Digital twinborn irrigation area-oriented water resource efficient utilization and rice population quality comprehensive evaluation method and system

Abstract

The invention discloses a digital twinborn irrigation district-oriented water resource efficient utilization and rice population quality comprehensive evaluation method and system, and relates to the technical field of irrigation district water resource efficient utilization and agricultural technology evaluation. Comprising the following steps: constructing a five-dimensional multi-level comprehensive evaluation index system, and determining a level framework; weighting the second-level indexes based on an analytic hierarchy process and calculating scores of the first-level indexes; a comprehensive integration model is established, the five first-level index scores serve as input, the nonlinear relation between the indexes is automatically learned, and an irrigation area comprehensive performance index is output; an evaluation system and a CatBoost model are integrated to a digital twinborn irrigation area platform, and multi-source data automatic access, model online reasoning and evaluation result space visualization are achieved. According to the method, the problems that an existing evaluation method is split in system, difficult in index quantification, depends on subjective authority assignment and cannot achieve dynamic spatialization display are solved, and intelligent and engineering decision support is provided for collaborative optimization of irrigation area water resources and crop production.

Description

Technical Field

[0001] This invention relates to the field of efficient water resource utilization and agricultural technology evaluation in irrigation districts, and in particular to a method and system for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts. Background Technology

[0002] Agricultural water resources are a crucial strategic resource for ensuring agricultural production. Currently, water scarcity and uneven spatial and temporal distribution are significant constraints on my country's food and ecological security. Efficient utilization of irrigation water resources refers to maximizing water use efficiency and minimizing waste through scientific and rational management and technology during farmland irrigation, thereby achieving sustainable agricultural development. As the center of my country's agricultural production, agricultural irrigation districts are a vital pillar for ensuring national food security. Therefore, the evaluation methods for efficient water resource utilization in irrigation districts and the improvement of rice population quality are pressing issues that my country urgently needs to address.

[0003] Currently, experts and scholars have conducted a series of studies on evaluation methods for efficient agricultural water use in irrigation areas. The Institute of Farmland Irrigation, Chinese Academy of Agricultural Sciences, has disclosed a comprehensive evaluation method, equipment, and medium for efficient agricultural water use in well-canal combined irrigation areas (Chinese Patent, Application No. 202311037066.3), which relates to the field of irrigation area water use evaluation technology. The method includes: identifying the target well-canal combined irrigation area and setting evaluation indicators based on the basic data of the target well-canal combined irrigation area. Northeast Agricultural University has disclosed a method for balanced and efficient allocation of water resources in large-area agricultural irrigation areas to address climate change (Chinese Patent, Application No. 202110906820.7), which takes into account the comprehensive effects of multiple objectives, handles the relationship between economic benefits from both subjective and objective perspectives, and comprehensively considers the uncertainty of multiple factors affecting runoff from both current and future perspectives, thereby achieving sustainable and efficient management of irrigation area water resources. However, current research lacks evaluation methods related to the efficient utilization of irrigation water resources and the improvement of rice irrigation quality, especially evaluation methods that comprehensively consider multiple factors such as irrigation system, soil moisture management, crop water requirements, irrigation facility improvement, water resource monitoring and management, water resource recycling, and scientific management and decision support. Through scientific and reasonable management and technical means, we can achieve the economical utilization and sustainable development of water resources.

[0004] The current technology has the following shortcomings: (1) An integrated irrigation district-scale evaluation method based on engineering monitoring and physical models is proposed. In existing studies, the improvement of rice population quality and the efficient utilization of irrigation district water resources are still in a state of fragmented evaluation, lacking a technical system that can achieve collaborative analysis using remote sensing inversion, sensor monitoring and digital twin platform data. This invention proposes an integrated technical evaluation method at the irrigation district scale, which integrates the canal system water conveyance and distribution process, soil-crop growth response, population structure indicators, ecological constraints and irrigation district operation data, and couples them through mechanistic models such as SWAT-MODFLOW, AquaCrop, and InVEST to achieve a systematic and engineering evaluation of "water resource utilization - crop population quality - ecological benefits", solving the problem of long-term separation between the two and lack of feasible technical paths.

[0005] (2) Constructing a five-dimensional, 54-item multi-level engineering indicator system for digital twin irrigation districts. Existing indicator systems are fragmented, unquantifiable, and fail to reflect the true operational status of irrigation districts. This invention constructs a multi-level indicator system consisting of 54 secondary indicators across five major categories: water conservancy, agronomy, management, new quality, and ecology. It clarifies the physical meaning, monitoring methods, and data sources of the indicators, covering various types of engineering data, including statistical business data (irrigation district water supply, planting area, crop yield), remote sensing image data (leaf area index, evapotranspiration, surface temperature), monitoring system data (flow rate, water level), and IoT terminal data (soil moisture, valve status, pump station operation status). This system links the "target layer—primary indicators—secondary indicators—basic data layer," solving the problem of inconsistent data sources and the inability to support multi-source data-driven analysis in traditional evaluation systems.

[0006] (3) A three-layer coupled calculation method is proposed and automated model service integration application is realized. Traditional methods are difficult to handle multi-source heterogeneous data and nonlinear coupling of multiple indicators, and manual weighting is highly subjective. This invention proposes a three-layer calculation method of single indicator normalization, sub-indicator AHP weighting, and comprehensive indicator CatBoost nonlinear integration to realize the dynamic fusion of multi-dimensional indicators. At the same time, the CatBoost model is embedded in the digital twin platform in .cbm format, combined with the PostGIS spatial database, RESTful / MQTT data interface and ArcGIS / SuperMap spatial calculation to realize automated evaluation, spatial mapping, dynamic visualization and decision support. This approach overcomes the problems of existing methods being unable to run in engineering scenarios for a long time, unable to update automatically and unable to display spatially, and provides an intelligent, engineering-oriented and implementable technical solution for the comprehensive performance evaluation of irrigation districts.

[0007] Therefore, proposing a method and system for the efficient utilization of water resources and comprehensive evaluation of rice population quality in digital twin irrigation districts, which can integrate multi-source data, provide objective quantitative evaluation, dynamic visualization, and be embedded in actual engineering platforms, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0008] In view of this, the present invention provides a method and system for the comprehensive evaluation of efficient water resource utilization and rice population quality in digital twin irrigation districts. It can realize a synergistic, objective, dynamic and spatial comprehensive evaluation of water resource utilization efficiency and rice population growth quality at the irrigation district scale, and provide intelligent support for management decision-making.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A method for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts includes the following steps: S1 Steps for constructing a multi-level comprehensive evaluation index system: Based on the concept of coupling water-crops-management-technology-ecosystem, establish an index system that includes a target layer, a primary index layer, a secondary index layer, and a basic data layer. The primary index layer includes water conservancy indicators, agronomic indicators, management indicators, new quality indicators, and ecological indicators. S2 Acquisition and Processing of Basic Data Steps: Based on the basic data layer, collect and integrate multi-source heterogeneous data from the irrigation area, including ground monitoring data, remote sensing inversion data, statistical and management data, and IoT terminal data. Standardize the raw data corresponding to each secondary indicator to achieve dimensionless and directional consistency. S3 Steps for Calculating the Score of Primary Indicators: For each primary indicator, the weight of each secondary indicator under it is determined by the analytic hierarchy process (AHP). Based on the standardized data and weights, the scores of the five primary indicators of water conservancy, agronomy, management, new quality and ecology are calculated by weighting. S4 Steps for constructing and applying the integrated model: Construct an integrated model based on the CatBoost gradient boosting decision tree algorithm, using the scores of the five primary indicators calculated in S3 as input features and the comprehensive performance index of the irrigation district as the output target, train the model, use the trained model to comprehensively evaluate the target irrigation district, and output the comprehensive performance index. S5 Results Visualization and Decision Support Steps: Integrate the evaluation results obtained in S4 into the digital twin irrigation district platform. Through the spatial computing and visualization module, generate and display the spatial distribution map and dynamic evolution information of the irrigation district's efficient water resource utilization and rice population quality improvement level.

[0010] The above methods may optionally include the following secondary indicators in S1: water conservancy indicators include the integrity rate of canal structures, irrigation water utilization rate, and crop water productivity; agronomic indicators include leaf area index, yield, and fertilizer utilization rate; management indicators include the degree of standardized management and water fee collection rate; new quality indicators include the degree of digital infrastructure improvement and intelligence level; and ecological indicators include the irrigation water quality compliance rate, fertilizer use rate, and water resource utilization rate.

[0011] Optionally, ground monitoring data can be collected using ultrasonic flow meters, pressure water level gauges, and FDR / TDR soil moisture monitors, with a sampling frequency of 10-30 minutes. Remote sensing inversion data can be obtained using Sentinel-2 MSI and Landsat-8 / 9OLI–TIRS satellite imagery, with evapotranspiration retrieved using the SEBAL energy balance model and leaf area index obtained using the NDVI-LAI empirical model.

[0012] Optionally, in the above method, quantitative positive indicators in S3 are normalized using the extreme value method, inverse indicators are normalized using reverse normalization, and qualitative indicators are quantified into 0-1 range values ​​using expert scoring or fuzzy membership functions.

[0013] The above method, optionally, includes S4 specifically: S401 Constructing the training sample set: , in, x i =[ W 1i , W 2i , W 3i , W 4i , W 5i ] indicates the first i The score vectors of the five primary indicators for each sample. y i The corresponding known comprehensive performance index; the CatBoost model in the th... t The prediction function for the round of iterations is: ; in, F t-1 ( x () represents the value predicted in the previous round. For learning rate, h t ( x ) represents the current round's weak learner, i.e., a symmetric decision tree; S402 initializes the CatBoost model parameters, including the learning rate, number of iterations, and tree depth; S403 optimizes the objective function L of the CatBoost model through iterative training. L The least squares error function is: , in, For the first t Model predictions after rounds of iterations; S404 inputs the scores of the five primary indicators of the irrigation district to be evaluated into the trained CatBoost model to obtain the comprehensive performance index of the irrigation district to be evaluated.

[0014] Optionally, S4 also includes: using the SHAP method to interpret the trained CatBoost model and quantify the contribution of the five primary indicators of water conservancy, agronomy, management, new quality and ecology to the comprehensive performance index.

[0015] Optionally, the above method uses the CGCS2000 or WGS84 coordinate system for spatial registration in S4, performs spatial calculations through ArcGIS Engine or SuperMap iServer, and uses Leaflet or MapboxGL for visualization on the front end.

[0016] A comprehensive evaluation system for efficient water resource utilization and rice population quality in digital twin irrigation districts, used to execute the method for efficient water resource utilization and rice population quality in digital twin irrigation districts described above, includes a data acquisition and fusion module, an index calculation and processing module, a comprehensive model service module, and a visualization and decision support module connected in sequence. The data acquisition and fusion module is used to acquire and fuse multi-source heterogeneous data from the digital twin irrigation district platform to form a basic data layer; The indicator calculation and processing module is used to standardize the basic data according to the preset multi-level comprehensive evaluation indicator system and calculate the scores of each level of indicator. The integrated model service module contains a pre-trained integrated model based on the CatBoost algorithm. It receives the scores of the five primary indicators from the indicator calculation and processing module, calls the model to perform calculations, and returns the comprehensive performance index. The visualization and decision support module, integrated into the digital twin irrigation district platform, is used to perform spatial raster rendering and dynamic visualization of the comprehensive performance index and scores of indicators at all levels.

[0017] The above system, optionally, includes a comprehensive model service module that stores and loads CatBoost models in .cbm format files, and supports receiving input data and returning evaluation results via RESTful API or MQTT interface.

[0018] The above-mentioned system, optionally, includes a visualization and decision support module based on a geographic information system platform, which can generate heat maps, iso-zoning maps, and time series evolution maps of comprehensive performance index, water resource utilization efficiency, and rice population quality improvement level.

[0019] As can be seen from the above technical solution, compared with the prior art, the present invention provides a method and system for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts, which has the following beneficial effects: (1) Achieve integrated evaluation: Integrate water resource utilization and rice population quality evaluation, and realize the systematic evaluation of water resource utilization, crop population quality and ecological benefits through multi-model coupling to solve the long-term problem of separation between the two; (2) Engineering of the indicator system: Construct a five-dimensional, 54-item multi-level indicator system, clarify the data sources and monitoring methods, support multi-source data-driven analysis, and solve the problem of fragmentation of traditional indicators; (3) Intelligent evaluation method: The three-layer coupling calculation method handles the nonlinear coupling of multi-source heterogeneous data, and the model is embedded in the digital twin platform to realize automated and dynamic evaluation, solving the problems of strong subjectivity and difficulty in engineering application of traditional methods; (4) Visualizing decision-making: Realizes spatial mapping and dynamic display of evaluation results, supports zonal diagnosis and decision support, and provides intuitive and scientific technical support for irrigation district management. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 This invention discloses a flowchart of a method for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts; Figure 2 This invention discloses an evaluation method for efficient utilization of irrigation water resources and improvement of rice population quality. Figure 3 This is a flowchart of the multi-source data acquisition and processing disclosed in this invention; Figure 4This invention discloses a structural block diagram of a system for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] In this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0024] To achieve a comprehensive evaluation of efficient water resource utilization and synergistic improvement in rice population quality in irrigation districts, this invention, based on the concept of "water-crop-management-ecology" system coupling, constructs a multi-level indicator system applicable to irrigation district scales, consisting of "target layer—primary indicator—secondary indicator," providing a quantitative basis for subsequent evaluation models and technical methods. Supported by multi-source data, this system comprehensively reflects water resource utilization efficiency, crop population quality improvement level, and management technical effectiveness from aspects such as irrigation project operation, agronomic management, management efficiency, digitalization level, and ecological benefits. It is a core component of the overall evaluation method of this invention.

[0025] Principles for constructing an indicator system: In constructing the irrigation district-scale index system, this invention follows the principles of scientific rigor, systematic approach, operability, and independence to ensure the objectivity and feasibility of the evaluation results.

[0026] Scientific principle: Each indicator should be able to scientifically reflect the inherent coupling relationship between water resource utilization and rice population quality improvement, and reflect the comprehensive impact of technical measures on yield and water efficiency; Systematic principle: The system covers the entire process of "water-crops-management-technology-ecology", taking into account both water allocation and distribution efficiency, as well as crop growth response and ecological constraints; The principle of operability: the indicator data can be obtained through irrigation area monitoring, remote sensing inversion, statistical data or intelligent sensing systems, and is quantifiable and repeatable; Independence principle: low correlation between indicators and strong information complementarity, avoiding redundant expressions and multicollinearity, and ensuring the clarity and reliability of the system.

[0027] The indicator system constructed based on the above principles has both a scientific and logical foundation and is easy to embed into crop and water resource models and subsequent software development.

[0028] See Figure 1 and Figure 2 As shown, this invention discloses a method for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts, comprising the following steps: S1 Steps for constructing a multi-level comprehensive evaluation index system: Based on the concept of coupling water-crops-management-technology-ecosystem, establish an index system that includes a target layer, a primary index layer, a secondary index layer, and a basic data layer. The primary index layer includes water conservancy indicators, agronomic indicators, management indicators, new quality indicators, and ecological indicators. S2 Basic Data Acquisition and Processing Steps: Based on the basic data layer, collect and integrate multi-source heterogeneous data from the irrigation area, including ground monitoring data, remote sensing inversion data, statistical and management data, and IoT terminal data. Standardize the raw data corresponding to each secondary indicator to achieve dimensionlessness and directional consistency. (Specifically, based on the basic data layer, collect and integrate multi-source heterogeneous data from the irrigation area, including ground monitoring data (such as flow rate, water level, and soil moisture), remote sensing inversion data (such as leaf area index and evapotranspiration), statistical and management data (such as water use plans and water prices), and IoT terminal data (such as valve status and pump station operating parameters). Standardize and forward-orientedly process the raw data corresponding to each secondary indicator.)

[0029] S3 Steps for Calculating the Score of Primary Indicators: For each primary indicator, the weight of each secondary indicator under it is determined by the analytic hierarchy process (AHP). Based on the standardized data and weights, the scores of the five primary indicators of water conservancy, agronomy, management, new quality and ecology are calculated by weighting. S4 Steps for constructing and applying the integrated model: Construct an integrated model based on the CatBoost gradient boosting decision tree algorithm, using the scores of the five primary indicators calculated in S3 as input features and the comprehensive performance index of the irrigation district as the output target, train the model, use the trained model to comprehensively evaluate the target irrigation district, and output the comprehensive performance index. S5 Results Visualization and Decision Support Steps: Integrate the evaluation results obtained in S4 into the digital twin irrigation district platform. Through the spatial computing and visualization module, generate and display the spatial distribution map and dynamic evolution information of the irrigation district's efficient water resource utilization and rice population quality improvement level.

[0030] Furthermore, the secondary indicators of water conservancy indicators in S1 include the integrity rate of canal structures, irrigation water utilization rate, and crop water productivity; the secondary indicators of agronomic indicators include leaf area index, yield, and fertilizer utilization rate; the secondary indicators of management indicators include the degree of standardized management and water fee collection rate; the secondary indicators of new quality indicators include the degree of digital infrastructure improvement and intelligence level; and the secondary indicators of ecological indicators include the irrigation water quality compliance rate, fertilizer use rate, and water resource utilization rate.

[0031] The primary indicator layer includes water conservancy indicators (reflecting engineering and water transmission efficiency), agronomic indicators (reflecting crop growth and agronomic management), management indicators (reflecting system and operational efficiency), new quality indicators (reflecting digitalization and intelligentization levels), and ecological indicators (reflecting green sustainability).

[0032] Furthermore, ground monitoring data were collected using ultrasonic flow meters, pressure water level gauges, and FDR / TDR soil moisture monitors, with a sampling frequency of 10-30 minutes. Remote sensing inversion data were obtained using Sentinel-2 MSI and Landsat-8 / 9 OLI–TIRS satellite imagery, with evapotranspiration retrieved using the SEBAL energy balance model and leaf area index obtained using the NDVI-LAI empirical model.

[0033] Furthermore, in S3, quantitative positive indicators are normalized using the extreme value method, negative indicators are normalized using the reverse method, and qualitative indicators are quantified into 0-1 range values ​​using expert scoring or fuzzy membership functions. Specifically, for each type of primary indicator, the weights of its subordinate secondary indicators are determined using the analytic hierarchy process (AHP), and the scores of the five primary indicators—water conservancy, agronomy, management, new quality, and ecology—are calculated based on the standardized data and the aforementioned weights.

[0034] Furthermore, S4 specifically includes: S401 Constructing the training sample set: , in, x i =[ W 1i , W 2i , W 3i , W 4i , W 5i] indicates the first i The score vectors of the five primary indicators for each sample. y i The corresponding known comprehensive performance index; the CatBoost model in the th... t The prediction function for the round of iterations is: ; in, F t-1 ( x () represents the value predicted in the previous round. For learning rate, h t ( x ) represents the current round's weak learner, i.e., a symmetric decision tree; S402 initializes the CatBoost model parameters, including the learning rate, number of iterations, and tree depth; S403 optimizes the objective function L of the CatBoost model through iterative training. L The least squares error function is: , in, For the first t Model predictions after rounds of iterations; S404 inputs the scores of the five primary indicators of the irrigation district to be evaluated into the trained CatBoost model to obtain the comprehensive performance index of the irrigation district to be evaluated.

[0035] Furthermore, S4 also includes: using the SHAP method to interpret the trained CatBoost model and quantify the contribution of five primary indicators—water conservancy, agronomy, management, new quality, and ecology—to the comprehensive performance index.

[0036] Furthermore, S4 employs CGCS2000 or WGS84 coordinate systems for spatial registration, performs spatial calculations using ArcGIS Engine or SuperMap iServer, and utilizes Leaflet or MapboxGL for visualization on the front end. Specifically, the evaluation results are integrated into the digital twin irrigation district platform. Through spatial calculation and visualization modules, spatial distribution maps, heat maps, and dynamic evolution information of the irrigation district's efficient water resource utilization and rice population quality improvement levels are generated and displayed, supporting zoning diagnosis and decision-making.

[0037] like Figure 2 As shown, the core of the comprehensive evaluation method provided by this invention lies in constructing an intelligent evaluation system that is deeply integrated with the digital twin irrigation district platform, moving from data to decision-making. The steps are described in detail below: S1: Constructing a multi-level comprehensive evaluation index system: The indicator system of this invention adopts a four-layer logical structure of "target layer - primary indicator layer - secondary indicator layer - basic data layer".

[0038] The target layer, namely "Comprehensive Performance of Efficient Water Resource Utilization and Rice Population Quality Improvement in Irrigation Districts," is the final output of the evaluation. This layer reflects the overall performance of the irrigation district in terms of water resource allocation efficiency, technology application level, crop production quality, and ecological environment coordination, and is the final calculation target of the evaluation method of this invention. The score results of the target layer can be used for horizontal comparisons between different irrigation districts, and also for vertical tracking of technological improvements within the same irrigation district.

[0039] The primary indicator layer comprises five dimensions: water conservancy indicators (W1), agronomic indicators (W2), management indicators (W3), new quality indicators (W4), and ecological indicators (W5). These five dimensions systematically cover the key aspects affecting irrigation district performance. Specifically, water conservancy indicators (W1) reflect the integrity of the engineering system, water transmission and distribution efficiency, and water resource supply security capacity, forming the foundation for efficient water use. Agronomic indicators (W2) reflect the degree of matching between rice population growth and agronomic management measures, a core element for improving crop quality. Management indicators (W3) reflect the efficiency of system implementation, operation and maintenance, and resource allocation, providing institutional support for achieving water-saving management and population regulation. New quality indicators (W4) reflect the depth of application of digital, intelligent, and information technologies in rice production and irrigation district regulation, representing an important pathway to achieving intelligent and precise management. Ecological indicators (W5) reflect the benefits of energy conservation, emission reduction, and green production, serving as ecological constraints and optimization directions to ensure the sustainable development of the irrigation district. The five primary indicators are independent of each other, yet form a closed-loop system: water conservancy indicators and agronomic indicators jointly determine the efficiency of the water-crop process; management indicators and new quality indicators reflect the regulatory capabilities of humans and technology; and ecological indicators, as a constraint feedback layer, correct the overall operation for greening and sustainability.

[0040] Secondary indicator layer: Each primary indicator is further subdivided into several quantifiable and obtainable specific indicators. Each secondary indicator has a clear definition, unit, and data source, which can be obtained through monitoring, remote sensing, or statistical data, and is calculable and traceable. For example: Water conservancy indicators (W1) include the integrity rate of canal structures, the integrity rate of wells, the utilization rate of wells, the ratio of actual irrigated area, the irrigation water volume per unit area, the water consumption per unit area, the irrigation water utilization rate, the crop water productivity, the irrigation water productivity, the water consumption productivity, the drainage rate, the flood drainage compliance rate, the water price, the water fee collection rate, and the internal recycling rate of the irrigation area.

[0041] Agronomic indicators (W2) include fertilizer utilization rate, plant height, leaf area index, chlorophyll content, machine planting rate, rotary tillage rate, straw return rate, agricultural film coverage rate, yield, quality, and root growth.

[0042] The management indicators (W3) include the percentage of standardized construction in the irrigation district, the degree of standardized management, the rate of sound management institutions, the number of employees per unit irrigation area, the operation, management and maintenance cost per cubic meter of irrigation water, the degree of informatization construction, the area rate of large-scale and intensive development, the water fee collection rate, and the water rights coverage rate.

[0043] The new quality indicators (W4) include the degree of improvement of digital infrastructure in irrigation districts, the allocation rate of digital professionals, the degree of improvement of disaster prevention, mitigation, early warning and monitoring, the level of intelligence, the matching rate of digital water measurement facilities, the degree of digital construction of pumping stations, and the coverage rate of digital control.

[0044] Ecological indicators (W5) include irrigation water quality compliance rate, groundwater level compliance rate, changes in groundwater over-extraction area, groundwater level compliance rate, fertilizer use rate, energy conservation level, greening rate, soil erosion rate, proportion of green and high-quality agricultural products, and water resource utilization rate.

[0045] The total number of secondary indicators can be adjusted according to the specific irrigation district conditions, for example, 54 items.

[0046] The foundational data layer contains the raw data sources that support the calculation of all the above indicators, including remote sensing imagery, sensor monitoring data, statistical annual reports, and IoT terminal data. Through data assimilation and standardization, it provides a unified input format for upper-layer calculations.

[0047] Step S2: Acquire and process basic data (see...) Figure 3 ): To ensure the operability and universality of the evaluation method, the indicator system constructed in this invention is supported by multi-source, multi-scale data fusion, combining ground monitoring, remote sensing inversion, statistical data, and digital twin platform data to achieve comprehensive information integration from single-point observation to regional scale. The correspondence between various data types and system levels is shown in Table 1: Table 1. Correspondence between various data types and system levels

[0048] The system automatically accesses multi-source data through various interfaces: The remote sensing data service is called through the RESTful API to obtain raster data such as LAI and ET. Specifically, (1) Remote sensing and management data interface: Remote sensing and management data is obtained through the RESTful API interface based on the HTTP / HTTPS protocol. This interface is used to retrieve remote sensing inversion raster data such as evapotranspiration, leaf area index, and vegetation coverage, and to retrieve irrigation area management data such as water price and water rights. The interface request parameters include time range, spatial area and index type, and the returned format is a unified identifier path of JSON data or raster data. The platform can call this interface as needed to automatically complete the data capture, parsing and storage. (2) Internet of Things monitoring data interface: This invention accesses irrigation area sensor data through the MQTT protocol. Water level gauges, flow meters, soil moisture meters, meteorological monitoring stations and other equipment continuously push data to the platform through preset topics. The platform, as the subscriber, parses the real-time data according to a unified structure, records the equipment number, collection time, latitude and longitude location, monitoring value, unit and other information, and writes it into the time series data table in the database to support minute-level irrigation area status monitoring. (3) Database Connection and Unified Storage: This invention uses PostgreSQL and PostGIS to establish a unified data storage system. Remote sensing inversion and model calculation results are stored in GeoTIFF raster format; basic geographic data such as irrigation area boundaries, canals, and fields are stored in Shapefile or GeoJSON format; management data and monitoring data are stored in structured tables. All data containing spatial attributes are indexed spatially to ensure rapid spatial query and visualization access within a large irrigation area. Historical evaluation results are recorded in the structure of "irrigation area identifier, management unit, evaluation time, and indicator value," providing a basis for trend assessment and rolling monitoring.

[0049] Retrieves structured data such as statistical reports, water rights, and water prices from the irrigation district management information system.

[0050] Obtain status data of equipment such as valves and pump stations from a digital twin platform.

[0051] After the acquired raw data is cleaned, format converted, and spatiotemporally registered, it is extracted and calculated according to the indicator definition. All secondary indicator values ​​are normalized to eliminate the influence of dimensions, so that all indicator values ​​fall within the [0,1] interval.

[0052] Quantitative indicator normalization

[0053] For positive indicators (higher values ​​indicate better performance), normalization is performed using the extreme value method: ; For inverse metrics (smaller values ​​indicate better performance), inverse normalization is used: ; in,X ij For the first i The evaluation object is in the first j The original values ​​under each indicator X ij This is the result of normalization.

[0054] Quantification of qualitative indicators

[0055] For non-numerical indicators such as management level, degree of digitalization, and ecological green level, expert scoring or fuzzy membership functions are used for quantification, with a scoring range of 0–1. Ultimately, all indicators are transformed into a unified dimensionless data matrix. R=[X ij ] As input for subsequent calculations.

[0056] Data fusion and processing: This invention aims to construct a unified data fusion system for typical rice irrigation areas, integrating and standardizing ground monitoring data, remote sensing inversion data, statistical and management data, and intelligent sensing data from multiple sources to achieve collaborative use of different time scales, spatial scales, and data types.

[0057] Ground monitoring data: Data including well and canal flow, canal water level, groundwater level, and soil moisture content are collected using ultrasonic flow meters, pressure water level gauges, and FDR / TDR soil moisture monitors installed at locations such as irrigation canals, farms, canals, and pumping stations. The typical sampling frequency is 10-30 minutes. Groundwater level monitoring uses in-hole pressure water level gauges at a monitoring depth of 20-80m to generate high-frequency time-series raw data.

[0058] Remote sensing inversion data: Using multi-source satellite imagery, including Sentinel-2 MSI (10–20 m) and Landsat-8 / 9 OLI–TIRS (30 m), regional-scale parameters such as leaf area index (LAI), evapotranspiration (ET), and land surface temperature (LST) were retrieved. ET was retrieved using the SEBAL energy balance model, and LAI was obtained using the NDVI-LAI empirical model. The remote sensing results were used to supplement information on crop growth, water consumption, and thermal conditions at the area scale.

[0059] Statistical and management data: It originates from the irrigation district's annual statistical reports, water pricing policies, water rights allocation, annual water use plans, standardized construction archives, and other business systems. It is used to describe the irrigation district's input and output, management system, and water allocation strategy, and to provide the management data required for model calibration and comprehensive evaluation.

[0060] IoT terminal data: Data comes from a digital twin irrigation district platform, field IoT terminals, and drone monitoring systems. It includes valve status, pump station operating parameters, field-level soil moisture, aerial imagery, and anomaly area identification results, enabling multi-dimensional dynamic perception of irrigation district water conditions, moisture levels, and facility operating status.

[0061] Spatial organization and platform embedding: This structure is specifically designed for applications at the "irrigation district scale." Each primary indicator can be calculated and aggregated at spatial zoning units (such as tributaries, main canals, and management areas) to form hierarchical evaluation results, achieving scale expansion from point to area. Through data interface connection with the digital twin irrigation district platform, monitoring data can be accessed in real time to complete automated evaluations, providing dynamic support for management decisions and realizing an automated closed loop from "data acquisition—model calculation—result display."

[0062] Scope of application: This system uses a typical rice irrigation area as its application scenario, focusing on evaluating the comprehensive effects of technologies for improving rice population quality, such as intermittent irrigation, precision fertilization, and intelligent management. It also possesses good scalability and portability. At the crop level, the adjustable agronomic parameters are applicable to major crops such as wheat, corn, and cotton; At the regional level, this can be extended to different types of irrigation districts or watersheds; At the system level, it can serve as an independent module of the digital twin agricultural water resources management platform, enabling automated evaluation and dynamic display.

[0063] In summary, the indicator system of this invention forms a multi-dimensional integration, regional expansion, and method embedding framework in terms of structural hierarchy, data support, and scope of application, laying the foundation for establishing an "Evaluation Method for High-Efficiency Water Resource Utilization and Rice Population Quality Improvement Technology" at the irrigation district scale.

[0064] Step S3: Calculate the scores for the primary indicators: For each type of primary indicator (such as water conservancy indicators), the weights of all its subordinate secondary indicators are determined using the analytic hierarchy process (AHP). First, a judgment matrix is ​​constructed: ; in, a ij Indicates the first i The indicator relative to the first j The importance of the indicators.

[0065] The relative importance of indicators is determined by expert scoring, then the weight vector is calculated and a consistency test is performed (requirements). CR <0.1). The largest eigenvalue can be obtained using the eigenvector method or the root method. With corresponding feature vectors Calculate the consistency index CI Consistency ratio CR : ; when CR If the result is less than 0.1, the matrix passes the consistency check. If it fails, the matrix is ​​revised.

[0066] After the test is passed, the overall score of this primary indicator is calculated using a weighted summation formula: , in, S k For the first k The overall score of the category indicators, w j For the first j The weight of each secondary indicator (individual indicator), These are normalized values. The results are denoted as follows: S 1 (Water Conservancy) S 2 (Agronomy) S 3 (manage), S 4 (New quality) and S 5 (Ecology). Five scores, W1 to W5, were calculated separately.

[0067] Step S4: Build and apply the integrated model: To comprehensively reflect the multidimensional characteristics of "water-crops-management-technology-ecology" at the irrigation district scale, this invention embeds different models at each indicator layer for computational support, forming a technical path of "model-driven-indicator quantification-result synthesis".

[0068] Hydraulic index calculation model: SWAT-MODFLOW coupled model

[0069] This model is used to simulate the joint regulation process of surface water and groundwater, and to calculate the water balance and water resource utilization rate of the irrigation area. The model inverts the transport and distribution efficiency through parameters such as irrigation water diversion, seepage, and evapotranspiration, providing quantitative support for water conservancy indicators. Parameters were determined through field measurements, SWAT-CUP automatic calibration (SUFI-2, GLUE), and comparative literature. The value ranges of key parameters are shown in Table 2. Table 2 Key Parameters of the SWAT-MODFLOW Coupled Model

[0070] Agronomic index calculation model: AquaCrop crop growth model

[0071] This method is used to simulate the population growth, yield, and water use efficiency of rice under different irrigation and management conditions, assess the impact of water regulation measures on population quality, and output results such as crop yield, water productivity, and leaf area index. Parameters were determined based on field trial measurements (plant height, LAI), the FAO user manual, and literature calibration, as shown in Table 3.

[0072] Table 3 Key parameters of the AquaCrop crop growth model

[0073] Ecological Indicator Calculation Model: InVEST Ecosystem Services Model

[0074] It is used to calculate the degree to which water-saving measures improve ecosystem service functions (such as water conservation, nutrient retention, and soil conservation), and to provide spatial quantitative results for ecological indicators.

[0075] Management and New Quality Indicator Calculation Model: Fuzzy-AHP Fuzzy Comprehensive Evaluation Model

[0076] By using fuzzy mathematics, qualitative indicators such as management level, system implementation, and information technology construction are fuzzily quantified to achieve a comprehensive evaluation of multiple factors.

[0077] Integrated Model: Analytic Hierarchy Process

[0078] This invention constructs a comprehensive fusion model based on CatBoost gradient boosting decision trees in the comprehensive evaluation process of efficient water resource utilization and rice population quality improvement technologies in irrigation areas. This model achieves nonlinear fusion and intelligent integration from secondary indicators to primary indicators and then to the target layer. While inheriting the logical structure of multi-level indicators, this model introduces machine learning algorithms to automatically learn the nonlinear relationships and coupling effects between indicators, thereby realizing a shift from "expert-based weighting" to "data-driven weight learning," significantly improving the objectivity, adaptability, and dynamic updating capability of the evaluation results.

[0079] The model input includes standardized scores for five primary indicators: water resources (W1), agronomy (W2), management (W3), crop quality (W4), and ecology (W5). These indicators are derived from physical models (SWAT-MODFLOW, AquaCrop, etc.), remote sensing inversion, field monitoring, and management statistics, comprehensively reflecting the irrigation district's performance in water resource utilization efficiency, crop population characteristics, operational management level, intelligence level, and ecological performance. The model uses the irrigation district's comprehensive performance index, Sttotal, as the output target. Through an ordered gradient boosting mechanism, it gradually optimizes the prediction residuals in multiple iterations, achieving a nonlinear mapping and weight self-learning between the input indicators and the target performance.

[0080] Traditional methods involve weighted summation of the scores of the five primary indicators using AHP, which remains linear and subjective. This invention introduces the CatBoost machine learning model to capture their non-linear coupling relationships.

[0081] Model training: Collect historical data or generate training sample sets through scenario simulation. , in, x i =[ W 1i , W 2i , W 3i , W 4i , W 5i ] indicates the first i The input vector of five primary indicators for each sample. y i The corresponding real overall performance (which can be determined from historical observations or simulation data) is used. Model hyperparameters are set (e.g., learning rate = 0.05, number of iterations = 200, tree depth = 6), and training is performed using the least squares error as the loss function.

[0082] Model Inference: Input the score vectors of the five primary indicators of the irrigation district to be evaluated into the trained CatBoost model, and the model automatically outputs a comprehensive performance index. S total This index comprehensively reflects the overall performance of the irrigation district. The CatBoost model in the... t The prediction function for the round of iterations is: ; in, F t-1 (x) This is the value predicted in the previous round. For learning rate, h t (x) This is the current round of weak learners (i.e., symmetric decision trees). The model minimizes the squared error loss function: ; Calculate the pseudo-residual in each round: ; Train the next tree using the residual as the target. h t (x)CatBoost's ordered gradient estimation method generates independent subsets based on the temporal or sequential order of samples, effectively preventing target leakage and gradient bias, and improving the model's stability and generalization ability under conditions of small samples and heterogeneous data. The final comprehensive performance prediction model is as follows: ; in, T For the number of iteration rounds, F 0 represents the initial constant term (generally the sample mean). The model automatically models the nonlinear coupling relationships between various indicators through iterative learning, thereby achieving intelligent prediction and dynamic optimization of overall performance.

[0083] Model Explanation: In the model output stage, Feature Importance and Shapley Additive Explanations (SHAP) methods are introduced to quantify the contribution of each primary indicator and its secondary indicators (W1-W5). Each indicator contributes to the final... The contribution and direction (positive or negative) of the values ​​are analyzed to understand the key factors influencing performance. The importance of each type of indicator is calculated based on its average contribution to the reduction of the loss function across different rounds of improvement: ; in, Indicates the first t Wheel of Life k The contribution of each indicator category to the reduction of the loss function is analyzed. The SHAP value is used to further explain the marginal impact of each indicator in different spatial units and time periods, providing data support for water resource regulation and crop quality improvement in irrigation areas. For example, the model can identify the dynamic pattern of increased contribution rates of water conservancy and agronomic indicators during the dry season, while the weights of ecological and new quality indicators increase during the wet season, providing a quantitative basis for precise regulation of irrigation areas in terms of "water conservation, stable yield, and ecological balance."

[0084] To ensure the stability and feasibility of model training, the key hyperparameters of the CatBoost model are explicitly configured: the learning rate is selected within the range of 0.03 to 0.10, and a value of 0.05 is used in this embodiment; the number of iterations is set to 200; the tree depth is 6 layers; the loss functions used are RMSE and MAPE; and Ordered Boosting is used for sampling to avoid target leakage in sequential data. Furthermore, this invention uses a grid search and 5-fold cross-validation method to jointly fine-tune the above hyperparameters, ensuring that the model has good generalization ability and convergence performance under different irrigation district sample sizes, different time scales, and heterogeneous data conditions. This invention provides clear and executable parameter settings for the CatBoost model, meeting the engineering deployment requirements for comprehensive irrigation district performance evaluation.

[0085] The integrated model of this invention is applicable not only to cross-regional learning of samples from multiple irrigation districts, but also to adaptive optimization under single irrigation district conditions. When samples originate from the same irrigation district, the model constructs a sample set using monitoring data from multiple time periods (annual, quarterly, and monthly) and data from multiple spatial units (canal systems, fields, and management areas), utilizing spatiotemporal heterogeneity to achieve parameter training and weight updates. This method can still capture the nonlinear relationships between indicators even with limited data samples, enabling dynamic evaluation and optimization of comprehensive performance, and possesses strong regional adaptability and potential for widespread application.

[0086] In terms of platform implementation, the CatBoost model is embedded in the model service module of the digital twin irrigation district platform in .cbm format, forming a closed-loop operation mechanism of "offline training - online inference - periodic retraining". During the model training phase, the optimal hyperparameters are determined through cross-validation and grid search; during the online inference phase, the system automatically calculates the comprehensive performance results after receiving the latest monitoring or remote sensing data. The platform supports quarterly or annual rolling updates and achieves adaptive evolution of parameters through an incremental training mechanism, enabling the model to dynamically reflect the impact of climate change, crop structure, and management strategy adjustments.

[0087] In summary, the CatBoost integrated model proposed in this invention achieves intelligent, dynamic, and interpretable evaluation systems while maintaining a hierarchical logical structure. Compared with the traditional Analytic Hierarchy Process (AHP), this model does not require manual weight setting; instead, it automatically learns the weights and coupling relationships of various indicators based on data-driven approaches, thus better reflecting the actual operating status of complex irrigation district systems.

[0088] Step S5: Results visualization and decision support.

[0089] The comprehensive performance index calculated above S totalThe scores of each primary and secondary indicator are combined with the spatial geographic information of the irrigation area.

[0090] In the GIS module of the digital twin irrigation district platform, S total Values ​​are mapped to different management units (such as the control area of ​​branch canals and fields), and rendered through color gradients (heat maps) or hierarchical zoning maps to intuitively show the spatial differences in performance.

[0091] It provides a timeline control that allows you to compare evaluation results from different years and quarters, showing the dynamic trend of performance changes.

[0092] It supports clicking to query any spatial unit, and displays its detailed indicator score radar chart, historical change curve and problem diagnosis report.

[0093] The visualization results can be directly used for the performance evaluation of irrigation district management departments, identification of water-saving areas, assessment of the effectiveness of technical measures, and future planning decisions.

[0094] Specifically, to achieve intuitive expression of evaluation results and intelligent management decision-making, this invention constructs a spatial computing and result display module in the digital twin irrigation district platform. This module organizes and dynamically visualizes the evaluation results in a spatial raster format, achieving integrated "data-model-space-decision" functionality. Using basic geographic information data of the irrigation district as a base map, this module integrates monitoring data, remote sensing inversion data, and model calculation results. Spatial registration and resolution resampling are performed through a unified coordinate system (CGCS2000 or WGS84) to ensure spatiotemporal consistency between different data sources.

[0095] The technical implementation of this module includes three aspects: data interface integration, model embedding method, and GIS visualization implementation.

[0096] Data interface and integration method: (1) Remote sensing and management data interface Remote sensing and management data are obtained through a RESTful API interface based on the HTTP / HTTPS protocol. This interface is used to retrieve remote sensing inversion raster data such as evapotranspiration, leaf area index, and vegetation cover, as well as irrigation district management data such as water prices and water rights. The interface request parameters include the time range, spatial region, and indicator type. The returned data is in JSON format or a unified identifier path for raster data. The platform can call this interface as needed to automatically complete data capture, parsing, and storage.

[0097] (2) IoT monitoring data interface

[0098] This invention accesses irrigation district sensor data via the MQTT protocol. Devices such as water level gauges, flow meters, soil moisture meters, and meteorological monitoring stations continuously push data to the platform through preset topics. The platform, acting as a subscriber, parses the real-time data according to a unified structure, recording information such as device number, collection time, latitude and longitude location, monitoring value, and unit, and writes it into a time-series data table in the database to support minute-level monitoring of the irrigation district's status.

[0099] (3) Database connectivity and unified storage

[0100] This invention employs PostgreSQL and PostGIS to establish a unified data storage system. Remote sensing inversion and model calculation results are stored in GeoTIFF raster format; basic geographic data such as irrigation district boundaries, canals, and fields are stored in Shapefile or GeoJSON format; management and monitoring data are stored in structured tables. All data containing spatial attributes are indexed spatially to ensure rapid spatial queries and visualization access across a large irrigation district. Historical evaluation results are recorded according to the structure of "irrigation district identifier, management unit, evaluation time, and indicator value," providing a foundation for trend assessment and rolling monitoring.

[0101] Model embedding process and service-oriented implementation

[0102] (1) Model file deployment

[0103] After training the CatBoost integrated model in an offline environment, the training results are stored in a dedicated model file format (.cbm) and deployed in the model library directory of the digital twin irrigation district platform so that the model service module can automatically load it upon startup.

[0104] (2) Model service operation

[0105] The platform has a dedicated model service module for receiving evaluation requests and invoking models for inference. The model input consists of standardized values ​​for five primary indicators (water resources, agronomy, management, new quality, and ecology), which are automatically retrieved from the database by the business service module. Upon receiving the input, the model service module performs a format consistency check on the data and returns the comprehensive performance index output by the model to the business module, while simultaneously writing it to the evaluation results database.

[0106] (3) Model update and replacement mechanism

[0107] This invention supports periodic retraining of the evaluation model. The platform can select new samples quarterly or annually to construct new model files. After the model replacement is completed, the model service module automatically loads the new model file upon the next system startup, enabling the model to adapt to changes in climate change, management systems, and planting structures.

[0108] GIS visualization and spatial rendering technology

[0109] (1) GIS platform and spatial data management

[0110] This invention employs ArcGIS Engine or SuperMap iServer as the core of spatial computing in the background to perform operations such as projection transformation, spatial clipping, raster calculation, and buffer generation. Raster data is managed in GeoTIFF format and published as a tileable map service, while vector data is managed in Shapefile or GeoJSON format. The platform improves the access speed of large-area data by establishing spatial indexes, achieving efficient map rendering and spatial querying.

[0111] (2) Front-end visual presentation

[0112] The platform's front end uses a map rendering engine (Leaflet or MapboxGL) to load base maps and thematic layers, and combines a chart library (such as ECharts) to display indicator analysis graphs such as line charts, bar charts, and radar charts. A linkage mechanism is used between the map and the charts; when a user selects any area on the map, the system will simultaneously display its time-series changes and indicator composition information to enhance analytical capabilities.

[0113] (3) Heatmap and isopleth rendering

[0114] For continuous indicators such as water consumption per unit area and water productivity, this invention uses spatial interpolation or kernel density estimation methods to generate heat maps, displaying the spatial gradient distribution of these indicators within the irrigation area. For categorical indicators such as comprehensive performance level or water resource utilization level, this invention generates graded zoning maps or contour maps through raster reclassification and contour line extraction. Different levels employ a graded color scheme to facilitate the identification of performance differences.

[0115] (4) Time series and dynamic evolution display

[0116] The platform features a timeline control, allowing users to query evaluation results for different years, quarters, or months. The system automatically triggers a layer refresh mechanism after monitoring data or model results are updated, enabling dynamic monitoring of the irrigation district's evolution across different time scales. Users can view historical changes through the timeline, achieving joint time-space analysis.

[0117] (5) Zoning diagnosis and decision support functions

[0118] This module supports zoning statistical analysis based on management areas, crop areas, water source areas, or irrigation zones, outputting indicators such as average performance, fluctuation coefficient, and contribution. The platform supports generating spatial heat maps and identifying problem areas, and can output decision analysis reports, providing a scientific basis for water-saving scheduling, facility maintenance, and technology promotion in irrigation districts.

[0119] Ultimately, the model results are presented in a spatial raster format on the integrated interface of the digital twin irrigation district platform. Combined with the timeline control, multi-period overlay comparisons and trend playback can be achieved. Users can load, adjust the transparency, and overlay different indicator layers through the layer management function. They can also call the analysis module to generate customized reports and performance radar charts, realizing full-process visualization from indicator quantification to spatial decision-making. This module has the advantages of strong real-time data, high spatial interpretability, convenient interaction, and strong decision support capabilities. It can provide irrigation district management departments with multi-scale, full-time comprehensive information services, directly serving various application scenarios such as water-saving scheduling, technology promotion, performance evaluation, and planning decisions.

[0120] The spatial computing and result display module of this invention realizes the transformation from model calculation results to spatial decision-making information through multi-source data fusion, spatial mapping calculation and multi-dimensional visualization. It constructs a closed-loop system that connects "monitoring-computation-display-decision-making", providing intuitive, scientific and intelligent management support for the efficient utilization of irrigation water resources and the improvement of crop population quality in irrigation areas.

[0121] and Figure 1 Corresponding to the method shown, this invention also discloses a comprehensive evaluation system for efficient water resource utilization and rice population quality in digital twin irrigation districts, used for implementing... Figure 1 The method shown is a comprehensive evaluation method for efficient water resource utilization and rice population quality in digital twin irrigation districts. See the structural diagram below. Figure 4 As shown, it includes a data acquisition and fusion module, an indicator calculation and processing module, a comprehensive model service module, and a visualization and decision support module connected in sequence. The data acquisition and fusion module is used to acquire and fuse multi-source heterogeneous data from the digital twin irrigation district platform to form a basic data layer; The indicator calculation and processing module is used to standardize the basic data according to the preset multi-level comprehensive evaluation indicator system and calculate the scores of each level of indicator. The integrated model service module contains a pre-trained integrated model based on the CatBoost algorithm. It receives the scores of the five primary indicators from the indicator calculation and processing module, calls the model to perform calculations, and returns the comprehensive performance index. The visualization and decision support module, integrated into the digital twin irrigation district platform, is used to perform spatial raster rendering and dynamic visualization of the comprehensive performance index and scores of indicators at all levels.

[0122] Furthermore, the integrated model service module stores and loads CatBoost models in .cbm format files, and supports receiving input data and returning evaluation results via RESTful API or MQTT interface.

[0123] Furthermore, the visualization and decision support module is implemented based on the geographic information system platform, which can generate heat maps, iso-zoning maps, and time series evolution maps of comprehensive performance index, water resource utilization efficiency, and rice population quality improvement level.

[0124] The present invention also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the following: Figure 1 The method shown.

[0125] The present invention also discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements as follows: Figure 1 The method described.

[0126] To further clarify the technical solution of this invention, the following examples of three typical irrigation districts, A, B, and C, illustrate the specific implementation process of the "Evaluation Method for High-Efficiency Utilization of Irrigation District Water Resources and Improvement of Rice Population Quality".

[0127] Indicator data collection: Representative irrigation districts A, B, and C were selected as the research subjects. Data on five primary indicators and their secondary indicators (water conservancy, agronomy, management, soil quality, and ecology) were collected through field measurements, questionnaires, and remote sensing inversion. Data sources included statistical data from irrigation district management departments, field monitoring data, hydrological and meteorological observation data, and remote sensing image inversion results.

[0128] Water conservancy indicators: Canal system structure integrity rate (0.19), well integrity rate (0.24), well utilization rate (0.26), actual irrigated area ratio (0.36), irrigation water volume per unit area (0.46), water consumption per unit area (0.45), irrigation water utilization rate (0.16), crop water productivity (0.26), irrigation water productivity (0.19), water consumption productivity (0.19), drainage rate (0.25), flood drainage compliance rate (0.69), water price (0.40), water fee collection rate (0.26), and internal recycling rate of irrigation area (0.14).

[0129] Agronomic indicators include fertilizer utilization rate (0.04), plant height (0.05), leaf area index (0.06), chlorophyll content (0.08), machine planting rate (0.10), rotary tillage rate (0.10), straw return rate (0.04), agricultural film coverage rate (0.06), yield (0.04), quality (0.04), and root growth (0.06).

[0130] The management indicators include the percentage of standardized construction in the irrigation area (0.03), the degree of standardized management (0.04), the rate of sound management institutions (0.05), the number of employees per unit irrigation area (0.08), the operation, management and maintenance cost per cubic meter of irrigation water (0.10), the degree of informatization construction (0.04), the rate of large-scale and intensive area development (0.06), the water fee collection rate (0.04), and the water rights coverage rate (0.06).

[0131] The new quality indicators include the degree of improvement of digital infrastructure in irrigation districts (0.95), the configuration rate of digital professionals (0.91), the degree of improvement of disaster prevention, mitigation, early warning and monitoring (0.82), the level of intelligence (0.90), the matching rate of digital water measurement facilities (0.93), the degree of digital construction of pumping stations (0.90), and the coverage rate of digital control (0.60).

[0132] Ecological indicators include irrigation water quality compliance rate (0.03), groundwater level compliance rate (0.04), changes in groundwater over-extraction area (0.04), pesticide use rate (0.05), fertilizer use rate (0.07), energy conservation level (0.07), greening rate (0.02), soil erosion rate (0.04), proportion of green and high-quality agricultural products (0.03), and water resource utilization rate (0.03).

[0133] All raw data are normalized and positiveized to form a standardized index matrix.

[0134] Secondary indicator aggregation and weight calculation (AHP method)

[0135] Under each primary indicator system, the weights of the included secondary indicators are calculated using the Analytic Hierarchy Process (AHP) to determine the importance of each indicator within the system.

[0136] Taking "water conservancy indicators" as an example, it includes 14 secondary indicators, including the integrity rate of canal system structures, the integrity rate of wells, the utilization rate of wells, the ratio of actual irrigated area, the irrigation water volume per unit area, the water consumption per unit area, the irrigation water utilization rate, the crop water productivity, the irrigation water productivity, the water consumption productivity, the drainage rate, the flood drainage compliance rate, the water price, and the water fee collection rate.

[0137] Due to the large dimension of the complete matrix (14×14), for ease of explanation, we will select the first six representative secondary indicators—canal structure integrity rate, well integrity rate, well utilization rate, actual irrigated area ratio, irrigation water volume per unit area, and water consumption per unit area—to construct a judgment matrix (see Table 4) for demonstration. Other secondary indicators are calculated using the same method.

[0138] Table 4 Judgment Matrix

[0139] The weight vector is obtained after standardization calculation: W (6) 1 = [0.36, 0.23, 0.16, 0.11, 0.08, 0.06]; The consistency test results are as follows: λ max =6.24, C . I =0.048, C . R =0.043 < 0.1; This indicates that the judgment matrix has good consistency and the weight allocation is effective. For the remaining eight secondary indicators under the water resources index, the same method is used to construct the judgment matrix and calculate the weights, ultimately obtaining the complete W. (14) 1. Vector. After aggregation, a comprehensive score of the primary water conservancy indicators is formed, which serves as one of the input features of the upper-level CatBoost integrated model.

[0140] Comprehensive calculation and model integration of primary indicators (CatBoost model): Using the scores of five primary indicators obtained from the aggregation of secondary indicators as input features, a comprehensive ensemble model is constructed using the CatBoost gradient boosting decision tree model. The model input includes five variables: water resources (W1), agronomy (W2), management (W3), quality (W4), and ecology (W5), and the output is a comprehensive performance index. S total .

[0141] The objective function of the model is defined as: ; in, F t (x) = F t-1 (x) + t (x) Iterative optimization is used to capture the nonlinear relationships between indicators.

[0142] During training, the model underwent 200 iterations with a learning rate of 0.05 and a loss function of least squares error. The training set comprised 70% of the model, and the validation set comprised 30%. The final comprehensive performance scores for each irrigation district are as follows: Irrigation District A: 0.589; Irrigation District B: 0.247; Irrigation District C: 0.208 The results showed that Irrigation District A had the best overall performance in terms of water resource utilization efficiency and improvement of rice crop quality.

[0143] Model result interpretation and platform integration implementation: To enhance the interpretability of the model, the SHAP (Shapley Additive Explanations) method was used to analyze the contribution rate of the indicators. The results show that the combined contribution rate of new quality and water conservancy indicators is approximately 72%, playing a leading role in the final performance improvement.

[0144] The trained CatBoost model is embedded in the digital twin irrigation district platform in .cbm file format, realizing the process of "offline training - online inference - quarterly retraining - visualization". The system can generate thematic maps such as "water resource utilization efficiency", "rice population quality improvement level", and "comprehensive performance index", realizing spatial dynamic visualization and decision support at the irrigation district scale.

[0145] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for efficient water resource utilization and comprehensive evaluation of rice population quality in digital twin irrigation districts, characterized in that, Includes the following steps: S1 Steps for constructing a multi-level comprehensive evaluation index system: Based on the concept of coupling water-crops-management-technology-ecosystem, establish an index system that includes a target layer, a primary index layer, a secondary index layer, and a basic data layer. The primary index layer includes water conservancy indicators, agronomic indicators, management indicators, new quality indicators, and ecological indicators. S2 Acquisition and Processing of Basic Data Steps: Based on the basic data layer, collect and integrate multi-source heterogeneous data from the irrigation area, including ground monitoring data, remote sensing inversion data, statistical and management data, and IoT terminal data. Standardize the raw data corresponding to each secondary indicator to achieve dimensionless and directional consistency. S3 Steps for Calculating the Score of Primary Indicators: For each primary indicator, the weight of each secondary indicator under it is determined by the analytic hierarchy process (AHP). Based on the standardized data and weights, the scores of the five primary indicators of water conservancy, agronomy, management, new quality and ecology are calculated by weighting. S4 Steps for constructing and applying the integrated model: Construct an integrated model based on the CatBoost gradient boosting decision tree algorithm, using the scores of the five primary indicators calculated in S3 as input features and the comprehensive performance index of the irrigation district as the output target, train the model, use the trained model to comprehensively evaluate the target irrigation district, and output the comprehensive performance index. S5 Results Visualization and Decision Support Steps: Integrate the evaluation results obtained in S4 into the digital twin irrigation district platform. Through the spatial computing and visualization module, generate and display the spatial distribution map and dynamic evolution information of the irrigation district's efficient water resource utilization and rice population quality improvement level.

2. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 1, characterized in that, The secondary indicators of water conservancy indicators in S1 include the integrity rate of canal structures, irrigation water utilization rate, and crop water productivity; the secondary indicators of agronomic indicators include leaf area index, yield, and fertilizer utilization rate; the secondary indicators of management indicators include the degree of standardized management and water fee collection rate; the secondary indicators of new quality indicators include the degree of digital infrastructure improvement and intelligence level; and the secondary indicators of ecological indicators include the irrigation water quality compliance rate, fertilizer use rate, and water resource utilization rate.

3. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 2, characterized in that, Ground monitoring data were collected using ultrasonic flow meters, pressure water level gauges, and FDR / TDR soil moisture monitors, with a sampling frequency of 10-30 minutes. Remote sensing inversion data were obtained using Sentinel-2 MSI and Landsat-8 / 9 OLI–TIRS satellite imagery. Evapotranspiration was retrieved using the SEBAL energy balance model, and leaf area index was obtained using the NDVI-LAI empirical model.

4. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 3, characterized in that, In S3, quantitative positive indicators are normalized using the extreme value method, and negative indicators are normalized using the reverse method. Qualitative indicators are quantified into 0-1 range values ​​using expert scoring or fuzzy membership functions.

5. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 4, characterized in that, S4 specifically includes: S401 Constructing the training sample set: ， in, x i =[ W 1i , W 2i , W 3i , W 4i , W 5i ] indicates the first i The score vectors of the five primary indicators for each sample. y i The corresponding known comprehensive performance index; the CatBoost model in the th... t The prediction function for the round of iterations is: ； in, F t-1 ( x () represents the previous round of predictions. For learning rate, h t ( x ) represents the current round's weak learner, i.e., a symmetric decision tree; S402 initializes the CatBoost model parameters, including the learning rate, number of iterations, and tree depth; S403 optimizes the objective function L of the CatBoost model through iterative training. L The least squares error function is: ， in, For the first t Model predictions after rounds of iterations; S404 inputs the scores of the five primary indicators of the irrigation district to be evaluated into the trained CatBoost model to obtain the comprehensive performance index of the irrigation district to be evaluated.

6. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 5, characterized in that, S4 also includes: using the SHAP method to interpret the trained CatBoost model and quantifying the contribution of five primary indicators—water conservancy, agronomy, management, new quality, and ecology—to the comprehensive performance index.

7. The method for efficient water resource utilization and comprehensive evaluation of rice population quality in a digital twin irrigation district according to claim 6, characterized in that, S4 uses CGCS2000 or WGS84 coordinate systems for spatial registration, performs spatial calculations through ArcGIS Engine or SuperMapiServer, and uses Leaflet or MapboxGL for visualization on the front end.

8. A comprehensive evaluation system for efficient water resource utilization and rice population quality in digital twin irrigation districts, characterized in that: The method for implementing the comprehensive evaluation of water resource efficiency and rice population quality in a digital twin irrigation district as described in any one of claims 1-7 includes a data acquisition and fusion module, an index calculation and processing module, a comprehensive model service module, and a visualization and decision support module connected in sequence. The data acquisition and fusion module is used to acquire and fuse multi-source heterogeneous data from the digital twin irrigation district platform to form a basic data layer; The indicator calculation and processing module is used to standardize the basic data according to the preset multi-level comprehensive evaluation indicator system and calculate the scores of each level of indicator. The integrated model service module contains a pre-trained integrated model based on the CatBoost algorithm. It receives the scores of the five primary indicators from the indicator calculation and processing module, calls the model to perform calculations, and returns the comprehensive performance index. The visualization and decision support module, integrated into the digital twin irrigation district platform, is used to perform spatial raster rendering and dynamic visualization of the comprehensive performance index and scores of indicators at all levels.

9. A comprehensive evaluation system for efficient water resource utilization and rice population quality in a digital twin irrigation district, as described in claim 8, is characterized in that... The integrated model service module stores and loads CatBoost models in .cbm format files, and supports receiving input data and returning evaluation results via RESTful API or MQTT interface.

10. A comprehensive evaluation system for efficient water resource utilization and rice population quality in a digital twin irrigation district, as described in claim 9, is characterized in that... The visualization and decision support module is based on a geographic information system platform and can generate heat maps, iso-zoning maps, and time series evolution maps of comprehensive performance index, water resource utilization efficiency, and rice population quality improvement level.

Images