand load of each demand node is converted into the lower bound constraint of the flow of the corresponding directed path in the standard flow network model. On the standard flow network model, a maximum flow optimization algorithm with lower bound constraints on flow and gain factor attributes is run to search for a feasible flow distribution scheme that minimizes the outflow of water source while satisfying the lower bounds of flow at all demand nodes.

9. The method according to claim 8, characterized in that, The solution process also includes bottleneck diagnosis and load adjustment steps: If there is no feasible network flow that satisfies the lower bound of the flow of all demand nodes in the single-time flow allocation subproblem during a specific time period, then the bottleneck directed path or bottleneck node set that restricts the global water supply capacity can be identified based on the maximum flow minimum cut theorem. Based on the preset irrigation priority rules, the gross water demand load of the demand nodes in the bottleneck area is dynamically reduced proportionally. The maximum flow optimization algorithm is rerun based on the reduced load until a feasible flow distribution scheme for the current time period is obtained.

10. The method according to claim 1, characterized in that, The generation of the control command sequence for the regulating actuator includes: Obtain the flow characteristic curves of each regulating gate in the irrigation area, and establish a mapping model between the flow through the gate, the water level difference before and after the gate, and the gate opening degree. Based on the optimized flow distribution sequence of each directed path, combined with the water level status data of real-time monitoring or simulation calculation, the target opening degree of each regulating gate in each time period is calculated in reverse using the mapping model. The target opening degree is encapsulated as a control command with a timestamp, and its rate of change is smoothed to conform to the mechanical operation specifications of the adjustment mechanism.

11. The method according to claim 10, characterized in that, The control command sequence is sent to the irrigation district control terminal to execute irrigation scheduling.

12. A system for intelligent irrigation decision-making in an irrigation district, characterized in that, The system is used to implement the method according to any one of claims 1-11, comprising: The canal system topology modeling module is used to construct a lossy directed network structure model that represents the dynamic water conveyance loss generated by water flow in a directed path, based on the spatial connection relationship and engineering parameters of the irrigation area water conveyance system; the network nodes in the lossy directed network structure model include water source nodes and demand nodes. The irrigation demand quantification module is used to determine the net irrigation demand in the future scheduling cycle based on the water demand characteristics of each water-using unit, and to calculate the gross water demand load at each demand node from the end demand by combining the water conveyance loss parameters in the lossy directed network structure model. The optimization decision modeling module is used to establish a dynamic water allocation model with multi-period spatiotemporal coupling characteristics, based on the topological connections of the lossy directed network structure model, the capacity upper bound constraints of each directed path, and the gross water demand load of each demand node, with the global optimization objective of minimizing the total water supply. The dynamic water allocation model embeds node flow balance constraints, which require that the net inflow of each directed path into a node, after being reduced by its corresponding water loss parameters, remains conserved with the total outflow from that node. The flow optimization solution module is used to solve the dynamic water allocation model and generate the optimized flow allocation sequence of each directed path in each time period within the scheduling cycle. The intelligent irrigation execution module is used to generate a control command sequence for adjusting the execution mechanism based on the optimized flow distribution sequence.

13. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 11.

Images