Intelligent irrigation method and device based on large language model

Abstract

The application provides an intelligent irrigation method based on a large language model, comprising: an agent initialization step, wherein a plurality of agents are defined based on a large language model, and each agent adopts a hybrid model architecture; a data collection and analysis step, wherein each agent respectively acquires multi-modal data, and performs fusion analysis on the multi-modal data through the hybrid model architecture to obtain an analysis result of each agent; an information sharing and collaboration step, wherein the multi-modal data and the analysis result uploaded by each agent are comprehensively processed to generate a corresponding irrigation instruction; and a dynamic control execution step, wherein irrigation equipment is controlled to automatically irrigate according to the irrigation instruction. The application also provides an intelligent irrigation device based on a large language model, a storage medium and an electronic device. Therefore, the application utilizes the reasoning capability of the large language model and the collaborative characteristics of the multi-agent architecture, completes comprehensive monitoring of farmland environment and intelligent irrigation decision-making through multi-agent collaboration, and enhances the self-adaptive capability of the system.

Description

Technical Field

[0001] This invention relates to the field of intelligent agricultural management technology, and in particular to an intelligent irrigation method, device, storage medium, and electronic device based on a large language model. Background Technology

[0002] In modern agriculture, smart irrigation technology is increasingly being used to improve water resource utilization efficiency. However, traditional smart irrigation systems typically employ a single control method, which is insufficient to cope with the complex and ever-changing agricultural environment. These systems are usually based on simple sensor control, making it difficult to achieve precise and real-time adjustments to irrigation plans. Typical smart irrigation methods include data feedback modes based on sensor monitoring and simple decision control modes based on predictive models, but these methods suffer from slow response speeds and poor environmental adaptability.

[0003] For example, existing smart irrigation technologies mostly focus on the collection of single data sources and the irrigation execution process, lacking a comprehensive consideration of multiple factors such as soil and climate. This results in insufficient irrigation precision and rigid operation methods when environmental changes are significant, making it difficult to meet the diverse needs of different farmland environments.

[0004] The main problems with existing irrigation systems are as follows:

[0005] (1) Single decision-making mode: Traditional systems often make decisions based on information from a single sensor, which makes them unable to adapt to complex scenarios with dynamic changes in multiple factors, especially climate fluctuations and changes in soil structure.

[0006] (2) Lack of collaborative control: The concept of multi-agent has not yet been applied, and there is insufficient coordination between various system modules, which easily leads to data silos, poor information exchange, and affects decision-making efficiency.

[0007] (3) Poor self-adaptability: Existing technologies lack efficient adaptive adjustment mechanisms, making it difficult to dynamically optimize irrigation processes according to real-time changes in agricultural needs, resulting in increased operating costs and resource waste.

[0008] In conclusion, the existing technology obviously has inconveniences and defects in practical use, so it is necessary to improve it. Summary of the Invention

[0009] To address the aforementioned shortcomings, the present invention aims to provide an intelligent irrigation method, device, storage medium, and electronic device based on a large language model. This method enables comprehensive monitoring of the farmland environment and intelligent irrigation decision-making through multi-agent collaboration, while also enhancing the system's adaptive capabilities.

[0010] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0011] In a first aspect, embodiments of the present invention provide an intelligent irrigation method based on a large language model, comprising:

[0012] The agent initialization step involves defining multiple agents based on a large language model, with each agent employing a hybrid model architecture.

[0013] In the data acquisition and analysis step, each of the intelligent agents acquires multimodal data and performs fusion analysis on the multimodal data through the hybrid model architecture to obtain their respective analysis results;

[0014] The information sharing and collaboration step involves comprehensively processing the multimodal data uploaded by each intelligent agent and the analysis results to generate corresponding irrigation instructions;

[0015] The dynamic control execution steps regulate the irrigation equipment to perform automatic irrigation according to the irrigation instructions.

[0016] According to the intelligent irrigation method based on a large language model of the present invention, the multimodal data includes environmental information and / or crop information;

[0017] The intelligent agent includes multiple analytical intelligent agents, management intelligent agents, and irrigation intelligent agents;

[0018] The data acquisition and analysis steps include:

[0019] Each of the aforementioned analytical agents acquires the required multimodal data through the Internet of Things acquisition layer, and performs fusion analysis on the multimodal data through the hybrid model architecture to obtain its own analysis results;

[0020] The information sharing and collaboration steps include:

[0021] Each of the aforementioned analytical agents uploads the multimodal data and the analysis results to the information sharing layer for sharing;

[0022] The management agent comprehensively processes the analysis results uploaded by each analysis agent, balances the decision weights of each agent through game theory strategies, and generates the corresponding irrigation instructions.

[0023] The dynamic control execution steps include:

[0024] The irrigation agent controls the irrigation equipment to perform automatic irrigation according to the irrigation command.

[0025] According to the intelligent irrigation method based on a large language model of the present invention, the environmental information includes meteorological information and / or soil information;

[0026] The plurality of analytical agents include:

[0027] The future meteorological intelligent agent will be used to analyze the collected meteorological information to produce meteorological trend analysis results.

[0028] The soil moisture intelligence agent is used to analyze soil moisture analysis results based on the collected soil information.

[0029] A soil nutrient intelligence agent is used to analyze the soil nutrient content based on the collected soil information.

[0030] A growth analysis agent is used to analyze the crop growth status based on the collected crop information and / or environmental information.

[0031] A growth period target intelligent agent is used to analyze the crop growth stage and set corresponding stage target analysis results based on the collected crop information and / or environmental information.

[0032] According to the intelligent irrigation method based on a large language model of the present invention, the steps of each of the analytical agents acquiring the multimodal data required by the Internet of Things acquisition layer include:

[0033] Each of the aforementioned analytical agents acquires the required multimodal data through the IoT acquisition layer;

[0034] Each of the aforementioned analytical agents performs spatiotemporal alignment and outlier correction on the multimodal data through a specialized algorithm layer;

[0035] Each of the analytical agents performs cross-validation on the corrected multimodal data.

[0036] The intelligent irrigation method based on a large language model according to the present invention further includes:

[0037] The reasoning enhancement step involves enhancing the reasoning of each agent through a global knowledge enhancement layer.

[0038] According to the intelligent irrigation method based on a large language model of the present invention, the reasoning reinforcement step further includes:

[0039] A knowledge graph is constructed based on a historical irrigation case database and the experience of agronomic experts through a global knowledge enhancement layer, and the decision-making paths of each agent are optimized through reinforcement learning.

[0040] According to the intelligent irrigation method based on a large language model of the present invention, the crop information includes the crop growth stage;

[0041] The method further includes:

[0042] In the adaptive adjustment step, for different crop growth stages and / or environmental information, each intelligent agent adaptively adjusts the analysis and control parameters by comparing historical data with real-time monitoring data.

[0043] Secondly, embodiments of the present invention provide an intelligent irrigation device based on a large language model constructed according to any one of the methods described above, the device comprising:

[0044] The agent initialization module is used to define multiple agents in a large language model, with each agent adopting a hybrid model architecture.

[0045] The data acquisition and analysis module is used to acquire multimodal data from each of the aforementioned intelligent agents, and to perform fusion analysis on the multimodal data through the hybrid model architecture to obtain their respective analysis results;

[0046] The information sharing and collaboration module is used to comprehensively process the multimodal data and analysis results uploaded by each of the intelligent agents to generate corresponding irrigation instructions;

[0047] The dynamic control execution module is used to regulate the irrigation equipment to perform automatic irrigation according to the irrigation command.

[0048] Thirdly, embodiments of the present invention provide a storage medium for storing a computer program for performing any of the methods described herein.

[0049] Fourthly, embodiments of the present invention provide an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor, when executing the computer program, implements any of the methods described above.

[0050] Therefore, the intelligent irrigation method based on a large language model of the present invention includes: defining multiple agents based on a large language model, each agent adopting a hybrid model architecture; each agent acquiring multimodal data and performing fusion analysis on the multimodal data through the hybrid model architecture to obtain its own analysis results; comprehensively processing the multimodal data and analysis results uploaded by each agent to generate corresponding irrigation instructions; and controlling irrigation equipment for automatic irrigation according to the irrigation instructions. Thus, the present invention utilizes the reasoning ability of the large language model and the collaborative characteristics of the multi-agent architecture, not only overcoming the limitations of single-agent decision-making, but also achieving comprehensive monitoring of the farmland environment and intelligent irrigation decision-making through multi-agent collaboration, and enhancing the system's adaptability. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the intelligent irrigation method based on a large language model provided in Embodiment 1 of the present invention.

[0052] Figure 2 This is a flowchart illustrating the intelligent irrigation method based on a large language model provided in Embodiment 2 of the present invention.

[0053] Figure 3 This is a schematic diagram of the multi-agent structure of the intelligent irrigation method based on a large language model provided in Embodiment 3 of the present invention;

[0054] Figure 4 This is a multi-agent collaborative architecture diagram of the intelligent irrigation method based on a large language model provided in Embodiment 4 of the present invention;

[0055] Figure 5 This is a schematic diagram of the structure of the intelligent irrigation device based on a large language model provided in Embodiment 5 of the present invention;

[0056] Figure 6 This is a schematic diagram of the structure of the intelligent irrigation device based on a large language model provided in Embodiment Six of the present invention;

[0057] Figure 7 This is a schematic diagram of the structure of the electronic device provided in Embodiment 7 of the present invention. Detailed Implementation

[0058] 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 accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0059] It should be noted that references to "an embodiment," "embodiment," "example embodiment," etc., in this specification refer to the described embodiment including specific features, structures, or characteristics, but not every embodiment must include these specific features, structures, or characteristics. Furthermore, such expressions do not refer to the same embodiment. Moreover, when describing specific features, structures, or characteristics in conjunction with embodiments, whether or not explicitly described, it is indicated that incorporating such features, structures, or characteristics into other embodiments is within the knowledge of those skilled in the art.

[0060] Furthermore, certain terms are used in the specification and subsequent claims to refer to specific components or parts. Those skilled in the art will understand that manufacturers may use different names or terms to refer to the same component or part. This specification and subsequent claims do not distinguish components or parts by differences in name, but rather by differences in function. The terms "comprising" and "including" used throughout the specification and subsequent claims are open-ended and should be interpreted as "including but not limited to." Additionally, the term "connection" here includes any direct and indirect electrical connection means. Indirect electrical connection means include connections made through other means.

[0061] The intelligent irrigation method based on a large language model provided by the present invention will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0062] This invention relates to the field of intelligent agricultural management technology, particularly to agricultural irrigation systems based on Large Language Models (LLM) and multi-agent systems, aiming to achieve unmanned, dynamically adaptive farmland water and fertilizer irrigation management. During research on intelligent agricultural irrigation technology, it was found that the aforementioned shortcomings are mainly caused by the overly simplistic control modes and insufficient information collaboration of traditional systems, lacking sufficient data analysis depth and dynamic adjustment capabilities. By studying the collaborative characteristics of multi-agent control architecture and the advantages of large language models in knowledge reasoning, the inventors designed a multi-agent autonomous control method based on large language models, effectively solving the dynamic adaptability problem of irrigation systems. This method not only overcomes the limitations of single-agent decision-making but also utilizes multi-agent collaboration to complete comprehensive monitoring and intelligent decision-making of the farmland environment, enhancing the system's adaptability.

[0063] Figure 1 This is a flowchart illustrating the intelligent irrigation method based on a large language model provided in Embodiment 1 of the present invention. The method includes:

[0064] Step S101, agent initialization step, defines multiple agents based on a large language model, and each agent adopts a hybrid model architecture.

[0065] Step S102, data acquisition and analysis step: each agent acquires multimodal data and performs fusion analysis on the multimodal data through a hybrid model architecture to obtain its own analysis results.

[0066] Preferably, the multimodal data includes environmental information and / or crop information. More preferably, the environmental information includes meteorological information and / or soil information. The crop information includes crop growth stage.

[0067] Step S103, the information sharing and collaboration step, involves comprehensively processing the multimodal data and analysis results uploaded by each agent to generate corresponding irrigation instructions.

[0068] Step S104: Dynamic control execution step, according to irrigation instructions, adjust the irrigation equipment to carry out automatic irrigation.

[0069] This invention proposes an autonomous irrigation method based on a large language model and multi-agent control, applied to complex agricultural irrigation scenarios. Through the division of labor and cooperation among multiple agents, it achieves intelligent management and dynamic control of farmland irrigation. The method defines multiple agents with specific functions, forming a collaborative multi-agent architecture. Each agent adopts a hybrid model architecture to enhance its adaptability in complex environments. Supported by an Internet of Things (IoT) data acquisition layer and a specialized algorithm layer, each agent dynamically senses and responds to changes in the external environment.

[0070] Figure 2 This is a flowchart illustrating the intelligent irrigation method based on a large language model provided in Embodiment 2 of the present invention. The method includes:

[0071] Step S201, Agent initialization step, defines multiple agents based on a large language model, and each agent adopts a hybrid model architecture.

[0072] Preferably, the plurality of intelligent agents includes a plurality of analytical intelligent agents 200, management intelligent agents 300, and irrigation intelligent agents 400, such as... Figure 3 As shown, each intelligent agent possesses the capabilities of perception, analysis, decision-making, execution, and memory. Each intelligent agent operates independently but shares data input and knowledge enhancement modules. More preferably, the plurality of analytical intelligent agents 200 include:

[0073] The future weather intelligence agent 210 is used to analyze weather trend analysis results based on collected weather information.

[0074] The Soil Moisture Intelligence Agent 220 is used to analyze soil moisture analysis results based on collected soil information.

[0075] The Soil Nutrient Intelligence Agent 230 is used to analyze soil nutrient content based on collected soil information.

[0076] The growth analysis agent 240 is used to analyze the crop growth status based on the collected crop information and / or environmental information.

[0077] The growth period target agent 250 is used to analyze the crop growth stage and set corresponding stage target analysis results based on the collected crop information and / or environmental information.

[0078] Step S202, data acquisition and analysis step: each agent acquires multimodal data and performs fusion analysis on the multimodal data through a hybrid model architecture to obtain its own analysis results.

[0079] Preferably, the data acquisition and analysis steps include:

[0080] Each analytical agent acquires the required multimodal data through the Internet of Things (IoT) acquisition layer, and performs fusion analysis on the multimodal data through a hybrid model architecture to obtain its own analysis results. Preferably, each analytical agent acquires environmental information and / or crop information in real time through the IoT acquisition layer. More preferably, the environmental information includes meteorological information and / or soil information, etc. The crop information includes crop growth stage, etc. For example, the future meteorological agent 210 predicts short-term meteorological trends based on meteorological sensor data, and the soil moisture agent 220 analyzes changes in soil moisture to provide a basis for irrigation decisions. For example, air temperature and humidity, soil moisture, soil nutrients, etc., are independently processed and analyzed to generate irrigation suggestions.

[0081] Better still, the steps for each analytical agent to acquire the required multimodal data through the IoT acquisition layer include:

[0082] (1) Each analytical agent acquires the required multimodal data through the Internet of Things acquisition layer.

[0083] (2) Each analytical agent performs spatiotemporal alignment and outlier correction on multimodal data through a professional algorithm layer.

[0084] (3) Each analytical agent performs cross-validation on the corrected multimodal data.

[0085] Preferably, the multimodal data includes environmental information and / or crop information. More preferably, the environmental information includes meteorological information and / or soil information. The crop information includes crop growth stage.

[0086] Step S203, Reasoning Enhancement Step, enhances the reasoning of each agent through the global knowledge enhancement layer to ensure the accuracy of decision-making and response speed.

[0087] Preferably, the reasoning enhancement step further includes:

[0088] By constructing a knowledge graph based on a historical irrigation case database and the experience of agronomic experts through a global knowledge enhancement layer, the decision-making paths of each agent are optimized through reinforcement learning.

[0089] Step S204, the information sharing and collaboration step, involves comprehensively processing the multimodal data and analysis results uploaded by each agent to generate corresponding irrigation instructions.

[0090] Preferably, the agents exchange data through an information sharing layer to achieve collaborative decision-making among multiple agents.

[0091] Preferably, the information sharing and collaboration steps include:

[0092] Each analytical agent uploads multimodal data and analysis results to the information sharing layer for sharing and reference by other agents.

[0093] The management agent comprehensively processes the analysis results uploaded by each analysis agent, balances the decision weights of each agent through game theory strategies, and generates corresponding irrigation instructions.

[0094] Step S205, dynamic control execution step, involves regulating the irrigation equipment for automatic irrigation according to irrigation instructions. Preferably, the irrigation equipment is regulated according to irrigation instructions to achieve unmanned water and fertilizer supply.

[0095] Preferably, the dynamic control execution step includes:

[0096] The irrigation agent controls the irrigation equipment to carry out automatic irrigation based on irrigation instructions.

[0097] Step S206, Adaptive Adjustment Step: For different crop growth stages and / or environmental information, each agent adaptively adjusts the analysis and control parameters by comparing historical data with real-time monitoring data.

[0098] Preferably, each agent adaptively adjusts its analysis and control parameters under different growth stages and soil conditions to cope with environmental changes.

[0099] Figure 4This is a multi-agent collaborative architecture diagram of the intelligent irrigation method based on a large language model provided in Embodiment 4 of the present invention. It mainly includes the Fuxi Brain, multiple agents, and several models. These models include a multi-modal general-purpose large model and a seed-adaptive model. The Fuxi Brain, as the intelligent decision-making core of the system, relies on the capabilities of the large model (such as multi-dimensional data processing and knowledge reasoning of the multi-modal general-purpose large model) to generate a dynamic optimal matrix [X,Y,…], and sends it to the dynamic optimal matrix dynamic MOA (execution network). The dynamic MOA is the execution carrier for multi-agent collaboration, used to respond to the control of the Fuxi Brain, adjust its own network parameters, and transform abstract decisions into a set of executable collaborative instructions. The multiple agents include a management agent and a multi-functional agent. The management agent, as the coordinator of agent collaboration, is responsible for receiving global decisions from the Fuxi Brain, breaking down tasks and allocating them to the multi-functional agents, while also collecting execution feedback and optimizing the collaboration process. The multi-functional agent, as the execution unit, possesses memory and dynamic collaboration capabilities, covering specific task execution.

[0100] This invention includes the following key technical points:

[0101] Key Point 1: Multi-Agent Hybrid Model Architecture. A hybrid model architecture is constructed by combining the generalization and reasoning capabilities of a large language model with agricultural algorithm models. This enhances the agent's adaptability to complex agricultural environments (such as multivariate meteorological coupling and soil heterogeneity), addressing the prediction bias problem of traditional single models in nonlinear dynamic scenarios. Leveraging the multimodal data processing capabilities of the large language model, it enables the fusion analysis of unstructured data (such as crop growth images) and structured data (such as sensor values), improving the decision-making dimensionality.

[0102] Key Point 2: Dynamic Sensing and Multimodal Data Fusion Mechanism. The IoT acquisition layer integrates heterogeneous data from multiple sources, including weather stations, soil sensors, and UAV remote sensing, and performs spatiotemporal alignment and outlier correction through a specialized algorithm layer. This achieves millisecond-level updates of farmland environmental parameters (e.g., soil moisture monitoring accuracy of ±2%), avoiding irrigation delays caused by traditional periodic sampling. Multimodal data cross-validation (e.g., matching remote sensing thermal imaging with ground sensor data) reduces the risk of misjudgments caused by the failure of a single data source.

[0103] Key Point 3: Enhanced Reasoning in the Global Knowledge Enhancement Layer. A knowledge graph is constructed based on a historical irrigation case database and the experience of agronomic experts, and reinforcement learning is used to optimize the agent's decision-making path. In sudden weather events (such as heavy rain), the reasoning response time is reduced to within 3 seconds, an 80% improvement over traditional systems. Knowledge distillation technology is used to compress large language model parameters, reducing the memory footprint of edge computing devices (such as field controllers) by 40%, making it suitable for low-computing-power scenarios.

[0104] Key Point 4: Information Sharing and Collaborative Decision-Making Mechanism. A distributed architecture is adopted to achieve real-time data exchange between agents, and a game theory strategy is introduced to balance the decision weights of each agent. This resolves multi-objective conflict problems (such as the contradiction between water conservation needs and high-yield goals), comprehensively optimizes irrigation schemes, and reduces redundant computational resource consumption through a dynamic priority adjustment mechanism (e.g., agents targeting planting during the growing season have the highest weight during the flowering stage).

[0105] Key Point 5: Adaptive Parameter Adjustment Mechanism. A fuzzy control rule base based on crop growth stages is designed, combined with a large language model to predict environmental change trends, enabling dynamic calibration of control parameters. Irrigation strategies are automatically switched at different crop growth stages (e.g., seedling stage, maturity stage), further improving water and fertilizer utilization efficiency.

[0106] To better understand the technical solution of this invention, a specific embodiment is provided. From October 2023 to December 2024, an autonomous irrigation method based on a large language model and multi-agent control was demonstrated on 200 mu of land in Shangkuli Farm, Hulunbuir State Farm, achieving the following significant technical effects:

[0107] (1) In terms of yield: the average yield per mu increased by 24.33% compared with dryland planting, and the benefit increased by 112.42 yuan / mu; compared with sprinkler irrigation planting, the average yield per mu increased by 2.27%, and the benefit increased by 12.76 yuan / mu, as shown in Table 1.

[0108]

[0109] Table 1. Production Increase Effect

[0110] (2) Water saving: Compared with sprinkler irrigation, the average water saving per mu is 36.7%, as shown in Table 2.

[0111]

[0112] Table 2 Water-saving effect

[0113] (3) In terms of saving manpower: Compared with sprinkler irrigation, it saves a total of 47.5% of manpower and saves 57 yuan per mu, as shown in Table 3.

[0114]

[0115] Table 3. Labor Saving Effect

[0116] (3) In terms of total income: compared with dryland planting, the income per mu increased by RMB43.79; compared with sprinkler irrigation planting, the income per mu increased by RMB109.2, as shown in Table 4.

[0117]

[0118] Table 4 Total Revenue

[0119] It should be noted that the intelligent irrigation method based on a large language model provided in this embodiment of the invention can be executed by an electronic device, a apparatus, or a control module within that apparatus for executing the method. This embodiment of the invention uses an apparatus executing the method as an example to illustrate the intelligent irrigation apparatus based on a large language model provided in this embodiment of the invention.

[0120] The intelligent irrigation device based on a large language model provided in this invention can achieve... Figures 1-2 The various processes implemented in the embodiment of the intelligent irrigation method based on a large language model shown are not described in detail here to avoid repetition.

[0121] Figure 5 This is a schematic diagram of the structure of an intelligent irrigation device based on a large language model provided in Embodiment 5 of the present invention. The device 100 includes an agent initialization module 10, a data acquisition and analysis module 20, an information sharing and collaboration module 30, and a dynamic control execution module 40, wherein:

[0122] The agent initialization module 10 is used to define multiple agents in a large language model, with each agent adopting a hybrid model architecture.

[0123] The data acquisition and analysis module 20 is used to acquire multimodal data from each agent and perform fusion analysis on the multimodal data through a hybrid model architecture to obtain the respective analysis results. Preferably, the multimodal data includes environmental information and / or crop information. More preferably, the environmental information includes meteorological information and / or soil information. The crop information includes the crop growth stage.

[0124] The information sharing and collaboration module 30 is used to comprehensively process the multimodal data and analysis results uploaded by each intelligent agent and generate corresponding irrigation instructions.

[0125] The dynamic control execution module 40 is used to regulate the irrigation equipment to perform automatic irrigation according to the irrigation command.

[0126] Figure 6 This is a schematic diagram of the structure of an intelligent irrigation device based on a large language model provided in Embodiment Six of the present invention. The device 100 includes an agent initialization module 10, a data acquisition and analysis module 20, an information sharing and collaboration module 30, a dynamic control execution module 40, a reasoning reinforcement module 50, and / or an adaptive adjustment module 60, wherein:

[0127] The agent initialization module 10 is used to define multiple agents in a large language model, each agent adopting a hybrid model architecture. Preferably, the multiple agents include multiple analysis agents 200, management agents 300, and irrigation agents 400, such as... Figure 3 As shown.

[0128] The data acquisition and analysis module 20 is used for each intelligent agent to acquire multimodal data, and to perform fusion analysis on the multimodal data through a hybrid model architecture to obtain their respective analysis results. Preferably, the multimodal data includes environmental information and / or crop information. More preferably, the environmental information includes meteorological information and / or soil information. The crop information includes crop growth stages. Preferably, each analytical intelligent agent acquires the required multimodal data through the IoT acquisition layer, and performs fusion analysis on the multimodal data through a hybrid model architecture to obtain its own analysis results.

[0129] Better yet, the analytical agent performs the following actions:

[0130] (1) Each analytical agent acquires the required multimodal data through the Internet of Things acquisition layer.

[0131] (2) Each analytical agent performs spatiotemporal alignment and outlier correction on multimodal data through a professional algorithm layer.

[0132] (3) Each analytical agent performs cross-validation on the corrected multimodal data.

[0133] The information sharing and collaboration module 30 is used to comprehensively process the multimodal data and analysis results uploaded by each agent to generate corresponding irrigation instructions. Preferably, each analysis agent uploads the multimodal data and analysis results to the information sharing layer for sharing. The management agent comprehensively processes the analysis results uploaded by each analysis agent, balances the decision weights of each agent through game theory strategies, and generates corresponding irrigation instructions.

[0134] The dynamic control execution module 40 is used to regulate the irrigation equipment for automatic irrigation according to irrigation instructions. Preferably, the irrigation intelligent agent regulates the irrigation equipment for automatic irrigation according to irrigation instructions.

[0135] The reasoning enhancement module 50 is used to enhance the reasoning of each agent through the global knowledge enhancement layer.

[0136] Preferably, the reasoning reinforcement step module 50 constructs a knowledge graph based on a historical irrigation case database and the experience of agronomic experts through a global knowledge enhancement layer, and optimizes the decision-making path of each agent through reinforcement learning.

[0137] The adaptive adjustment module 60 is used to adaptively adjust the analysis and control parameters of each agent by comparing historical data with real-time monitoring data for different crop growth stages and / or environmental information.

[0138] Preferably, the plurality of intelligent agents includes a plurality of analytical intelligent agents 200, management intelligent agents 300, and irrigation intelligent agents 400, such as... Figure 3 As shown. More preferably, the plurality of analytical agents 200 include:

[0139] The future weather intelligence agent 210 is used to analyze weather trend analysis results based on collected weather information.

[0140] The Soil Moisture Intelligence Agent 220 is used to analyze soil moisture analysis results based on collected soil information.

[0141] The Soil Nutrient Intelligence Agent 230 is used to analyze soil nutrient content based on collected soil information.

[0142] The growth analysis agent 240 is used to analyze the crop growth status based on the collected crop information and / or environmental information.

[0143] The growth period target agent 250 is used to analyze the crop growth stage and set corresponding stage target analysis results based on the collected crop information and / or environmental information.

[0144] The intelligent irrigation device based on a large language model provided in this invention includes: an agent initialization module for defining multiple agents based on a large language model, each agent adopting a hybrid model architecture; a data acquisition and analysis module for each agent to acquire multimodal data and perform fusion analysis on the multimodal data through the hybrid model architecture to obtain their respective analysis results; an information sharing and collaboration module for comprehensively processing the multimodal data and analysis results uploaded by each agent to generate corresponding irrigation instructions; and a dynamic control execution module for regulating the irrigation equipment for automatic irrigation according to the irrigation instructions. Thus, this invention utilizes the reasoning ability of a large language model and the collaborative characteristics of a multi-agent architecture to not only overcome the limitations of single-agent decision-making but also to achieve comprehensive monitoring of the farmland environment and intelligent irrigation decision-making through multi-agent collaboration, thereby enhancing the system's adaptability.

[0145] The present invention also provides a storage medium for storing, for example, Figures 1-4 The computer program for any of the intelligent irrigation methods based on a large language model. For example, computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the present invention through the operation of the computer, and can achieve the same technical effect. To avoid repetition, they will not be described in detail here. The program instructions for invoking the methods of the present invention may be stored in a fixed or removable storage medium, and / or transmitted through data streams in broadcast or other signal carrying media, and / or stored in the storage medium of a computer device operating according to the program instructions.

[0146] According to one embodiment of the present invention, the present invention also provides such a Figure 7The illustrated electronic device 400 may optionally include a storage medium 200 for storing a computer program and a processor 300 for executing the computer program. When the computer program is executed by the processor 300, it implements any of the aforementioned intelligent irrigation methods based on a large language model, triggering the electronic device 400 to execute methods and / or technical solutions based on the foregoing embodiments, achieving the same technical effect. To avoid repetition, further details are omitted here. It should be noted that the electronic devices in this embodiment include mobile electronic devices and non-mobile electronic devices. For example, mobile electronic devices may be mobile phones, tablets, laptops, PDAs, in-vehicle electronic devices, wearable devices, super mobile personal computers, netbooks, or personal digital assistants, etc., while non-mobile electronic devices may be servers, network-attached storage (NAS), personal computers (PCs), televisions (TVs), ATMs, or self-service machines, etc. This embodiment does not specifically limit the scope of the invention.

[0147] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present invention (including associated data structures) can be stored in a computer-readable recording medium, such as RAM memory, a magnetic or optical drive, a floppy disk, or similar devices. Furthermore, some steps or functions of the present invention can be implemented in hardware, for example, as circuitry that works with a processor to perform the various steps or functions.

[0148] This invention can be implemented on a computer as a computer-based method, or in dedicated hardware, or a combination of both. Executable code or portions thereof for the method according to the invention can be stored on a computer program product. Examples of computer program products include memory devices, optical storage devices, integrated circuits, servers, online software, etc. Optionally, the computer program product includes non-transitory program code components stored on a computer-readable medium so as to execute the method according to the invention when the program product is executed on a computer.

[0149] In an optional embodiment, the computer program includes computer program code components adapted to perform all the steps of the method according to the invention when the computer program is run on a computer. Optionally, the computer program is embodied on a computer-readable medium.

[0150] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of the present invention is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0151] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A smart irrigation method based on a large language model, characterized in that, include: The agent initialization step involves defining multiple agents based on a large language model, with each agent employing a hybrid model architecture. In the data acquisition and analysis step, each of the intelligent agents acquires multimodal data and performs fusion analysis on the multimodal data through the hybrid model architecture to obtain their respective analysis results; The information sharing and collaboration step involves comprehensively processing the multimodal data uploaded by each intelligent agent and the analysis results to generate corresponding irrigation instructions; The dynamic control execution steps regulate the irrigation equipment to perform automatic irrigation according to the irrigation command; The multimodal data includes environmental information and / or crop information; The intelligent agent includes multiple analytical intelligent agents, management intelligent agents, and irrigation intelligent agents; The data acquisition and analysis steps include: Each of the aforementioned analytical agents acquires the required multimodal data through the Internet of Things acquisition layer, and performs fusion analysis on the multimodal data through the hybrid model architecture to obtain its own analysis results; The information sharing and collaboration steps include: Each of the aforementioned analytical agents uploads the multimodal data and the analysis results to the information sharing layer for sharing; The management agent comprehensively processes the analysis results uploaded by each analysis agent, balances the decision weights of each agent through game theory strategies, and generates the corresponding irrigation instructions. The dynamic control execution steps include: The irrigation agent controls the irrigation equipment to perform automatic irrigation according to the irrigation command; The steps by which each of the aforementioned analytical agents acquires the multimodal data required by the IoT acquisition layer include: Each of the aforementioned analytical agents acquires the required multimodal data through the IoT acquisition layer; Each of the aforementioned analytical agents performs spatiotemporal alignment and outlier correction on the multimodal data through a specialized algorithm layer; Each of the analytical agents performs cross-validation on the corrected multimodal data; The method further includes: The reasoning enhancement step involves enhancing the reasoning of each agent through a global knowledge enhancement layer. The crop information includes the crop growth stage; The method further includes: In the adaptive adjustment step, for different crop growth stages and / or environmental information, each intelligent agent adaptively adjusts the analysis and control parameters by comparing historical data with real-time monitoring data.

2. The intelligent irrigation method based on a large language model according to claim 1, characterized in that, The environmental information includes meteorological information and / or soil information; The plurality of analytical agents include: The future meteorological intelligent agent will be used to analyze the collected meteorological information to produce meteorological trend analysis results. The soil moisture intelligence agent is used to analyze soil moisture analysis results based on the collected soil information. A soil nutrient intelligence agent is used to analyze the soil nutrient content based on the collected soil information. A growth analysis agent is used to analyze the crop growth status based on the collected crop information and / or environmental information. A growth period target intelligent agent is used to analyze the crop growth stage and set corresponding stage target analysis results based on the collected crop information and / or environmental information.

3. The intelligent irrigation method based on a large language model according to claim 1, characterized in that, The reasoning enhancement step further includes: A knowledge graph is constructed based on a historical irrigation case database and the experience of agronomic experts through a global knowledge enhancement layer, and the decision-making paths of each agent are optimized through reinforcement learning.

4. A smart irrigation device based on a large language model constructed according to the method described in any one of claims 1 to 3, characterized in that, The device includes: The agent initialization module is used to define multiple agents in a large language model, with each agent adopting a hybrid model architecture. The data acquisition and analysis module is used to acquire multimodal data from each of the aforementioned intelligent agents, and to perform fusion analysis on the multimodal data through the hybrid model architecture to obtain their respective analysis results; The information sharing and collaboration module is used to comprehensively process the multimodal data and analysis results uploaded by each of the intelligent agents to generate corresponding irrigation instructions; The dynamic control execution module is used to regulate the irrigation equipment to perform automatic irrigation according to the irrigation command.

5. A storage medium, characterized in that, Used to store a computer program for performing the method according to any one of claims 1 to 3.

6. An electronic device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 3.

Images