High-standard farmland intelligent construction effect evaluation system and method and electronic equipment

Abstract

The invention provides a high-standard farmland intelligent construction effect evaluation system and method and electronic equipment, and the system comprises a plurality of modules which are distributed at a terminal sensing layer, an edge calculation layer and a cloud service layer. The data acquisition module comprises a data acquisition terminal, Internet of Things sensing terminals and an intelligent execution terminal, wherein the Internet of Things sensing terminals and the intelligent execution terminal are arranged in blocks of a to-be-evaluated high-standard farmland; the data preprocessing module is arranged on an edge computing layer and comprises an edge computing gateway; and the weight dynamic calculation module, the evaluation analysis module and the result output and feedback module are arranged on the cloud service layer and are used for realizing calculation, data storage, model training and evaluation service release. The invention further provides an evaluation method using the evaluation system. According to the technical scheme, the method is high in practicability and wide in adaptability, can effectively solve key problems in current high-standard farmland intelligent construction evaluation, and has good social benefits and industrial popularization potential.

Description

Technical Field

[0001] This invention belongs to the field of smart agriculture technology, specifically relating to an evaluation system, method, and electronic equipment for the effectiveness of intelligent construction of high-standard farmland. Background Technology

[0002] According to the terminology explanation of the Chinese national standard (GB / T33130—2024), high-standard farmland refers to arable land that is flat, contiguous, well-equipped, water-saving and efficient, equipped with agricultural electricity, suitable for cultivation, fertile soil, ecologically friendly, and highly resilient to disasters, and is adapted to modern agricultural production and management methods, ensuring stable and high yields regardless of drought or flood. The construction of high-standard farmland is a core measure to ensure food security, and intelligent upgrading (smart agriculture construction) has become a key direction for improving the quality and efficiency of high-standard farmland. Its core lies in achieving precise perception, intelligent decision-making, and efficient execution of farmland production through technologies such as the Internet of Things, big data, and artificial intelligence. High-standard farmland construction accounts for more than half of the arable land area and plays a vital supporting role in agricultural production. Practice has shown that the construction of high-standard farmland has not only significantly improved farmland infrastructure conditions and enhanced soil fertility, providing important support for food supply, but also strengthened the farmland's disaster resistance and mitigation capabilities, promoted the economical and intensive use of agricultural resources, driven the transformation and upgrading of agricultural production, and helped farmers increase their income.

[0003] Currently, the evaluation of the effectiveness of high-standard farmland construction largely focuses on traditional dimensions such as land consolidation and water conservancy facilities, while the evaluation system for the achievements of smart agriculture construction has many shortcomings and defects. For example, existing evaluation methods suffer from problems such as: vague and general indicator systems, rigid and singular evaluation methods, and a disconnect between evaluation and application. Existing evaluation methods cannot reflect the actual effectiveness of intelligent facilities, relying mainly on manual verification and static data statistics, and cannot provide guidance for the iterative upgrading of intelligent systems.

[0004] Therefore, there is an urgent need to construct a high-standard farmland intelligent construction effectiveness evaluation system with clear indicators, scientific methods, and dynamic adjustment capabilities to address the shortcomings of existing technologies. Summary of the Invention

[0005] To alleviate or solve the shortcomings and problems of existing technologies, this invention provides a system and method for evaluating the effectiveness of intelligent construction of high-standard farmland.

[0006] According to one aspect of the present invention, a high-standard farmland intelligent construction effectiveness evaluation system is proposed. The evaluation system may include multiple modules distributed across a terminal sensing layer, an edge computing layer, and a cloud service layer: a data acquisition module, which may be distributed in the terminal sensing layer and may include data acquisition terminals, various IoT sensing terminals deployed in different blocks of the high-standard farmland to be evaluated, and intelligent execution terminals, capable of collecting soil data, meteorological data, crop growth data, and execution layer operation data; a data preprocessing module, which may be deployed in the edge computing layer and may include an edge computing gateway, capable of preprocessing, cleaning, and standardizing data from the data acquisition module; and a weight dynamic calculation module, an evaluation analysis module, and a result output and feedback module deployed in the cloud service layer, capable of realizing computation, data storage, model training, and evaluation service publishing; wherein, the data acquisition module can collect hardware information, installation location, and runtime information of each IoT sensing terminal; the data acquisition module may also be equipped with a manual data entry terminal and a port for obtaining data from an external big data platform.

[0007] In an evaluation system according to a specific embodiment of this application, the evaluation analysis module constructs a multi-model fusion evaluation and dynamic correction system, specifically including: performing a first comprehensive calculation on the scores of each evaluation index after dynamic correction based on a fuzzy comprehensive evaluation model to reduce the impact of fluctuations in a single index on the evaluation results; performing nonlinear correction on the first comprehensive calculation results based on a neural network model to learn the mapping relationship between historical evaluation results and actual construction effectiveness; and generating the final comprehensive evaluation score by weighted fusion of the output results of different models.

[0008] According to another aspect of this application, an electronic device is also proposed, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the program, implements the functions of the high-standard farmland intelligent construction effectiveness evaluation system as described above. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings: Figure 1 The diagram shows a functional module schematic of a specific embodiment of the intelligent construction effectiveness evaluation system for high-standard farmland according to the present invention; Figure 2 This is a schematic diagram of a specific implementation of a method for evaluating the effectiveness of intelligent construction of high-standard farmland according to the present invention; Figure 3 This is a schematic diagram illustrating another specific embodiment of the method for evaluating the effectiveness of intelligent construction of high-standard farmland according to the present invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0011] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0012] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0013] It should also be noted that, unless otherwise specified, the term "connection" in this document may refer not only to a direct connection but also to an indirect connection with an intermediary. The wording related to serial numbers and designations in this disclosure, such as "first," "second," "S100," "S200," and similar expressions, are used only for distinguishing purposes in the description and should not be construed as indicating or implying relative importance or implicitly indicating the number or necessary order of the indicated technical features.

[0014] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0015] According to one aspect of the present invention, a high-standard farmland intelligent construction effectiveness evaluation system is proposed, the evaluation system may include multiple modules distributed in the terminal sensing layer, edge computing layer and cloud service layer: The data acquisition module can be distributed in the terminal sensing layer. The data acquisition module can include data acquisition terminals, IoT sensing terminals and intelligent execution terminals arranged in each block of the high-standard farmland to be evaluated, and can be used to collect soil data, meteorological data, crop growth data and execution layer operation data. A data preprocessing module, which can be deployed in the edge computing layer, may include an edge computing gateway, which is capable of preprocessing, cleaning and standardizing data from the data acquisition module. The weight dynamic calculation module, evaluation analysis module, and result output and feedback module, which are deployed in the cloud service layer, can be used to realize calculation, data storage, model training and evaluation service publishing. The data acquisition module can collect information such as hardware information, installation location, and runtime of each IoT sensing terminal; the data acquisition module can also be equipped with a manual data entry terminal and a docking port for obtaining data from an external big data platform.

[0016] Preferably, the data acquisition module can collect farmland sensing data and execution equipment operation data in real time; the edge computing gateway of the data preprocessing module performs protocol conversion and local caching between different terminals; the manual data entry terminal can receive the entered expert scoring data and farmer ledger data; the docking port can be used to connect to and obtain data from agricultural and rural big data platforms and GIS platforms.

[0017] Preferably, in an evaluation system according to a specific embodiment of this application, the data preprocessing module can also be configured to perform real-time indicator calculation and processing and localized emergency evaluation processing. In the real-time indicator calculation and processing, the data preprocessing module can monitor the real-time performance indicators of the online rate and command response timeliness of each IoT sensing terminal and intelligent execution terminal, triggering an early warning when the real-time indicator falls below a predetermined threshold. In the localized emergency evaluation processing, the data preprocessing module can perform localized evaluation processing under network outage conditions, generating a preliminary assessment report of emergency effectiveness, and automatically synchronizing the evaluation results to the cloud after network recovery. Preferably, the data cleaning and standardization performed by the data preprocessing module can include: outlier removal, missing value completion, and unit unification and classification quantification.

[0018] Preferably, in an evaluation system according to a specific embodiment of this application, the weight dynamic calculation module can generate subjective weight coefficients based on expert scoring data from the manual data entry terminal, calculate objective weight coefficients based on data from a predetermined time period, optimize the fusion coefficients through a genetic algorithm, match dynamic correction coefficients on a predetermined periodic day to complete the correction of indicator scores, calculate the comprehensive score through multi-model fusion, and determine the level according to the rules.

[0019] In an evaluation system according to a specific embodiment of this application, the evaluation analysis module can match basic correction coefficients and specific correction coefficients based on crop growth period, environmental conditions, and technological maturity to dynamically correct indicator scores; generate preliminary results through fuzzy comprehensive evaluation, and generate a comprehensive score by hierarchical weighted summation after neural network correction; automatically check the threshold of first-level indicators and core veto items, and determine the final evaluation level according to the rule of "threshold priority, irreversible downgrade".

[0020] Preferably, in the evaluation and analysis module, dynamic influencing factors related to crop growth period, environmental conditions, and technological maturity can be introduced to calibrate the original scores of the indicators; at the same time, specific correction coefficients related to equipment type, crop type, and disaster level can also be introduced to achieve dynamic adaptation to different production scenarios, technological stages, and regional conditions.

[0021] In an evaluation system according to a specific embodiment of this application, the evaluation analysis module constructs a multi-model fusion evaluation and dynamic correction system, specifically including: performing a first comprehensive calculation on the scores of each evaluation index after dynamic correction based on a fuzzy comprehensive evaluation model to reduce the impact of fluctuations in a single index on the evaluation results; performing nonlinear correction on the first comprehensive calculation results based on a neural network model to learn the mapping relationship between historical evaluation results and actual construction effectiveness; and generating the final comprehensive evaluation score by weighted fusion of the output results of different models.

[0022] Preferably, the evaluation system may further include a system management and security module; the IoT sensing terminal may include: soil moisture / fertility sensors, weather stations, crop growth cameras, etc., supporting communication protocols such as RS485, LoRa, and NB-IoT to meet the data acquisition accuracy requirements of the sensing layer; the intelligent execution terminal may include: intelligent irrigation valves, variable fertilizer applicators, drone plant protection equipment, agricultural machinery Beidou positioning terminals, etc., for collecting execution layer operation command logs, energy consumption data, and collaborative operation parameters; the data acquisition terminal may include: edge acquisition gateways and rugged handheld PADs.

[0023] The following section, in conjunction with the accompanying drawings and specific embodiments, further illustrates the evaluation system and method for the effectiveness of intelligent construction of high-standard farmland according to this application.

[0024] In a specific implementation plan of the intelligent construction evaluation system for high-standard farmland according to this application, a fully automated evaluation system is constructed, with a three-level distributed architecture of "cloud-edge-device" as the core support. This system is tightly coupled with the entire chain of technical frameworks, including "indicator system-data processing-weight calculation-dynamic correction-comprehensive evaluation-optimization feedback." Through the collaborative linkage of the three-level nodes, the system not only solves the pain points of limited network bandwidth and high real-time data requirements in farmland, but also meets the needs of large-scale data storage and complex algorithm calculations, ensuring that the evaluation process is accurate, efficient, and dynamically closed-loop.

[0025] In one specific embodiment of the high-standard farmland intelligent construction effectiveness evaluation system according to the present invention (hereinafter referred to as "the system"), it includes multiple modules distributed across the terminal sensing layer, the edge computing layer, and the cloud service layer. See also Figure 1The diagram shown illustrates the functional modules of a specific embodiment of the intelligent construction effectiveness evaluation system for high-standard farmland according to the present invention. The system comprises a terminal sensing layer (edge ​​side), an edge computing layer (side side), and a cloud service layer (cloud side) as its overall architecture.

[0026] Terminal perception layer The terminal perception layer is the source node for system data collection. It corresponds to the core requirements of the multi-source data collection module and is responsible for connecting various intelligent devices and collection terminals in high-standard farmland to achieve accurate and efficient collection of data from all dimensions.

[0027] The hardware components of the terminal perception layer may include: IoT sensing terminals include, for example, soil moisture / fertility sensors, weather stations, and crop growth cameras, which support communication protocols such as RS485, LoRa, and NB-IoT to meet the data acquisition accuracy requirements of the sensing layer. Intelligent execution terminals: such as intelligent irrigation valves, variable fertilizer applicators, drone plant protection equipment, agricultural machinery Beidou positioning terminals, etc., are used to collect execution layer operation command logs, energy consumption data, and collaborative operation parameters; Data acquisition terminals: For example, they can be industrial-grade edge acquisition gateways (with built-in libraries of more than 10 mainstream agricultural IoT device protocols, supporting multi-protocol parsing), rugged handheld PADs (used for manual supplementation of expert scores, farmer ledgers, and other data), etc.

[0028] The main features of the terminal perception layer are: Real-time device status monitoring: For example, the online status of devices can be monitored based on a heartbeat mechanism. When a device is offline for more than 1 hour, an SMS / platform alarm will be automatically triggered to ensure the continuity of data collection. Multi-source data acquisition: Automatically collects soil, meteorological, and crop growth data from the sensing layer, as well as operation data from the execution layer, while also supporting manual data entry via handheld PAD; Local data preprocessing: Perform preliminary outlier filtering on the collected data, and cache the data locally on the edge gateway for ≥72 hours to avoid data loss due to network interruption; Protocol standardization conversion: Convert the proprietary protocols of different devices into the system standard MQTT protocol to ensure seamless data interoperability with the edge layer and cloud layer.

[0029] Edge computing layer The edge computing layer addresses the localized processing needs of the data preprocessing module, enabling data cleaning, real-time indicator calculation, and emergency evaluation tasks, thereby reducing cloud computing pressure and network transmission costs. It can be deployed in high-standard farmland management sub-centers. For example, the hardware deployment carriers of the edge computing layer can be edge computing gateways and local area network switches. Preferably, the edge computing gateway supports lightweight GPU computing, enabling lightweight data computation for artificial intelligence models.

[0030] The core functions of the edge computing layer mainly include: data cleaning and standardization, outlier removal, missing value completion, unit unification and classification quantification, real-time indicator calculation, and localized emergency evaluation. For example: When removing outliers, the "2σ principle" can be used for equipment operation and maintenance data, the "3σ principle" can be used for perception / benefit data, and the sliding window anomaly detection algorithm can be used for time series data (such as soil moisture). The abnormal data types (equipment failure / transmission interruption / human error) are automatically marked and synchronized to the equipment ledger.

[0031] When imputing missing values, time-series data (soil moisture, energy consumption) can be imputed using, for example, an LSTM neural network model or other similar neural network models (imputed error ≤ 5%), while non-time-series data (expert scores, farmer ledgers) can be filled with the mean of the same type of indicator. The imputed data is specially marked. When performing dimensional unification and classification quantification, the system automatically identifies indicators as "positive / negative / interval type / exclusive formula type" and calls the corresponding quantification formula to normalize indicators of different dimensions into standard scores of 0-100. Subjective and objective integrated indicators are comprehensively quantified according to preset weights (60%-80% objective data, 20%-40% expert scores). When calculating real-time indicators, real-time indicators such as "device online rate" and "command response timeliness rate" are calculated on a minute-by-minute basis, and local warnings are triggered when they fall below the threshold. During localized emergency assessments, localized assessments are supported even when the network is down. Once the network is restored, the assessment results are automatically synchronized to the cloud. In extreme scenarios, a preliminary assessment report on the effectiveness of the emergency can be generated within 10 minutes.

[0032] Cloud service layer (cloud side) The cloud service layer serves as the "core computing and service node" of this system, centrally carrying the core functions of the dynamic weight calculation module, intelligent evaluation and analysis module, result output and feedback module, and system management and security module, and realizing complex calculation, data storage, model training and evaluation service publishing.

[0033] The hardware and software support for cloud service layers can be divided into: Hardware configuration: cloud server cluster (CPU ≥ 64 cores / unit, memory ≥ 256GB / unit, SSD storage ≥ 10TB / unit), GPU computing nodes (support model training acceleration), distributed storage system (ensure secure data storage). Software architecture: Containerized deployment based on Kubernetes, supporting elastic scaling of services; adopts a multi-database architecture of MySQL (structured data) + Redis (cached data) + MongoDB (unstructured data); integrates TensorFlow, PyTorch deep learning framework and Scikit-learn machine learning library.

[0034] The core functions of the cloud service layer are reflected in: full data storage and management, dynamic weight calculation, multi-model fusion evaluation, and evaluation result output and feedback. Preferably, they are also reflected in system management and security protection. For example: When storing and managing all data, the system stores nearly 5 years of full-chain data in partitions according to the "project-plot-time" dimension, supports second-level retrieval of hundreds of millions of data points, and has a data storage period of ≥5 years. When performing dynamic weight calculation, the Delphi method for subjective weight calculation, the entropy weight method for objective weight calculation, and the genetic algorithm are combined to optimize the weight, automatically generate hierarchical weights and display them visually. When conducting multi-model fusion evaluation, the fuzzy comprehensive evaluation and BP neural network correction model are run, and the comprehensive score is calculated and the level is accurately determined by combining dynamic correction coefficients. When outputting and providing feedback on evaluation results, multi-dimensional visualization charts and standardized / customized evaluation reports are generated, tiered optimization suggestions are pushed out, and the progress of rectification is tracked.

[0035] The system management and security protection module provides functions such as fine-grained permission management, encrypted data transmission and storage, disaster recovery backup, and operation and maintenance monitoring to ensure the stable and secure operation of the system.

[0036] This system is based on a closed-loop process of "data acquisition-processing-evaluation-feedback," and preferably includes six core functional modules: data acquisition module, data preprocessing module, dynamic weight calculation module, evaluation and analysis module, result output and feedback module, and system management and security module. Each module is functionally independent yet collaboratively linked. The following descriptions exemplify the specific implementation of each module. It is understood that the neural network models, data processing methods, quantization processes, and weight calculation methods described herein are exemplary and can be replaced by methods capable of achieving equivalent or similar functions. Specific equivalent artificial intelligence models or numerical processing methods are also within the scope of protection of this application.

[0037] 1. Data Acquisition Module This module serves as the data input entry point for the multi-source system, enabling the fusion of data from multiple channels through "automatic terminal acquisition + platform interface integration + manual data entry," thus ensuring the integrity and timeliness of the evaluation data.

[0038] Sub-function 1: Device access management Supports one-click registration and protocol adaptation for smart devices, and has a built-in protocol library for more than 10 mainstream agricultural IoT devices, making it compatible with smart devices of different brands and types. Visualized equipment ledger management displays information such as equipment model, installation location, runtime, and fault records, and supports filtering and querying by plot and equipment type; Monitor device communication status in real time, and automatically trigger SMS / platform alarms when the device is offline for more than 1 hour to ensure uninterrupted data collection.

[0039] Sub-function 2: Platform data integration It has pre-installed interfaces with third-party platforms such as agricultural and rural big data platforms, GIS geographic information platforms, and smart agriculture decision-making systems to enable rapid data integration. Automatically synchronizes system interface data such as farmland plot information, historical yield data, and decision model operation logs, and supports configuring the synchronization frequency according to time period (5min / 1h / 1d); It supports integration with the financial system to synchronize data on benefits such as investment in intelligent farmland construction and farmers' income. After synchronization, it automatically verifies the data format and integrity, and automatically marks abnormal data and initiates retransmission.

[0040] Sub-function 3: Manual data entry Expert scoring subsystem: Supports back-to-back scoring by expert groups of 5 or more people. The system automatically removes outlier scores that deviate from the mean by more than 2 standard deviations and calculates the mean. The scoring progress and statistical results are displayed in real time. Farmer ledger entry: Provides structured forms to support data entry and batch import of labor input, pesticide and fertilizer usage, financial input, etc. Data audit function: Performs format verification and logical audit on manually entered data (such as alerts for abnormal input-output ratio values) to ensure the accuracy of manually entered data.

[0041] 2. Data Preprocessing Module This module is responsible for standardizing the collected raw data, providing a high-quality data foundation for subsequent evaluation calculations.

[0042] Sub-function 1: Exception Data Handling Multi-strategy outlier removal: Differentiated anomaly identification strategies are adopted for different types of data. The "2σ principle" is used for equipment operation and maintenance data, the "3σ principle" is used for perception / benefit data, and the sliding window anomaly detection algorithm is used for time series data (such as soil moisture) to ensure the accuracy of anomaly identification. Abnormal cause recording: Automatically marks abnormal data types (equipment failure / transmission interruption / human error) and synchronizes them to the equipment ledger, providing a basis for subsequent equipment maintenance and data traceability.

[0043] Sub-function 2: Missing value completion Time series data completion: Based on the LSTM neural network model, time series data such as soil moisture and energy consumption are completed with a completion error of ≤5%, ensuring the continuity of time series data; Non-time series data completion: Non-time series data such as expert scores and farmer ledgers are filled with the average of the same type of indicators to avoid the impact of missing single data on the overall evaluation; Complete data labeling: Special labels are applied to complete data to clearly distinguish between original collected data and complete data, ensuring data traceability.

[0044] Sub-function 3: Data standardization and classification quantification Indicator type adaptation: Automatically identifies indicators as "positive / negative / range / dedicated formula type" and calls the corresponding quantitative formula to ensure that the quantitative logic matches the indicator characteristics; Dimensional unification: Different dimensional indicators (such as MTBF duration and energy consumption reduction rate) are normalized to a standard score of 0-100 points to eliminate the impact of dimensional differences on the evaluation results; Subjective and objective data fusion: For subjective and objective fusion indicators, a comprehensive quantification is completed according to preset weights (60%-80% objective data, 20%-40% expert scores), taking into account both data objectivity and expert experience value.

[0045] 3. Dynamic Weight Calculation Module This module serves as the "weight core engine" of the evaluation system, enabling automatic calculation of subjective and objective weights and intelligent optimization of combined weights, ensuring the scientific nature and dynamic adaptability of the weights.

[0046] Sub-function 1: Subjective weight calculation (Delphi method) Expert Management: Supports expert information entry, professional field classification (agricultural engineering, AI algorithms, equipment operation and maintenance, etc.), and expert scoring permission allocation to ensure full professional coverage of the expert team; Multi-round scoring management: Automatically initiates a three-round Delphi method scoring process, and synchronously provides experts with the statistical results of the group scoring (mean, standard deviation, coefficient of variation), which facilitates the experts to revise their scoring opinions; Subjective weight generation: Based on the final scores from experts, complete the normalization calculation of subjective weight levels for first-level, second-level, and third-level indicators to ensure that the weights at the same level sum to 1, and generate a weight calculation report containing detailed scores.

[0047] Sub-function 2: Objective weight calculation (entropy weight method) Data dispersion analysis: Automatically extract 6 sets of monthly data samples from the past year, construct a data matrix and complete the standardization process, calculate the information entropy and difference coefficient of each indicator, and the larger the difference coefficient, the stronger the indicator's discriminative power; Objective weight generation: Based on the difference coefficient, the objective weight of the indicator is transferred and normalized hierarchically. It supports automatic quarterly updates of objective weights to ensure that the weights change in sync with the actual characteristics of the data. Weight rationality verification: Automatically verify whether the sum of the weights of indicators at the same level is 1. If an anomaly is found, a recalculation is triggered to ensure the consistency of the weight logic.

[0048] Sub-function 3: Combinatorial weight optimization (genetic algorithm) Fusion coefficient optimization: Using "the goodness of fit between the evaluation score and the actual performance reference value" as the objective function, the subjective-objective weight fusion coefficient α is optimized through a genetic algorithm to improve the scientific nature of the weights; Combined weight generation: according to the formula The final combined weights are generated, taking into account both expert experience and the objective characteristics of the data. Weight visualization: Display the weight percentage of each indicator in a hierarchical pie chart / bar chart, support comparison of historical weight change trends, and intuitively present the dynamic adjustment of weights.

[0049] 4. Intelligent evaluation and analysis module This module is the "core evaluation module" of the system, which realizes dynamic evaluation and accurate level determination through multi-model fusion, ensuring the accuracy and rationality of the evaluation results.

[0050] Sub-function 1: Calculation of dynamic correction coefficient Automatic matching of basic correction coefficients: The basic correction coefficients are automatically matched based on crop growth stage (e.g., wheat grain-filling stage), environmental conditions (e.g., drought / flood), and technological maturity (e.g., new technology adaptation period). It is adaptable to different production scenarios and technology application stages; Dedicated correction coefficients for precise adaptation: Automatically matched based on equipment brand, crop type, disaster level, regional economic level, etc. , , , Specific coefficients are added to improve the scenario adaptability of the evaluation results; Corrected score calculation: according to the formula Complete the indicator score correction, and limit the score to the range of 0-100 points after correction to ensure the reasonableness of the score.

[0051] Sub-function 2: Multi-model fusion evaluation Overall Score: A weighted sum of the corrected scores of the three-level indicators multiplied by the combined weights of the three levels. Primary indicator score: a weighted sum of scores adjusted by the subordinate tertiary indicators. Sub-function 3: Level determination and veto item verification Candidate grade matching: Based on the overall score, candidates are initially matched for excellent / good / pass / fail (excellent ≥ 90 points, good 80-89 points, pass 60-79 points, and fail < 60 points). Multi-level threshold verification: Automatically check the scoring threshold of primary indicators (e.g., all primary indicators must be ≥85 points for the excellent level) and core veto items (e.g., perception coverage <90% triggers the excellent level veto) to ensure the rigor of the level determination. Final grade determination: The final grade is generated according to the rule of "threshold priority and irreversible downgrade", and a grade determination report is generated at the same time (including the threshold meeting status and the triggering of veto items) to ensure that the determination results are traceable.

[0052] 5. Results Output and Feedback Module This module is responsible for displaying evaluation results in multiple formats, generating reports, and pushing optimization suggestions, thus realizing a closed loop of "evaluation-feedback-optimization" and enhancing the application value of evaluation results.

[0053] Sub-function 1: Visualization of evaluation results Multi-dimensional chart display: Hierarchical Score Radar Chart: Displays the score distribution of 4 primary indicators and 19 secondary indicators, intuitively presenting the weak links in intelligent construction; Indicator weight distribution chart: The weight distribution of the core three-level indicators is displayed in a pie chart, clearly showing the degree of influence of each indicator on the evaluation results; Historical trend comparison chart: Supports comparison of evaluation scores and grade changes across different quarters / years, tracking changes in the effectiveness of intelligent construction; Spatial distribution heat map: Based on the GIS platform, it displays the distribution of intelligent effectiveness scores of different plots, realizing regional difference analysis; Multi-terminal adaptation: Supports visualization display on web, mobile (APP / mini-program) and large screen terminals (farmland management center), with real-time data synchronization to meet the viewing needs of different scenarios.

[0054] Sub-function 2: Intelligent Report Generation Automatic generation of standardized reports: Supports the generation of "Evaluation Report on the Effectiveness of Intelligent Construction of High-Standard Farmland". The report includes modules such as evaluation overview, detailed indicator scores, grade determination results, analysis of weak links, and optimization suggestions, meeting the needs of routine evaluation. Customized report output: Supports custom report templates according to user needs (such as acceptance reports, annual assessments), and can be exported in PDF / Word / Excel formats to adapt to different application scenarios; Report signing and archiving: Supports electronic signing function, and the generated reports are automatically archived to the cloud database. It supports retrieval by project number and time, which facilitates report management and traceability.

[0055] Sub-function 3: Pushing optimization suggestions Tiered suggestion generation: Based on the "score range - indicator type - optimization priority" mechanism, it automatically generates P1 (urgent), P2 (important) and P3 (general) level optimization suggestions, which are highly targeted; Multi-channel push: Supports pushing optimization suggestions to farmland management managers and technical maintenance personnel via SMS, platform messages, email, etc., to ensure that suggestions reach them in a timely manner; Rectification tracking: Supports the input of rectification measures and progress, enabling closed-loop tracking and effectiveness review of optimization suggestions, and promoting continuous optimization of intelligent construction.

[0056] 6. System Management and Security Module This module ensures stable system operation and data security, enabling full-process access control, security protection, and operation and maintenance monitoring to ensure reliable system operation.

[0057] Sub-function 1: User permission management Multi-role permission allocation: Set roles such as super administrator, farmland administrator, expert user, and operation and maintenance personnel, and refine permissions such as data viewing, scoring and evaluation, report export, and system configuration to achieve fine-grained permission control; Operation log recording: Record all user operations (such as data entry, weight modification, report export). The log includes the operator, operation time, and operation content, and is retained for ≥3 years to ensure operation traceability.

[0058] Sub-function 2: Data security protection Data encryption: Transmitted data is encrypted using the TLS 1.3 protocol, and stored data is encrypted using the AES-256 encryption algorithm to ensure data transmission and storage security; Access control: Platform access is restricted based on IP whitelists, and critical operations (such as weight modification) require dual authorization to prevent unauthorized access and operation risks; Disaster recovery backup: Cloud data supports daily incremental backup and weekly full backup. Backup data is stored on off-site servers and supports one-click recovery to ensure no data loss.

[0059] Sub-function 3: System operation and maintenance monitoring Equipment status monitoring: Real-time monitoring of terminal equipment online rate and communication status, automatic alarm when equipment is abnormal, facilitating timely troubleshooting; System performance monitoring: Monitors cloud server CPU, memory, and storage usage, edge gateway operating status, and triggers operation and maintenance alarms when anomalies occur to ensure system performance stability; Model iteration management: Supports online training and version updates for BP neural networks and LSTM completed models, retains historical model versions for backtracking, and ensures continuous optimization of model adaptability.

[0060] See Figure 2 As shown, Figure 2 This is a schematic diagram illustrating a specific embodiment of the method for evaluating the effectiveness of intelligent construction of high-standard farmland according to the present invention. In another aspect of this application, a method for evaluating the effectiveness of intelligent construction of high-standard farmland is proposed, employing the evaluation system described above. In one specific embodiment, the evaluation method includes the following steps: S100: Set the evaluation time period, evaluation indicators, and set the evaluation start date; S200: Data acquisition steps: Real-time acquisition of farmland sensing data and execution equipment operation data; use edge gateway to complete protocol conversion and local caching; acquisition of data such as expert scores and farmer ledgers; connection to third-party platforms to synchronously obtain data from agricultural and rural big data platforms and GIS platforms; all data are filtered by the edge layer and then synchronized to the cloud to ensure data integrity and timeliness.

[0061] S300: Data preprocessing steps: The cloud-based data preprocessing module performs multi-strategy outlier removal on the received data, fills in missing values ​​according to rules, automatically identifies the indicator type and completes standardization and classification quantification, and generates an initial score of 0-100 for the indicator, providing a high-quality data foundation for subsequent evaluation calculations; S400: Dynamic weight calculation steps: The weight calculation module automatically calls the Delphi method expert scoring data, and after at least two rounds of evaluation, selects the best three rounds of evaluation to generate subjective weights; based on a predetermined number of monthly data samples, the objective weights are calculated using the entropy weight method; with the fitting degree between the evaluation score and the actual performance reference value as the target, the fusion coefficient is optimized through a genetic algorithm to generate the final combined weights; S500: Dynamic evaluation steps: The evaluation and analysis module automatically matches the basic correction coefficient and the specific correction coefficient based on crop growth period, environmental conditions, and technological maturity to complete the dynamic correction of the indicator score; it generates preliminary results through fuzzy comprehensive evaluation, and after being corrected by neural network, it generates a comprehensive score according to the hierarchical weighted summation formula; it automatically checks the threshold of the first-level indicator and the core veto item, and determines the final evaluation level according to the rule of "threshold priority, irreversible downgrade". S600: Results Feedback Steps: The results output module generates multi-dimensional visualization charts and standardized / customized evaluation reports, and generates hierarchical optimization suggestions based on the evaluation results.

[0062] This system employs a fully automated processing logic encompassing "data acquisition - preprocessing - weight calculation - dynamic evaluation - result feedback," with each step deeply integrated with the functional modules to ensure efficient and accurate evaluation. In a specific embodiment, the steps are as follows: 1. Data Acquisition Phase Edge sensing devices collect farmland sensing data and execution equipment operation data in real time, and the edge gateway completes protocol conversion and local caching; manual input of expert scores, farmer ledgers and other data via handheld PADs; the system connects to third-party platforms (agricultural and rural big data platforms, GIS platforms, etc.) to synchronously obtain relevant data; all data is initially filtered by the edge layer and then synchronized to the cloud to ensure data integrity and timeliness.

[0063] 2. Data Preprocessing Stage The cloud-based data preprocessing module performs outlier removal using multiple strategies on the received data, fills in missing values ​​according to rules, automatically identifies indicator types and completes standardization and classification quantification, and generates initial scores for indicators ranging from 0 to 100, providing a high-quality data foundation for subsequent evaluation calculations.

[0064] 3. Dynamic weight calculation stage The weight calculation module automatically calls the Delphi method expert scoring data and generates subjective weights through three rounds of evaluation; based on six monthly data samples from the past year, objective weights are calculated using the entropy weight method; with the goal of the fit between the evaluation score and the actual performance reference value, the fusion coefficient is optimized through a genetic algorithm to generate the final combined weights, ensuring that the weights are scientific and reasonable.

[0065] 4. Dynamic Evaluation Phase The evaluation and analysis module automatically matches basic correction coefficients and specific correction coefficients based on crop growth period, environmental conditions, and technological maturity to dynamically correct indicator scores; it generates preliminary results through fuzzy comprehensive evaluation, corrects them through a BP neural network, and generates a comprehensive score according to a hierarchical weighted summation formula; it automatically checks the thresholds and core veto items of the first-level indicators and determines the final evaluation level according to the rule of "threshold priority, irreversible downgrade".

[0066] 5. Results Feedback Phase The results output module generates multi-dimensional visualization charts and standardized / customized evaluation reports. Based on the evaluation results, it generates hierarchical optimization suggestions and pushes them to relevant personnel through multiple channels. It also supports the input of rectification progress and effectiveness review, realizing a closed loop of "evaluation-feedback-optimization" and promoting the continuous improvement of the quality and efficiency of intelligent construction.

[0067] The system based on this disclosure also establishes a four-level evaluation architecture: "perception layer—decision layer—execution layer—benefit layer," encompassing 4 primary indicators, 19 secondary indicators, and 26 tertiary indicators. All indicators have clear quantitative standards and data sources. This system covers the entire process from data collection, intelligent decision-making, precise execution to comprehensive benefits, achieving a refined and structured evaluation of the effectiveness of intelligent construction of high-standard farmland.

[0068] Multi-level comprehensive evaluation index system This invention is based on the core elements of smart agriculture construction in high-standard farmland, and constructs a three-level indicator framework according to a full-chain coverage and hierarchical decomposition approach. The first-level indicators anchor key links in intelligent construction, the second-level indicators break down core dimensions, and the third-level indicators are specific, directly quantifiable indicators. There are 4 first-level indicators, 19 second-level indicators, and 26 third-level indicators.

[0069] The primary indicators are the perception layer, decision-making layer, execution layer, and benefit layer.

[0070] In a specific implementation plan, the indicators at each level and their corresponding reference numbers (marked in parentheses) are shown in Table 1 for example: Table 1: Evaluation Index System Since these evaluation indicators are not the focus of this invention, they are described here merely to make the technical solution of this application more complete. Related technical terms can be understood by referring to commonly known technical content by those skilled in the art. Further description and explanation are not provided here.

[0071] The following section uses a specific exemplary implementation plan, taking a smart agriculture demonstration project for high-standard farmland in a certain county as an example, to illustrate the evaluation system and method for the effectiveness of intelligent construction of high-standard farmland. Figure 3 This is a schematic diagram illustrating the evaluation method for the effectiveness of intelligent construction of high-standard farmland in this specific implementation.

[0072] Example 1 In a high-standard farmland smart agriculture demonstration project in a certain county, covering an area of ​​1,500 mu, it is divided into 10 planting blocks (numbered B1-B10, each block is 150 mu) according to the principle of "grid management". The core planting mode is "winter wheat-summer corn" two crops a year (wheat grain filling / harvest period from March to June each year, and corn sowing / jointing period from July to October), which is suitable for the growth characteristics of typical crops in the plains of the province.

[0073] The specific configuration of this high-standard farmland intelligent construction effectiveness evaluation system is as follows: IoT sensing layer: Each block is equipped with 5 soil moisture / fertility sensors (50 in total, monitoring depth 30cm), 1 small weather station (10 in total, monitoring temperature, humidity, precipitation, and wind speed), and 2 crop growth monitoring cameras (20 in total, supporting AI pest and disease identification). The communication protocol is compatible with LoRa / NBIoT. Intelligent execution layer: Each block is equipped with 2 sets of intelligent irrigation valve groups (20 sets in total, supporting precise control of pulse solenoid valves), 1 variable fertilizer applicator (10 in total, with Beidou positioning and metering module), and 1 take-off and landing point for plant protection drones (10 in total, with matching drones). Intelligent decision-making layer: Deploy a localized "farmland brain" platform, integrating irrigation / fertilization decision-making models (trained based on wheat / corn growth period parameters in the province), remote equipment control modules, and data visualization dashboards, supporting API integration with the provincial agricultural and rural big data platform and the county's agricultural and rural affairs bureau's regulatory system.

[0074] The first evaluation of the effectiveness of intelligent construction is now being carried out based on the method and system of this invention. The evaluation period is from March 1 to October 31 of the evaluation year, that is, from the wheat grain filling / harvest period to the end of the corn sowing / jointing period.

[0075] Preferably, the evaluation method for the effectiveness of intelligent construction of high-standard farmland according to this application is as follows: A complete technical framework has been constructed, encompassing "indicator system - data processing - weight calculation - dynamic correction - comprehensive evaluation - optimization feedback," forming a cycle of "evaluation-feedback-optimization." This framework includes six parts: constructing a multi-level comprehensive evaluation indicator system, multi-source data collection and preprocessing, determining indicator weights based on a combined weighting method, dynamic correction of multi-dimensional coefficients, intelligent comprehensive evaluation results, and a three-dimensional optimization suggestion push mechanism.

[0076] 1. Construct a multi-level comprehensive evaluation index system A three-tiered indicator framework is constructed. First-tier indicators anchor key aspects of intelligent construction; second-tier indicators break down core dimensions; and third-tier indicators are specific, directly quantifiable metrics. There are 4 first-tier indicators, 19 second-tier indicators, and 26 third-tier indicators. See Table 1 above for details.

[0077] Based on the three-level indicator system of this invention, combined with the soil characteristics of 10 blocks (B1-B4 are alluvial soil, B5-B10 are sandy ginger black soil) and crop rotation patterns, the quantitative standards and data sources of each indicator are clarified to ensure that the indicators are collectable, verifiable and traceable.

[0078] 2. Multi-source data acquisition and preprocessing 2.1 Data Acquisition Based on the equipment configuration and data requirements for the intelligent construction of high-standard farmland, a multi-source data acquisition mode of "terminal sensing + system interface + manual data entry" is constructed, covering the entire chain of data from sensing, decision-making, execution, and benefits, as detailed below: 2.1.1 Terminal-sensed data (real-time collection of partitioned data): Five soil sensors in each block collect soil moisture / fertility data every 30 minutes; ten weather stations collect temperature, humidity, precipitation, and wind speed data every 15 minutes; and 20 cameras take crop images twice a day at 9:00 AM and 3:00 PM (for AI pest and disease identification). Smart irrigation valve groups, variable fertilizer applicators, and drones upload operation logs (irrigation flow, fertilizer application, and operation trajectory) in real time, with BeiDou positioning data accuracy ≤1 meter. Each block's edge gateway locally caches data for ≥72 hours, automatically storing it when the network is interrupted and uploading it to the cloud in batches after network recovery to ensure data integrity.

[0079] 2.1.2 System Interface Data (Unified Cloud Interface) Connect to the provincial agricultural and rural big data platform where the county is located to obtain meteorological warning data for the month evaluated by the higher-level city (such as high temperature warning in July and rainstorm warning in September) and average wheat / corn growth data for the whole province; connect to the county's GIS platform to obtain geographical data such as soil type and elevation difference (maximum elevation difference ≤ 4 meters) for tracing abnormal data; connect to the county's agricultural and rural affairs bureau's supervision system to obtain data such as project financial investment, traditional planting energy consumption benchmark, and farmer training records to assist in the calculation of benefit-level indicators.

[0080] 2.1.3 Manual data entry (precise data collection based on role) A five-person expert group was formed to score subjective and objective indicators such as "accuracy of multi-crop adaptation" back-to-back (on a scale of 1-10), and the scoring sheets were encrypted and submitted to the system. Each block of growers recorded their labor situation and agricultural product sales revenue, and financial personnel recorded equipment investment (hardware + software) and operation and maintenance costs, which were calculated based on a two-year depreciation period. The expert team conducted on-site verification once a quarter, recorded the equipment operation status and abnormal crop growth, and supplemented the explanation of the reasons for the abnormal data.

[0081] 2.2 Data Preprocessing Standardization processing was carried out on the data characteristics of 26 tertiary indicators to ensure controllable data quality: 2.2.1 Outlier Removal: Equipment Maintenance Data (MTBF): Using the "2σ principle", erroneous data (number of incorrectly recorded faults) of one irrigation valve fault statistics in Block B5 was removed and marked as "human input error". Sensing data (soil fertility): Two sensor calibration deviation data points in block B8 were removed using the "3σ principle" and marked as "equipment calibration not updated in time"; A total of 8 abnormal data entries were removed from 10 blocks, with an abnormality rate of 0.6% (below the acceptable threshold of 1%). The original records were retained for tracing.

[0082] The aforementioned "2σ rule" is a well-known rule of thumb in normal distribution, stating that the probability of a value falling within two standard deviations (σ) of the mean (μ) is approximately 95.44%. This rule is based on the statistical properties of standard deviation: standard deviation measures the dispersion of data; a larger σ indicates more dispersed data, while a smaller σ indicates more concentrated data. Similarly, the 3σ rule covers approximately 99.73% of the data, is more stringent, and is often used in high-precision control scenarios.

[0083] 2.2.2 Missing value completion Time series data: For the missing 2-hour soil moisture data of blocks B1-B4 due to a certain regional network interruption, the data was completed using an LSTM model (completeness error ≤ 5%), and marked as "model complete". Non-time series data: One expert score for "accuracy of multi-crop adaptation" is missing. The score is filled with the average of the scores of the other four experts (8.6 points), and is marked "mean filling".

[0084] 2.3 Classification and Quantification 2.3.1 Quantification of Objective and Measurable Indicators Quantification method for objective and measurable indicators: There are 24 objective and measurable indicators, including 15 positive indicators, 4 negative indicators, 3 interval indicators, and 2 indicators with proprietary formulas designed in this invention.

[0085] Positive indicators (15 items): The core feature is that "the value is positively correlated with the effect". The higher the value of this type of indicator, the better the effect of intelligent construction.

[0086] The core characteristic of the negative indicators (4 items) is that "the value is negatively correlated with the effectiveness". The quantitative standards all include "≤" (missing rate, intervention rate). The lower the value of these indicators, the better the effectiveness of intelligent construction.

[0087] All three range-type indicators are "error / deviation indicators". Their effectiveness depends on whether they fall within a reasonable range, rather than simply "higher / lower". The values ​​of these indicators must fall within a specific error or deviation range; otherwise, the effectiveness will decrease.

[0088] Specific formula indicators (2 items): These indicators need to be derived from multi-dimensional data and cannot be quantified by general positive, negative or interval formulas. They include the mean time between failures (MTBF) of equipment and the speed of disaster emergency response.

[0089] 2.3.2 Subjective and Objective Integrated Indicators Objective data for this type of indicator cannot fully cover all evaluation dimensions and needs to be supplemented and calibrated by subjective ratings (experts / users). It is determined by a combination of objective data (60%-80%) and expert ratings (20%-40%).

[0090] The two subjective-objective integrated indicators include multi-crop adaptation accuracy and production decision accuracy. A method combining objective data and expert scoring is employed.

[0091] Formula for a combined subjective and objective indicator: : The quantitative score of the kth tertiary indicator, ranging from 0 to 100 points. : The objective score of the k-th indicator (calculated based on actual data, ranging from 0 to 100 points); : Subjective score of the k-th indicator (mean of expert review, range 0-100 points); The objective weighting is dynamically set based on the characteristics of the indicator's "objective data availability" and "subjective evaluation necessity" (adjustable from 60% to 80%).

[0092] The objective scoring method, based on the indicator type (technical effectiveness, ease of use, promotion effectiveness, etc.), employs corresponding objective data quantification methods (such as weighted summation, ratio calculation, and achievement rate statistics) to convert the raw data into a standardized score of 0-100. Subjective score calculation method: A five-member expert panel was formed (based on the principle of "professional matching": technical indicators were matched with technical experts, and extension indicators were matched with agricultural extension experts); the experts scored each other back-to-back on a 1-10 scale (1 point = extremely unsuitable, 10 points = completely suitable). Calculation formula: The t-th expert's score Anomaly Handling: If an expert's score deviates from the mean by more than 2 standard deviations, that score is removed and the mean is recalculated (to avoid the influence of subjective bias). Based on whether the indicator quantification process relies on subjective evaluations such as expert ratings and user feedback, the three-level indicators are classified and quantified into objectively measurable indicators (24 items) and subjectively-objectively integrated indicators (2 items), and quantification rules are formulated for each.

[0093] 3. Determining indicator weights based on the combined weighting method The scientific nature of the indicator weights directly determines the reliability of the evaluation results. This invention adopts a combined weighting method of "subjective weight + objective weight + fusion coefficient". The Delphi method is used to determine the subjective weights, the entropy weight method is used to determine the objective weights, and the fusion coefficient is optimized through a genetic algorithm to finally obtain a scientific and reasonable combined weight.

[0094] 3.1 Determining Subjective Weights using the Delphi Method The system and methodology appropriately allocate expert group members to match the core requirements of the Delphi method in integrating multi-domain consensus and reducing subjective bias, and are highly compatible with the indicator characteristics and hierarchical logic of the evaluation system.

[0095] In this embodiment, an expert group of 11 people was formed, including 2 people specializing in AI algorithms, 2 people specializing in IoT engineering, 2 people specializing in planting, 1 person specializing in equipment operation and maintenance, 2 people specializing in agricultural economics, and 2 people specializing in agricultural technology promotion.

[0096] 3.1.1 Implementation of the three-round Delphi method First round of evaluation (anonymous preliminary evaluation): Each expert is provided with an "Indicator Importance Scoring Sheet," which clarifies the indicator definition, quantitative standards, and evaluation system positioning (e.g., perception layer = data foundation, decision-making layer = core brain). Experts score indicators independently within the same level, and different levels score independently.

[0097] Second round of evaluation (feedback and adjustments): Data summary: Calculate the average score for each indicator. Standard deviation Coefficient of variation ,formula: K: Number of experts: 11.

[0098] : The score given by the t-th expert to the k-th indicator.

[0099] Coefficient of variation reflects the dispersion of individual expert opinions. The smaller the group, the more consistent their opinions.

[0100] Feedback should provide experts with "statistical results of group scoring" (mean, standard deviation, coefficient of variation), with annotations. For indicators with high opinion dispersion, invite experts to explain the reasons for adjustment or revise the scores. Output the adjusted expert score sheet.

[0101] This example uses a primary indicator: CV≤0.15 meets the requirements and no adjustment is needed.

[0102] Third round of review (final confirmation): Repeat the second round of summary calculation, if all indicators ( And the coordination coefficient ( (As required by GB / T19582-2008, reflecting the overall consistency of expert opinions), then the experts confirm the final score; if there are still indicators ( An online consultation meeting was held, and a consensus was reached before confirmation.

[0103] Coordination coefficient The overall consistency of expert evaluations of indicators at the same level is quantified to compensate for the shortcomings of CV which only focuses on local dispersion. The value range is [0,1]. The closer W is to 1, the more consistent the ranking logic of the importance of the indicator at this level is among all experts.

[0104] : Coordination coefficient, with a value range of [0,1]; K: Number of experts; Number of indicators at the same level; : The total ranking value of expert scores for the j-th indicator (summed after ranking all expert scores for this indicator). For indicators with the same score, first, determine the ranking number of the indicator based on the score from highest to lowest. Then, take the average of the ranking numbers of all indicators with the same score as their common ranking.

[0105] The calculated concordance coefficient W = 0.395. This value is lower than the generally accepted high consensus threshold (e.g., W ≥ 0.7), but considering the following two points, the expert panel can still confirm the validity of the scoring results: the coefficients of variation (CV) for all indicators are less than 0.07, indicating that the experts' scores for the importance of each indicator are highly concentrated; and the concordance coefficient is positive, indicating a certain trend of consensus among the experts. After review and approval by the expert panel, the low dispersion of expert opinions and the validity of the scoring results are deemed conclusive, thus ending the Delphi process.

[0106] 3.1.2 Subjective weight normalization (hierarchical transmission) By using "hierarchical weighting + normalization", expert scores are converted into subjective weights for each level (ensuring that the weights of the same level sum to 1).

[0107] 3.2 Calculation of Objective Weights using the Entropy Weight Method The entropy weight method calculates objective weights based on the "dispersion" of actual monitoring data for indicators. Without any human intervention, it effectively counteracts the bias of subjective judgment, serving as a crucial supplement and optimization to subjective weights. This shifts weight allocation from "experience-driven" to "data-driven," ensuring that evaluation results better reflect the actual operational status of intelligent high-standard farmland construction. The core logic is: the higher the data dispersion (i.e., the greater the difference in indicators across different samples), the stronger the indicator's ability to distinguish effectiveness, and the higher its objective weight.

[0108] The calculation process of the entropy weight method is as follows: raw data → standardization processing → index weight → entropy value → difference coefficient → hierarchical normalization, following strict statistical mathematical logic.

[0109] The specific algorithms involved are not the focus of this invention and will not be described in detail in this paper.

[0110] 3.3 Combined Weights The final quantitative weight, obtained by weighting "subjective weight + objective weight" through a fusion coefficient, is used to measure the comprehensive contribution of each level of indicator (level 1, level 2, and level 3) to the effectiveness evaluation. This dual empowerment of experience-based guidance and data-driven evidence preserves the strategic judgments of industry experts on the core dimensions of intelligent construction while leveraging the discrete characteristics of underlying monitoring data to correct subjective biases. Ultimately, this achieves a weight allocation model where expert experience sets the direction and objective data provides strong support, ensuring that the evaluation results align with both industry technology guidelines and the actual operational status of intelligent farmland construction.

[0111] Formula for weighting and combining three-level indicators: : The weight of the j-th tertiary indicator combination; : Subjective weight of the j-th tertiary indicator; : The objective weight of the j-th tertiary indicator; The fusion coefficient, obtained through genetic algorithm optimization, .

[0112] Genetic algorithm optimizes fusion coefficient Calculation method: The mean square error (MSE) is minimized using the error minimization function: MSE( The smaller the value, the better. The lower the evaluation score, the closer it is to the actual operational effectiveness of high-standard farmland.

[0113] Sample size (as determined in this invention) (10 plots of land) : for the first Each sample in the fusion coefficient The overall score below; : No. Reference value for actual performance of each sample (output increase rate × 0.4 + labor saving rate × 0.3 + resource utilization efficiency improvement rate × 0.3).

[0114] Constraints: (Ensure that the combined weights are non-negative and sum to 1 after normalization).

[0115] This embodiment It can be further adjusted based on the availability of actual results data.

[0116] Formula for weighting and combining secondary indicators: : No. Weighting of each secondary indicator combination; : No. Subjective weighting of each secondary indicator; : No. Objective weights for each secondary indicator; The fusion coefficient, obtained through genetic algorithm optimization, .

[0117] The formula for merging the weights of primary indicator combinations is similar to that for merging the weights of secondary indicator combinations.

[0118] 4. Dynamic correction of multi-dimensional coefficients This invention integrates key influencing factors such as crop growth patterns, environmental changes, and technological iterations to construct a multi-dimensional correction system that combines general adaptation with specific optimization to correct the original quantitative scores of indicators. The basic correction coefficient covers common influencing factors for all indicators, while the indicator-specific correction coefficient precisely adjusts for the unique characteristics of different types of indicators. This upgrades the evaluation results from "static quantification" to "dynamic adaptation," improving the accuracy and rationality of the evaluation.

[0119] 4.1 Basic Correction Factor (Applicable to all indicators) The basic correction coefficient focuses on common influencing factors across all indicators, automatically identifying them through objective data or defining their values ​​according to industry standards, requiring no manual intervention and ensuring the objectivity and intelligence of the evaluation benchmark. Its core function is to provide a dynamic calibration benchmark for all evaluation indicators, eliminating systematic biases of common variables on the evaluation results.

[0120] The basic adjustment coefficient includes three core dimensions, including the reproductive period coefficient. Environmental coefficient Technology Maturity Coefficient .

[0121] Fertility period coefficient ( The dependence on and application effects of intelligent technologies vary significantly at different growth stages of crops, and the effectiveness of technologies during critical growth stages better reflects the construction results.

[0122] Environmental coefficient The natural environment is the fundamental scenario for the operation and decision-making models of intelligent equipment. Different environmental conditions have fundamentally different impacts on technological effectiveness: extremely unfavorable environments inhibit the stability of equipment operation and the applicability of models; while stressful environments (drought, high temperatures, etc.) are precisely the key scenarios where intelligent technologies can demonstrate their resilience; special soil conditions increase the difficulty of technology application. Value selection rules: Technology Maturity Ratio The application lifecycle of intelligent technologies (equipment, models) has a phased characteristic of "adaptation period - stable period - mature period - lagging period": new technologies are in the adaptation and optimization stage, and their operational stability may be affected by the adaptation of the scenario; in the mid-term, the technology operates stably and its performance is balanced; the iteration of mature technologies slows down and there may be technological lag; lagging technologies cannot meet the current intelligent standards and need to be strengthened.

[0123] 4.2 Indicator-Specific Correction Coefficient Five types of dedicated correction coefficients were designed to address the characteristics of different levels and types of indicators: Equipment Type Coefficient (K) 型 ), crop type coefficient (K) 作 ), disaster severity coefficient (K) 灾 ), regional economic coefficient (K) 经 ), and the application coefficient of archives (K) 档 ) Equipment type coefficient (K) 型 ( ): Quantify the differences in core performance such as operational stability and fault recovery capabilities of smart devices of different brands and qualities, and then correct the exclusive parameters of the original scores of device indicators. The core function is to eliminate the interference of "differences in device hardware foundation" on the evaluation results.

[0124] Crop type coefficient (K) 作 This is a set of crop adaptation indicators for decision-making levels. It quantifies the difficulty and technical value of intelligent decision-making models in adapting to the growth needs of different types of crops, and corrects the specific parameters of the model's decision-making indicator scores. The core orientation is to encourage the extension of intelligent technology to diverse crop scenarios.

[0125] Disaster severity coefficient (K) 灾For emergency response indicators at the execution level, this system quantifies the differences in response efficiency of intelligent execution systems (such as emergency irrigation and drone-based plant protection) under different levels of natural disaster scenarios, corrects the specific parameters for emergency indicator scores, and its core function is to highlight the value of technical support in extreme scenarios.

[0126] Regional economic coefficient (K) 经 For the promotion and benefit-related indicators at the benefit level, the differences in the acceptance, promotion difficulty and input-output ratio of intelligent technologies in regions with different economic development levels are quantified, and the specific parameters for the scores of benefit-related indicators are corrected. The core function is to balance the evaluation bias caused by uneven regional development.

[0127] Archive application coefficient (K) 档 For the digital asset indicators at the benefit level, the difference between the "application value" and "archival value" of farmland digital archives is quantified, and the exclusive parameters for the scores of archival indicators are corrected. The core orientation is to promote the transformation of digital archives from "static archiving" to "dynamic empowerment".

[0128] 4.3 Dynamic Correction Calculation The design employs a combination of "product of basic coefficients + superposition of specific coefficients," where the raw quantified score is first adjusted by a basic correction coefficient (K). 生 ×K 环 ×K 技 This enables dynamic calibration of the full-indicator evaluation benchmark, and then uses an indicator-specific correction coefficient (K) to achieve this. 专属 This allows for precise adjustments based on individual differences, ultimately resulting in a corrected score (Q). 修正后得分 ) Constraints: The corrected score shall not exceed 100 points and shall not be lower than 0 points.

[0129] 5. Comprehensive Evaluation Results of Intelligentization Effectiveness The intelligent effectiveness comprehensive evaluation result of this invention is based on a full-chain indicator system of "perception-decision-execution-benefit", a multi-dimensional dynamic correction model and a combined weighting method. It calculates the comprehensive score of the indicators through indicator data collection, dynamic correction calculation and comprehensive weight fusion calculation. At the same time, it constructs a three-dimensional judgment logic based on the comprehensive score range, with the first-level indicator threshold as a premise and the core veto item as a constraint, and classifies and grades the evaluation results.

[0130] 5.1 Overall Score The overall score is a weighted sum of the corrected scores of the three-level indicators multiplied by the combined weights of the three levels. It directly reflects the overall effectiveness of intelligent construction. The formula is as follows: 5.3 Level Determination The grading method of this invention is based on a three-dimensional quantitative grading logic: "comprehensive score range as the foundation, primary indicator threshold as a prerequisite, and core veto items as constraints." The grading process is a fully automated grading logic based on "quantitative data → layered verification → gradient downgrading → closed-loop locking." It uses comprehensive score, primary indicator performance, and core veto items as the three core criteria, and follows the principle of "gradual verification from high to low, with downgrading triggered by certain conditions," ensuring that the grading results are unique, reproducible, and consistent with the actual construction level of the project. The grading is divided into four levels: Excellent, Good, Qualified, and Unqualified.

[0131] 5.3.2 Core Veto Items The core parameters of perception coverage, production decision accuracy, disaster emergency response speed, and unit operation energy consumption reduction rate are selected as cross-level general core indicators to form a rigid constraint chain of "basic capabilities - core functions - guarantee effectiveness - policy adaptation".

[0132] 1. Core Parameter Perception Coverage (Data Foundation): As the "data entry point" for intelligent construction, it directly determines the effectiveness of the decision-making and execution layers. If the coverage rate is less than 80%, key data such as soil moisture and crop growth will be missing, leading to "no data to rely on and no clear direction." Therefore, it is listed as an entry-level veto item for all levels. The threshold for the excellent level is raised to 90%, highlighting the data integrity advantage of benchmark projects and supporting high-level intelligent decision-making.

[0133] 2. Production Decision Accuracy (Smart Brain): This is the core value of the intelligent system, directly related to crop yield and resource utilization efficiency. If the accuracy rate is less than 70%, decisions regarding irrigation, fertilization, etc., may deviate from actual needs, even causing yield losses, thus violating the core goal of "improving quality and efficiency." Therefore, the threshold gradually increases from qualified to excellent levels (70%→85%), reflecting the logic that "the higher the level, the stricter the requirement for decision accuracy."

[0134] 3. Disaster Emergency Response Speed ​​(Execution): This is a key test of the effectiveness of intelligent technology implementation and directly relates to the resilience of high-standard farmland. If the response speed is less than 50 points (corresponding to a time consumption of more than 80 seconds), it will be unable to respond promptly to disasters such as droughts and floods, and the protective value of intelligent technology cannot be realized. Therefore, differentiated thresholds are set according to levels (50 points → 70 points → 80 points) to highlight the rapid response capabilities of outstanding projects in extreme scenarios.

[0135] 4. Unit Operation Energy Consumption Reduction Rate (Policy Adaptation): Closely aligned with the national "dual-carbon" strategy and the requirements for green agricultural development, this is a significant added value of intelligent construction. If the reduction rate is <5%, the energy-saving and efficiency-enhancing goals have not been achieved, which is inconsistent with policy guidance. Therefore, it is listed as a veto item for qualified and above levels, and the excellent level implicitly requires a higher level of "energy consumption reduction rate ≥10%", promoting the transformation of intelligent construction from "high-efficiency production increase" to "high-efficiency and low-carbon".

[0136] 5.3.3 Degradation Rules and Rigid Enforcement Mechanism The core veto items are designed according to the principle of "the higher the level, the stricter the constraints". Each level sets three key indicator thresholds, covering three dimensions: "basic capabilities, core functions, and target results". Triggering any veto item will initiate the downgrade process. To avoid subjectivity and controversy in level determination, four rigid enforcement rules are formulated to clarify the downgrade logic, path and boundaries, and ensure that the determination results are objective and fair.

[0137] 5.3.1 Priority Determination Rules The execution order is: "Preliminary matching of comprehensive scores → Verification of primary indicator thresholds → Verification of core veto items → Final grade locking": 1. First step: Determine the candidate level based on the overall score; 2. Second step: Check the threshold of the first-level indicator corresponding to the candidate level (e.g., excellent requires all 4 first-level indicators to be ≥85 points). If the threshold is not met, directly downgrade to the next candidate level (e.g., if 1 first-level indicator is 83.20 points, downgrade to "good"). There is no need to check the rejection items. 3. Third step: For candidate levels that meet the threshold requirements, verify the core rejection items (such as the three indicators such as MTBF of the equipment for the "Good" level). If any rejection item is triggered, start the downgrade process. 4. Fourth step: Repeat "threshold check → veto item verification" until a level with no threshold violations and no veto items triggered is found, which is the final level.

[0138] 5.3.2 Downshift Path Rules • Regular downgrade: When only one veto item is triggered or one threshold is not met, the downgrade is carried out in adjacent levels of "Excellent → Good → Pass → Fail" (e.g., Excellent level triggers "Model self-learning optimization rate = 78 points" → downgrades to Good). • Downgrade across levels: When two or more veto items are triggered or two or more thresholds are not met, the level can be downgraded across levels (e.g., Excellent level triggers "Core parameter perception coverage = 88% + production decision accuracy = 82 points" → directly downgraded to qualified; Good level has 1 threshold not met + 1 veto item → directly downgraded to unqualified). • Extreme downgrade: If the candidate level is "Excellent", "Good" or "Qualified", but triggers 3 core veto items, or fails to meet the bottom line indicators such as core parameter perception coverage <80% or production decision accuracy <70%, it will be directly downgraded to "Unqualified".

[0139] 5.3.3 Irreversibility and Boundary Value Rules 1. Irreversible rule: Level determination only supports "downgrading from high to low," and does not support "upgrading from low to high." 2. Boundary Value Rule: Both the score and the threshold are reserved to two decimal places, adopting the principle of "closed on the left and open on the right": "≥X" includes the value of X, and "<X" does not include the value of X (example: Perception Coverage Rate = 90.00%, does not trigger the veto item of the excellent level "<90%"; Comprehensive Score = 80.00 points, meets the good level range); 3. Data Missing Handling: When the core veto item index data is missing, it is processed as "triggering the veto item" (example: Production Decision Accuracy data is missing → judged as <70 points, triggering the unqualified veto item); When the non-veto item index data is missing, it is calculated as 50 points, and the score of the corresponding first-level index shall not exceed 80% of its weight at most.

[0140] 6. Three-Dimensional Optimization Suggestion Push Mechanism Priority refers to the hierarchical division of the intelligent construction optimization tasks according to the core attributes of the evaluation indicators, the scope of problem influence, and the urgency of rectification. Its core role is to guide the rational allocation of optimization resources and the sorting of implementation order, ensuring that key problems are solved first and important needs are mainly met, and ultimately maximizing the optimization effect.

[0141] Construct a three-dimensional linkage push mode of score range - indicator type - optimization priority, where the priority sorting rule is: P1 (urgent) > P2 (important) > P3 (general), and it can automatically match and push targeted optimization suggestions. For details, see Figure 3 as shown in

[0142] In the weight dynamic calculation module of the present invention, the genetic algorithm, as an intelligent optimization algorithm, is used to solve the optimization problem of the fusion coefficient of subjective weight and objective weight. It adopts the way of bionic evolution to achieve global optimization search, especially suitable for the optimization problem in the multi-dimensional weight parameter space.

[0143] The genetic algorithm described above is a class of random search methods inspired by the natural selection and genetic mechanism in the biological world, mainly including basic operations such as selection, crossover, and mutation. It can explore multiple candidate solutions in parallel in the search space and promote the evolution of the population to a better solution through fitness evaluation. Select suitable individuals as parents according to the fitness values of individuals for generating the next generation. By recombining the "gene" information of two parent individuals, new individuals with new characteristics are generated to accelerate the search process. Random perturbations are made to some individual gene loci to maintain population diversity and avoid falling into local optima.

[0144] The weight fusion process corresponding to these terms in the present invention is roughly as follows: Chromosome Encoding: Represent the fusion coefficient vector as the chromosome of the genetic algorithm; Fitness function definition: The error between the comprehensive evaluation score calculated based on the combined weights and the reference value of the actual effect is used as the fitness assessment; Iterative evolution: By repeatedly performing selection, crossover, and mutation operations, new candidate solutions for weight fusion coefficients are continuously generated. Termination condition: Evolution stops when the fitness value meets the preset threshold or the maximum number of iterations is reached, and the optimized fusion coefficient is output.

[0145] Genetic algorithms, with their strong global search capabilities, no gradient requirements, and applicability to complex multi-objective optimization, can effectively avoid getting trapped in local optima in the optimization of fused weight parameters.

[0146] To make the technical solutions in this application clearer, the combination of "subjective weight + objective weight + fusion coefficient" used in this disclosure is explained. However, it should be understood that these statistical and weighting methods are not the main focus of this invention. Their primary purpose is to make the technical solutions in this application clearer. Any omissions or omissions in the relevant content of this disclosure do not affect the specific implementation of this application. Even without these explanations, the technical solutions in this application should be clear and complete to those skilled in the art.

[0147] In summary, compared to existing evaluation methods, the construction effectiveness evaluation method of this invention achieves a comprehensive, quantifiable, dynamic, and closed-loop evaluation of the achievements in the construction of smart agriculture in high-standard farmland. The evaluation method and system disclosed herein cover the entire chain from data perception, intelligent decision-making, precise execution to comprehensive benefits, ensuring that each indicator has a clear definition, quantitative standards, and reliable data sources, overcoming the shortcomings of traditional evaluations that "emphasize equipment configuration while neglecting operational efficiency." The method in this disclosure integrates subjective and objective information, employs a dynamically adapted scientific weighting method, and utilizes optimization algorithms to determine the optimal integration ratio, thereby improving the scientific and objective nature of indicator weight allocation and avoiding the limitations of weights relying on subjective experience.

[0148] The method in this application can achieve intelligent and automated comprehensive evaluation and accurate diagnosis of effectiveness. Based on the aforementioned indicator system, weight and correction mechanism, it automatically calculates the comprehensive score and the score of each stage, and gives an objective level assessment according to the preset multi-dimensional judgment rules (comprehensive score, first-level indicator threshold, core veto item), accurately locating the shortcomings in construction.

[0149] Compared to existing technologies, this patent offers the following significant advantages: it specifically addresses core issues in the evaluation of intelligent construction of high-standard farmland, such as vague indicators, singular methods, and a disconnect between evaluation and application, forming significant advantages in multiple dimensions. Specifically, it proposes a dynamic weighting method that integrates subjective and objective weights; it establishes a multi-dimensional dynamic correction mechanism to improve the adaptability of evaluation scenarios, introducing basic correction factors such as growth period coefficient, environmental coefficient, and technology maturity coefficient, as well as specific correction coefficients for equipment type, crop type, and disaster level, achieving dynamic adaptation to different production scenarios, technological stages, and regional conditions. This mechanism ensures that the matching degree between evaluation results and actual farmland operation exceeds 90%, effectively avoiding the limitations of static "one-size-fits-all" evaluation.

[0150] A closed-loop feedback mechanism of "evaluation-diagnosis-optimization" has been implemented. A fully automated, cloud-edge-device collaborative evaluation system has been developed. Based on a three-tiered "cloud-edge-device" architecture, the system automates the entire process, including automatic data collection, intelligent preprocessing, dynamic weight calculation, evaluation model fusion, result visualization output, and feedback push. Preliminary comparisons show that compared to traditional manual evaluation, work efficiency is increased by over 70%, and the evaluation cycle is shortened from 15-20 working days to 3-5 working days, significantly reducing the operational threshold and implementation costs.

[0151] The above detailed description further illustrates the purpose, technical solution, and beneficial effects of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A system for evaluating the effectiveness of intelligent construction of high-standard farmland, characterized in that, The system comprises multiple modules distributed across the terminal perception layer, edge computing layer, and cloud service layer: The data acquisition module is distributed in the terminal sensing layer. The data acquisition module includes a data acquisition terminal, various Internet of Things sensing terminals and intelligent execution terminals arranged in each block of the high-standard farmland to be evaluated, and is used to collect soil data, meteorological data, crop growth data and execution layer operation data. A data preprocessing module is deployed in the edge computing layer. The data preprocessing module includes an edge computing gateway, which performs preprocessing, data cleaning, and standardization on the data from the data acquisition module. The weight dynamic calculation module, evaluation analysis module, and result output and feedback module are deployed in the cloud service layer to realize calculation, data storage, model training and evaluation service publishing. The data acquisition module can collect hardware information, installation location, and runtime information of each IoT sensing terminal; the data acquisition module can also be equipped with a manual data entry terminal and a docking port for obtaining data from an external big data platform.

2. The evaluation system according to claim 1, characterized in that, The data acquisition module collects farmland sensing data and execution equipment operation data in real time; the edge computing gateway of the data preprocessing module performs protocol conversion between different terminals. The manual data entry terminal receives the expert scoring data; the interface port is used to connect to and acquire data from the agricultural and rural big data platform and the external geographic information system platform.

3. The evaluation system according to claim 2, characterized in that, The data preprocessing module is also configured to perform real-time indicator calculation and localized emergency evaluation. In the real-time indicator calculation and processing, the data preprocessing module monitors the real-time indicators of the online rate and instruction response timeliness of each IoT sensing terminal and intelligent execution terminal, and triggers an early warning when the real-time indicator is lower than a predetermined threshold. In the localized emergency evaluation process, the data preprocessing module performs localized evaluation processing while the network is offline, generating a preliminary assessment report of the emergency effectiveness.

4. The evaluation system according to any one of claims 1-3, characterized in that, The data cleaning and standardization performed by the data preprocessing module include: outlier removal, missing value completion, and unit unification and classification quantification; The weight dynamic calculation module generates subjective weight coefficients based on expert scoring data from the manual data entry terminal, calculates objective weight coefficients based on data from a predetermined time period, and optimizes the fusion coefficients through a genetic algorithm. The system matches dynamic correction coefficients on predetermined cycle days to correct indicator scores, calculates comprehensive scores through multi-model fusion, and determines the level according to rules.

5. The evaluation system according to claim 4, characterized in that, The evaluation and analysis module matches basic correction coefficients and specific correction coefficients based on crop growth period, environmental conditions, and technological maturity to dynamically correct indicator scores. Preliminary results are generated through fuzzy comprehensive evaluation, and after being corrected by a neural network, a comprehensive score is generated by hierarchical weighted summation. In the evaluation and analysis module, dynamic influencing factors related to crop growth period, environmental conditions, and technological maturity are introduced to calibrate the original scores of the indicators. At the same time, specific correction coefficients related to equipment type, crop type, and disaster level are also introduced to achieve dynamic adaptation to different production scenarios, technological stages, and regional conditions.

6. The evaluation system according to claim 5, characterized in that, In the evaluation and analysis module, a multi-model fusion evaluation and dynamic correction system is constructed, specifically including: The first comprehensive calculation of the scores of each evaluation indicator after dynamic correction is performed based on the fuzzy comprehensive evaluation model to reduce the impact of fluctuations in a single indicator on the evaluation results. The first comprehensive calculation results are nonlinearly corrected based on a neural network model in order to learn the mapping relationship between historical evaluation results and actual construction results; By weighting and fusing the outputs of different models, a final comprehensive evaluation score is generated.

7. The evaluation system according to claim 6, characterized in that, The system also includes a system management and security module; The IoT sensing terminals include: soil moisture / fertility sensors, weather stations, and crop growth cameras; The intelligent execution terminal includes: intelligent irrigation valve, variable fertilizer applicator, UAV plant protection equipment, agricultural machinery Beidou positioning terminal, etc., which are used to collect execution layer operation instruction logs, energy consumption data, and collaborative operation parameters; The data acquisition terminal includes: an edge acquisition gateway and a rugged handheld PAD.

8. A method for evaluating the effectiveness of intelligent construction of high-standard farmland, employing the evaluation system according to any one of claims 1 to 9, comprising the following steps: Set the evaluation time period, evaluation indicators, and evaluation start date; Data acquisition steps: Real-time collection of farmland sensing data and execution equipment operation data; protocol conversion and local caching using edge gateways; collection of expert scores, farmer ledgers, and other data; synchronization with third-party platforms to obtain data from agricultural and rural big data platforms and GIS platforms; all data is filtered by the edge layer and then synchronized to cloud storage. Data preprocessing steps: The cloud-based data preprocessing module performs multi-strategy outlier removal, fills in missing values, automatically identifies indicator types and completes standardization and classification quantification, and generates initial indicator scores of 0-100 points, providing a data foundation for subsequent evaluation calculations; The dynamic weight calculation process involves the weight calculation module automatically calling the Delphi method expert scoring data and generating subjective weights through at least two rounds of evaluation. Based on a predetermined number of monthly data samples, objective weights are calculated using the entropy weight method; with the goal of matching the evaluation scores with the actual performance reference values, the fusion coefficients are optimized using a genetic algorithm to generate the final combined weights. Dynamic evaluation steps: The evaluation and analysis module automatically matches the basic correction coefficient and the specific correction coefficient based on crop growth period, environmental conditions, and technological maturity to complete the dynamic correction of the indicator score; Preliminary results are generated through fuzzy comprehensive evaluation. After being corrected by a neural network, a comprehensive score is generated according to the hierarchical weighted summation formula. The threshold of the first-level indicators and the core veto items are checked to determine the final evaluation level. Results feedback steps: The results output module generates multi-dimensional visualization charts and standardized / customized evaluation reports, and generates hierarchical optimization suggestions based on the evaluation results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the functions of the evaluation system for the effectiveness of intelligent construction of high-standard farmland as described in any one of claims 1 to 6.

Images