Intelligent agent cooperation method and device, system, and agricultural task execution method and device

Abstract

The disclosure provides an agent cooperation method and device, system, and agricultural task execution method and device, and relates to the technical fields of artificial intelligence, natural language processing, computer vision, and the like. The specific implementation scheme is as follows: multi-source agricultural data of a crop in a historical period and a current growth stage is collected; a spatio-temporal feature graph is constructed based on the multi-source agricultural data; a crop stress index and a crop growth index are calculated based on the spatio-temporal feature graph; a target agent is determined from a set of agents based on the crop stress index and the crop growth index; the crop growth index and the crop stress index are sent to the target agent, so that the target agent executes a corresponding agricultural task on the crop based on the crop growth index and the crop stress index, and relevant agents in the set of agents are linked based on environmental data, thereby improving the globality and collaboration of agricultural decision-making.

Description

TECHNICAL FIELD

[0001] The present disclosure belongs to the technical field of computers, and particularly relates to the technical fields of natural language processing, artificial intelligence, computer vision, etc. In particular, an agent collaboration method and device, system, agricultural task execution method and device, electronic equipment, and computer readable storage medium. BACKGROUND

[0002] With the development of agricultural modernization, crop models and agricultural information systems play an increasingly important role in agricultural production. Traditional crop growth models are mainly based on mechanism modeling methods. These models establish mathematical equations based on crop physiological and ecological principles to simulate the entire growth process of crops from sowing to harvesting. However, such models rely on a large number of parameter calibrations and usually require expert experience, making it difficult to adapt to complex environmental changes and large-scale applications.

[0003] In recent years, with the development of artificial intelligence, especially deep learning, some agricultural AI models have been proposed to solve specific problems, such as using convolutional neural networks for pest and disease identification, using time series models for yield prediction, and monitoring soil moisture based on remote sensing data. However, these "small models" are often optimized for a specific task and lack cross-task collaboration capabilities, forming so-called "task islands."

[0004] In addition, traditional models lack self-adaptive ability when facing dynamic environmental changes and cannot automatically adjust model parameters to cope with differences in different regions or climate conditions. At the same time, due to the diversity of agricultural data sources and the complexity of the structure, how to effectively integrate multi-modal heterogeneous data such as remote sensing, Internet of Things sensors, weather stations, and field management logs has become a difficulty in current research. SUMMARY

[0005] The present disclosure provides an agent collaboration method and device, system, agricultural task execution method and device, electronic equipment, and computer readable storage medium.

[0006] According to a first aspect, an agent collaboration method is provided, which includes: collecting multi-source agricultural data of a crop in a historical period and a current growth stage; based on the multi-source agricultural data, constructing a spatio-temporal feature graph; based on the spatio-temporal feature graph, calculating a crop stress index and a crop growth index; based on the crop stress index and the crop growth index, determining a target agent from a set of agents; sending the crop growth index and the crop stress index to the target agent, so that the target agent performs a corresponding agricultural task on the crop based on the crop growth index and the crop stress index, and links relevant agents in the set of agents based on environmental data.

[0007] According to a second aspect, an agricultural task execution method is provided, the method comprising: receiving a crop growth index and a crop stress index of a crop; performing an agricultural task on the crop based on the crop growth index and the crop stress index; obtaining environmental data when performing the agricultural task; and linking a relevant agent in a set of agents based on the environmental data.

[0008] According to a third aspect, an agent coordination apparatus is provided, the apparatus comprising: an acquisition unit configured to acquire multi-source agricultural data of a crop in a historical period and a current growth stage; a graph construction unit configured to construct a spatio-temporal feature graph based on the multi-source agricultural data; an index construction unit configured to calculate a crop stress index and a crop growth index based on the spatio-temporal feature graph; a determination unit configured to determine a target agent from a set of agents based on the crop stress index and the crop growth index; and a sending unit configured to send the crop growth index and the crop stress index to the target agent, so that the target agent performs a corresponding agricultural task on the crop based on the crop growth index and the crop stress index, and links a relevant agent in the set of agents based on environmental data.

[0009] According to a fourth aspect, an agricultural task execution apparatus is provided, the apparatus comprising: a receiving unit configured to receive a crop growth index and a crop stress index of a crop; an execution unit configured to perform an agricultural task on the crop based on the crop growth index and the crop stress index; an obtaining unit configured to obtain environmental data when performing the agricultural task; and a linking unit configured to link a relevant agent in a set of agents based on the environmental data.

[0010] According to a fifth aspect, an agent coordination system is provided, the system comprising: a central agent and a set of agents; the central agent is configured to acquire multi-source agricultural data of a crop in a historical period and a current growth stage; construct a spatio-temporal feature graph based on the multi-source agricultural data; calculate a crop stress index and a crop growth index based on the spatio-temporal feature graph; determine a target agent from the set of agents based on the crop stress index and the crop growth index; and send the crop growth index and the crop stress index to the target agent; the target agent performs a corresponding agricultural task on the crop based on the crop growth index and the crop stress index, and links a relevant agent in the set of agents based on environmental data.

[0011] According to a sixth aspect, an electronic device is provided, the electronic device comprising: at least one processor; and a memory connected with the at least one processor in communication, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in any implementation manner of the first aspect or the second aspect.

[0012] According to a seventh aspect, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method according to any implementation of the first aspect or the second aspect.

[0013] The embodiment of the present disclosure provides an agent cooperation method and device. Firstly, multi-source agricultural data of a crop in a historical period and a current growth stage are collected. Secondly, a spatio-temporal feature graph is constructed based on the multi-source agricultural data. Thirdly, a crop stress index and a crop growth index are calculated based on the spatio-temporal feature graph. Fourthly, a target agent is determined from a set of agents based on the crop stress index and the crop growth index. Finally, the crop stress index and the crop growth index are sent to the target agent, so that the target agent performs a corresponding agricultural task on the crop based on the crop stress index and the crop growth index, and links relevant agents in the set of agents based on environmental data. The method can accurately reflect the growth state and stress situation of the crop by constructing the spatio-temporal feature graph based on the multi-source agricultural data and further generating the crop stress index and the crop growth index. The target agent is determined from the initial agents based on the indexes, and relevant data is sent to the target agent, so that the target agent can perform the agricultural task according to the accurate data, and relevant agents in the set of agents are linked, realizing the intelligent, accurate and collaborative of agricultural production. This not only improves the efficiency of agricultural production and reduces the cost of manpower, but also can timely deal with various problems in the growth process of the crop, improve the yield and quality of the crop, and enhance the stability and sustainability of agricultural production, providing strong support for the development of smart agriculture.

[0014] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0015] The accompanying drawings are used to better understand the present scheme and do not limit the present disclosure. Among them:

[0016] Figure 1 is a flowchart of an embodiment of the agent cooperation method according to the present disclosure;

[0017] Figure 2 is a structural schematic diagram of the cooperation between the set of agents in the present disclosure;

[0018] Figure 3 is a flowchart of an embodiment of the agricultural task execution method according to the present disclosure;

[0019] Figure 4 is a structural schematic diagram of an embodiment of the agent cooperation device according to the present disclosure;

[0020] Figure 5 is a structural schematic diagram of one embodiment of the agricultural task execution device of the present disclosure;

[0021] Figure 6 is a structural schematic diagram of an intelligent agent coordination system of the present disclosure;

[0022] Figure 7 is a block diagram of an electronic device for implementing an intelligent agent coordination method of an embodiment of the present disclosure. DETAILED DESCRIPTION

[0023] Unless otherwise specifically defined, the term "comprises" or its variations such as "comprising" or "comprises" as used throughout this specification and claims, are to be construed as open-ended, including the stated elements or components, but not excluding additional elements or components.

[0024] The technical solutions of the present disclosure are described below through specific embodiments. It should be understood that the one or more steps mentioned in the present disclosure do not exclude other methods and steps before and after the combination steps, or other methods and steps can be inserted between these explicitly mentioned steps. It should also be understood that these examples are only used to illustrate the present disclosure and do not limit the scope of the present disclosure. Unless otherwise specified, the numbering of each method step is only for the purpose of identifying each method step, and is not limited to the arrangement order of each method or the scope of the implementation of the present disclosure. Changes or adjustments of the relative relationship, without substantial changes in technical content, can also be considered as the scope of the implementation of the present disclosure.

[0025] The raw materials and instruments used in the embodiments are not specifically limited in source, and can be purchased on the market or prepared according to conventional methods well known to those skilled in the art.

[0026] In the prior art, the problems faced by the agricultural field are:

[0027] Weak data fusion capability: Agricultural data sources are diverse, including remote sensing images, sensors, weather data, and farm operation records. Existing models have difficulty in efficiently fusing these multi-modal data, affecting the model prediction accuracy and decision quality.

[0028] Task islandization is serious: existing agricultural models run independently, such as plant protection models for pest control, irrigation models for water management, and nutrition models for nitrogen balance, lacking information exchange and collaborative decision-making mechanisms between each other, making it difficult to form a globally optimal planting management strategy.

[0029] Strong dependence on artificial: Model construction and parameter calibration are highly dependent on expert experience, lack of real-time data-based self-feedback and correction ability, leading to poor model adaptability, difficult to cope with climate change and regional differences. Traditional crop models are based on limited experimental data fitting, which is difficult to accurately simulate the complex crop-environment interaction relationship, especially under atypical climate conditions.

[0030] In view of the defects in the prior art, the present disclosure proposes an agent cooperation method, which fully utilizes the data fusion capability, and the information interaction and collaborative decision mechanism between each other. Based on the real-time data-based self-feedback and correction ability, the agricultural production efficiency is improved, the labor cost is reduced, and various problems in the crop growth process can be timely responded to, the crop yield and quality are improved, the stability and sustainability of agricultural production are enhanced, and strong support is provided for the development of smart agriculture. Figure 1 The flow 100 of one embodiment of the agent cooperation method according to the present disclosure is shown, and the agent cooperation method comprises the following steps:

[0031] Step 101, collect multi-source agricultural data of crops in historical period and current growth stage.

[0032] In this embodiment, the multi-source agricultural data refers to a data set from different sources, different types and related to crops in historical period and current growth stage. These data may include meteorological data (such as temperature, humidity, rainfall, etc.), soil data (such as soil fertility, pH, texture, etc.), remote sensing image data (for monitoring crop growth, land use, etc.), agricultural production activity data (such as sowing time, fertilizer amount, irrigation record, etc.) and market data (such as agricultural product price, market demand, etc.). The integration of multi-source agricultural data can provide a more comprehensive perspective and richer information for agricultural production, which is helpful to realize precision agricultural management and decision support.

[0033] In this embodiment, the agent cooperation method runs on the center agent which can control each agent in the intelligent agent set. Through the coordination of the center agent and the collaborative work between the related agents, the intelligent control of crops can be realized.

[0034] In this embodiment, the collection of multi-source agricultural data of the crop in the historical period and the current growth stage can be realized by integrating various technologies and means. First, by screening historical data, multi-source agricultural data of the crop in the historical period is obtained; satellite remote sensing technology is used to obtain macro data such as soil moisture, vegetation index, and land use type of the crop in the current growth stage; at the same time, through the deployment of Internet of Things sensor network in farmland, real-time collection of soil temperature, humidity, nutrient content, and meteorological data such as air temperature, precipitation, and light intensity of the crop in the current growth stage is realized; in addition, combined with unmanned aerial remote sensing, high-resolution farmland images are taken for accurate identification of the growth status and pest distribution of the crop in the current growth stage; through field research by agricultural experts, human management data such as crop varieties, planting modes, and irrigation and fertilization records are collected; finally, these data from different channels are converged to an agricultural big data platform, and after cleaning, fusion, and analysis, comprehensive and accurate data support is provided for agricultural production decision-making, helping the development of intelligent agriculture.

[0035] In step 102, a spatio-temporal feature map is constructed based on multi-source agricultural data.

[0036] In this embodiment, the spatio-temporal feature map is a tool or model that visualizes and analyzes the features of data in time and space dimensions. The spatio-temporal feature map can show the rules and patterns of data changes over time in different geographical locations (spatial dimension). For example, in the field of agriculture, the spatio-temporal feature map can show the distribution of crop growth conditions in a region changing with the seasons, or the changing trend of soil fertility in different plots in different years. By constructing the spatio-temporal feature map, key factors affecting agricultural production and their dynamic changes can be more intuitively identified, thereby providing scientific basis for the optimal allocation of agricultural resources, pest warning, yield prediction, etc.

[0037] In this embodiment, based on multi-source agricultural data, the spatio-temporal feature map is constructed. First, the data from different sources needs to be standardized to ensure consistency in data format, unit, and time resolution. Then, the geographic information system (GIS) is used to associate the data with geographic location information to form a spatialized first data set. Next, through time series analysis method, the changing characteristics of the first data set in time dimension are extracted, such as seasonal fluctuations and long-term trends. Finally, the spatial and temporal features are combined, and the first data set is generated into a spatio-temporal feature map using visualization tools (such as maps and charts), which intuitively shows the dynamic change rules of agricultural data in different regions and time periods, providing strong support for agricultural production decision-making.

[0038] Optionally, the step 102 further comprises: before constructing the spatio-temporal feature map, associating the multi-source agricultural data with the agricultural knowledge graph to obtain associated data, standardizing the associated data, associating the associated data with geographic location information using a geographic information system to form a spatialized second data set; extracting the variation characteristics of the second data set in the time dimension by a time series analysis method. Finally, the spatial and temporal characteristics are combined, and the second data set is generated into a spatio-temporal feature map using a visualization tool.

[0039] Step 103, based on the spatio-temporal feature map, calculating the crop stress index and the crop growth index.

[0040] In this embodiment, the crop stress index (CSI) is an index for quantifying the degree of stress that crops are subjected to. Stress refers to the adverse environmental factors (such as drought, high temperature, salinity, pests and diseases, etc.) that crops are subjected to during growth. The higher the stress index, the more severe the stress that crops are subjected to, and the greater the negative impact on their growth and development.

[0041] In this embodiment, the crop growth index (CGI) is an index for measuring the growth status and health level of crops. It usually combines the physiological characteristics of crops (such as chlorophyll content, leaf area, biomass, etc.) and the comprehensive influence of environmental factors (such as light, temperature, moisture, etc.) on crop growth. The higher the growth index, the better the growth status and the higher the health level of crops.

[0042] In this embodiment, the CSI and CGI are time series variation indexes with a range of 0-1. In CGI, 0 represents the beginning of seed germination after sowing, and 1 represents the maturity of crops. In CSI, 0 represents severe stress that cannot grow normally, and 1 represents normal growth of crops without stress. The plant protection, irrigation, nutrition, and yield estimation agent dynamically adjusts the strategy according to CSI and CGI. During the growth of crops, CGI changes from 0 to 1, and the corresponding plant protection scheme, irrigation scheme, nutrition scheme, and dynamic yield estimation are triggered as the value changes.

[0043] In this embodiment, based on the spatiotemporal feature map, the crop stress index and the crop growth index are calculated. First, key spatiotemporal feature parameters such as soil moisture, temperature, light intensity, leaf color, etc. are extracted from the map. Then, according to the known relationship between these parameters and crop growth and stress, the corresponding mathematical model is established. For example, by analyzing the degree of inhibition of crop growth when soil moisture is below a certain threshold, the contribution weight of drought stress to the index is determined; at the same time, the health status of the crop is evaluated by combining the spectral reflectance change of leaf color, so as to assign values to the growth index. Finally, these weighted parameters are integrated to form an index value that can reflect the overall stress and growth status of the crop.

[0044] Step 104, based on the crop stress index and the crop growth index, the target agent is determined from the agent set.

[0045] In this embodiment, the execution subject dynamically adjusts resources based on the real-time changes of the crop stress index CSI and the crop growth index CGI. There are two ways to perform resource scheduling, both of which are applied. One is: based on the growth index CGI to determine the growth stage of the crop, based on the CGI index, the corresponding agent task is called to perform plant protection, irrigation, nutrition, and dynamic yield estimation measures in different stages of crop growth; the other is when the crop stress index CSI is less than 1, according to the growth stage corresponding to the crop growth index, the corresponding target agent is triggered.

[0046] In this embodiment, an agent is a system or model that can perceive, decide, and execute specific tasks. An agent can be a sensor network that monitors the health status of crops, an algorithm model that analyzes satellite remote sensing data, or a decision system that contains pre-set rules and algorithms.

[0047] In this embodiment, the crop stress index and the crop growth index are two key indicators for evaluating agents. The crop stress index may reflect the impact of adverse environmental factors (such as drought, pests and diseases, etc.) on crop growth during the growth process, and the higher the value, the greater the stress degree; while the crop growth index measures the growth status and health level of the crop, and the higher the value, the better the growth condition. In specific implementation, the system will consider these two indexes according to the pre-set rules or algorithms, evaluate and screen the initial agent, and get the target agent. For example, it may prefer to select an agent with a lower stress index and a higher growth index as the target agent, because it means that the agent can maintain a good growth state in the face of adverse environments, and is considered as the optimal choice. The agent set includes multiple agents, such as Figure 2As shown, the intelligent agents include: a plant protection intelligent agent, an irrigation intelligent agent, a nutrition intelligent agent, and a yield intelligent agent, wherein the plant protection intelligent agent is used for disease identification and disease control; the irrigation intelligent agent is used for balancing water and providing an irrigation plan for crops; the nutrition intelligent agent is used for balancing nutrients and providing fertilization recommendations; and the yield intelligent agent is used for yield prediction of crops and generation of a yield map of crops in each region. The target intelligent agent is at least one intelligent agent with a specific function selected from the plurality of intelligent agents by the central intelligent agent, that is, the target intelligent agent includes one or more intelligent agents each having a corresponding function.

[0048] The plant protection intelligent agent monitors the occurrence and formation of diseases and pests based on a spore instrument, a pest condition instrument, and an image recognition algorithm, predicts disease spread risks in combination with meteorological data, and generates a prevention plan in advance. The plant protection intelligent agent is triggered according to the CSI and CGI indexes, and corresponding agricultural tasks are performed in different stages as follows: 0: after sowing to before emergence - weed control; 0.1: emergence stage - synergistic effect and weed control; 0.3: jointing stage - aphids, Diabrotica virgifera, corn borer, leaf diseases, and aerial application adjuvants; 0.5: silk emergence stage - stress resistance and quality improvement, aphids, Diabrotica virgifera, corn borer, and leaf diseases. In addition, the plant protection intelligent agent also runs a disease and pest monitoring and early warning program according to the CSI index, monitoring equipment, and meteorological factors, and performs targeted disease and pest control, so that the above-mentioned agricultural treatments are more professional. The agricultural tasks of the plant protection intelligent agent also include matching disease types with compatibility of control agents, constructing a corresponding knowledge base for the disease types corresponding to the diseases and pests in the conventional agricultural treatment and the early warning, and containing information such as disease and pest category introduction, occurrence law, and control scheme.

[0049] The agricultural tasks of the irrigation intelligent agent include: real-time analysis of soil water content, meteorological data, and crop water requirement curve; combination of crop stress index and growth index to analyze the relationship between crop water requirement and soil water content, dynamic adjustment of irrigation plan; output of irrigation recommendation to the control system to realize automatic irrigation scheduling. The irrigation intelligent agent is triggered according to different CSI and CGI, the transpiration of crops is calculated according to meteorological data, and the irrigation amount required by crops is calculated according to the growth of crops and soil water content as shown in formula (1).

[0050] Irrigation amount = Kc·ETo·A·η (1)

[0051] In formula (1), Kc is a crop coefficient (reflecting the demand characteristics of crops), and Kc is different for different CSI; A is the irrigation area (m 2 ); η is the irrigation efficiency (the efficiency of drip irrigation is 0.9); ETo is obtained according to the standard method of Penman-Monteith for calculating the reference crop evapotranspiration.

[0052] The nutrition intelligent agent agricultural task includes: analyzing soil nutrient content, crop growth stage, target yield; combining stress index and growth index to analyze crop nutrient and soil nutrient supply relationship, dynamically recommending fertilizer type, dosage and application time; fusing remote sensing NDVI data to judge leaf nutrition status, and assisting in diagnosing nutrient deficiency symptoms. The nutrition intelligent agent is triggered according to CGI and CSI. The nitrogen content of crop leaves is dynamically changed at different growth stages of crops. The normal nitrogen content standard of each CGI stage is calculated according to the historical nitrogen content level of crop leaves, and then the current nitrogen content level of crop leaves is compared to give a fertilizer recommendation, which is increased or decreased on the basis of normal topdressing amount.

[0053] The yield estimation intelligent agent is based on crop growth model and historical yield data to train a prediction model. Its agricultural tasks include: dynamically predicting yield combining with current climate conditions, management measures, stress index and other factors; fusing remote sensing NDVI data to support multi-scale prediction (plot level, regional level). The dynamic yield estimation intelligent agent is started during the grain formation period according to the change value of CGI and CSI, and NDVI, weather, CSI and management data are used to evaluate the yield change.

[0054] Optionally, the step 104 includes: detecting whether the crop is in a stress state based on the crop stress index; in response to detecting that the crop is in a stress state, selecting an intelligent agent that can eliminate the stress state from the intelligent agent set; at the same time, determining the growth state of the crop based on the crop growth index, and selecting an intelligent agent that is beneficial to the maturation of the crop from the intelligent agent set, and taking the intelligent agent that eliminates the stress state and the intelligent agent that is beneficial to the maturation of the crop as the target intelligent agent.

[0055] Step 105, sending the crop growth index and the crop stress index to the target intelligent agent, so that the target intelligent agent executes corresponding agricultural tasks on the crop based on the crop growth index and the crop stress index, and links related intelligent agents in the intelligent agent set based on environmental data.

[0056] In this embodiment, the target intelligent agent refers to an intelligent device or software system with autonomous decision-making and execution capability in the agricultural system, such as an intelligent irrigation system, an automatic fertilization robot, an agricultural unmanned aerial vehicle, etc. It can receive data and execute corresponding agricultural tasks on crops according to preset rules or algorithms.

[0057] In this embodiment, the environmental data refers to various data related to the growth environment of crops, such as soil moisture, temperature, light intensity, carbon dioxide concentration, etc. These data are collected by sensors and transmitted to the system for evaluating whether the environmental conditions are suitable for crop growth.

[0058] In this embodiment, the relevant agent refers to other intelligent devices or systems in the agent set that work with the target agent, such as meteorological monitoring agents, greenhouse control agents, etc. The relevant agent and the target agent can achieve collaborative work through data sharing and linkage mechanism. It should be noted that the target agent is determined each time, and the corresponding relevant agent is also different.

[0059] In this embodiment, in the agricultural automation management system, the growth data and environmental data of crops are collected by sensors, and the crop growth index and the crop stress index are calculated. These indexes are sent to the target agent, such as the intelligent irrigation system. The target agent determines whether the crop is in the vigorous growth period according to the crop growth index, and determines whether the crop is water-deficient or subjected to other stress according to the crop stress index. If it is found that the crop is water-deficient, the target agent will automatically start the irrigation task, and at the same time, the environmental data (such as soil humidity) is sent to the relevant agent, such as the greenhouse control agent. The greenhouse control agent adjusts the ventilation and sunshade equipment according to the soil humidity data to optimize the environmental conditions in the greenhouse, thereby realizing the linkage between multiple agents and jointly ensuring the healthy growth of crops.

[0060] In this embodiment, the agent set includes four agents: plant protection agent, irrigation agent, nutrition agent, and yield agent, which work together to jointly affect the growth and development of crops. The plant protection agent relies on irrigation and humidity control, the nutrition agent resists stress and predicts early warning, the irrigation agent responds to plant protection efficacy protection, nutrition dissolution demand, and prediction of critical water demand; the nutrition agent needs to adapt to irrigation migration, plant protection and fertilizer interaction, and prediction of fertilizer demand; the yield agent needs to combine plant protection loss, irrigation stress, and nutrition surplus and deficit data for prediction.

[0061] The embodiment of the present disclosure provides an agent collaboration method. First, multi-source agricultural data of crops in historical periods and current growth stages are collected. Second, a spatio-temporal feature map is constructed based on the multi-source agricultural data. Third, a crop stress index and a crop growth index are calculated based on the spatio-temporal feature map. Fourth, a target agent is determined from an agent set based on the crop stress index and the crop growth index. Finally, the crop growth index and the crop stress index are sent to the target agent, so that the target agent performs corresponding agricultural tasks on the crops based on the crop growth index and the crop stress index, and links the relevant agents in the agent set based on the environmental data. This collaborative processing method realizes the efficiency, accuracy and flexibility of information processing in the agricultural scene, effectively improves the decision-making efficiency and management level of agricultural production, and provides strong support for the intelligent development of agriculture.

[0062] In some optional implementations of the present disclosure, the multi-source agricultural data includes remote sensing data, Internet of Things data, field management data, meteorological data, and pest and disease monitoring data of historical periods and current growth stages. Based on the multi-source agricultural data, constructing the spatio-temporal feature graph includes: data preprocessing of the remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest and disease monitoring data; using a spatio-temporal attention mechanism to align the frequency and spatial resolution of the preprocessed remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest and disease monitoring data to obtain aligned data; and constructing a spatio-temporal feature graph based on the aligned data.

[0063] In this optional implementation, as shown in Figure 2 the remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest and disease monitoring data are multi-source heterogeneous data, and the data preprocessing includes data cleaning and cross-source alignment. The cross-source alignment includes unified spatio-temporal reference, matched resolution, semantic alignment, and quality evaluation. Through cross-source alignment, the remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest and disease monitoring data can be converted into high-quality fusion data sets that can be directly used for crop condition monitoring, yield prediction, or pest and disease early warning. For example, the Internet of Things data in the multi-source agricultural data can be used to align the remote sensing data, the field management data, the meteorological data, and the pest and disease monitoring data, and some key data can be extracted from the aligned remote sensing data, the field management data, the meteorological data, and the pest and disease monitoring data to construct a spatio-temporal feature graph.

[0064] In this optional implementation, a spatio-temporal feature graph is constructed by integrating agricultural data from multiple sources. Specifically, these data include remote sensing data (such as crop growth reflected in satellite images) of historical time periods, data collected by Internet of Things devices (such as soil humidity and temperature sensor data), field management records (such as the time and location of fertilization and irrigation operations), meteorological data (weather change information), and pest and disease monitoring data (occurrence of pests and diseases). To integrate these different types of data into a unified spatio-temporal feature graph, a spatio-temporal attention mechanism is used. This mechanism can adjust and align the frequency (time interval of data collection) and spatial resolution (spatial precision of data) of different data sources according to their importance in time and space. For example, remote sensing data may have a lower time frequency but a higher spatial resolution, while Internet of Things data may have a high time frequency but a limited spatial range. Through the spatio-temporal attention mechanism, the frequency and spatial resolution of these data can be adjusted to a consistent level, generating a comprehensive spatio-temporal feature graph for more comprehensive analysis and understanding of various factors and their relationships in the agricultural production process.

[0065] In some optional implementations of the present disclosure, the calculation of the crop stress index and the crop growth index based on the spatiotemporal feature atlas includes: obtaining a normalized vegetation index and a maximum normalized vegetation index based on a remote sensing unit in the spatiotemporal feature atlas; calculating a vegetation index relative value based on the normalized vegetation index and the maximum normalized vegetation index; calculating a water stress index based on available water and crop water requirement in the spatiotemporal feature atlas; calculating a nutrient stress index based on crop growth nitrogen utilization rate and crop theoretical nitrogen requirement in the spatiotemporal feature atlas; calculating a temperature stress index based on actual effective accumulated temperature in the growth period and crop optimum accumulated temperature in the spatiotemporal feature atlas; calculating the crop stress index based on the vegetation index relative value, the water stress index, the nutrient stress index, and the temperature stress index; calculating an average temperature based on temperature in the spatiotemporal feature atlas; calculating a soil photosynthetic capacity value based on soil water content in the spatiotemporal feature atlas; calculating a fertilizer photosynthetic capacity value based on total nitrogen content of soil and fertilization in the spatiotemporal feature atlas; calculating a precipitation photosynthetic capacity value based on precipitation in the spatiotemporal feature atlas; calculating a temperature photosynthetic capacity value based on the average temperature; and calculating the crop growth index based on the vegetation index relative value, the soil photosynthetic capacity value, the fertilizer photosynthetic capacity value, the precipitation photosynthetic capacity value, and the temperature photosynthetic capacity value.

[0066] In the optional implementation, the stress state of the crop is considered from the internal aspects such as water, nutrients, pests and diseases, and the index relative value NDVI is added to reflect the growth state of the crop from the appearance; the multi-modal data is mapped to a vector representation in a unified dimension to obtain the crop stress index as shown in formula (2).

[0067] CSI = f(Sndvi,Sw,Sn,St)(2)

[0068] In formula (2), Sndvi is the vegetation index relative value, Sw is the water stress index, Sn is the nutrient stress index, and St is the temperature stress index. Specifically, the calculation process of the crop stress index can be obtained in the manner of formula (3).

[0069] CSI = a×Sndvi×b×Sw×c×Sn×d×St(3)

[0070] In formula (3), a, b, c, and d are dynamic weights, each stress factor ranges from 0 to 1, 1 represents no stress, and 0 represents severe stress. The core of the calculation can adopt a geometric mean instead of an arithmetic mean, which embodies the nonlinear superposition effect of the stress factors (such as the interaction between water and nutrients). The weights a, b, c, and d are dynamically adjusted through AI learning, the stress factors are input, and the yield reduction rate of the best yield in history is output through calculation.

[0071] In this optional implementation, the remote sensing normalized vegetation index (NDVI) is a vegetation index obtained by remote sensing technology, which is used to reflect the growth condition and coverage of vegetation. Its value range is usually between -1 and 1, and the higher the value, the better the growth condition of the vegetation. The maximum normalized vegetation index is the highest value of the remote sensing normalized vegetation index in the historical period (such as the crop growing season), which is used as a reference standard.

[0072] In this optional implementation, the vegetation index relative value can be the ratio of the remote sensing normalized vegetation index to the maximum normalized vegetation index, which is used to measure the level of the current vegetation growth condition relative to the best state. Alternatively, the vegetation index relative value can also be the value calculated by formula (4).

[0073] Sndvi = (NDVIt-NDVImin) / (NDVImax-NDVImin) (4)

[0074] In formula (3), Sndvi is the vegetation index relative value, reflecting the strength of photosynthetic capacity, NDVIt is the current remote sensing NDVI value, NDVImax is the maximum NDVI of the same period in the past history period, and NDVImin is the bare soil NDVI value, which is generally 0.1.

[0075] In this optional implementation, the water stress index is an index calculated based on available water and crop water requirement, which is used to reflect the degree of stress on crops due to water deficiency or excess. As shown in formula (5), Sw represents the water stress index.

[0076] Sw = min(1,TAWact / TAWreq) (5)

[0077] In formula (5), TAWact is the actual available water in the root zone (field management data), which is obtained by soil sensors (or the publicly available remote sensing soil moisture content value). TAWreq is the crop water requirement (field management data), which is calculated by potential evapotranspiration ET0 x crop coefficient Kc.

[0078] In this optional implementation, the nutrient stress index is an index calculated based on the utilization rate of crop growth nitrogen and the theoretical nitrogen requirement of crops, which is used to reflect the degree of stress on crops due to nutrient deficiency or excess. As shown in formula (6), the nutrient stress index is Sn.

[0079] Sn = min(1,Nup / Ndm) (6)

[0080] In formula (6), Nup is the actual nitrogen content for crop growth (field management data), Nup = (Soil_Nmin + Fertilizer_N x Efficiency) x (1 - Leaching_loss), Efficiency is the utilization efficiency of nitrogen, and the utilization efficiency of nitrogen is different for different crops, such as the utilization efficiency of corn is 50%. Ndm is the theoretical nitrogen requirement of crops, Ndm = k x (1- ) x Nmax, c and k are crop parameters.

[0081] In this optional implementation, the temperature stress index is an index calculated based on the difference between the actual effective accumulated temperature in the growth period and the optimal accumulated temperature of crops, and is used to reflect the degree of stress on crops due to temperature inadaptation. As shown in formula (7), St represents temperature stress.

[0082] St = exp(-0.5 x |GDDt - GDDopt| / σ) (7)

[0083] In formula (7), GDDt is the actual effective accumulated temperature in the growth period (belongs to meteorological data), GDDopt is the optimal accumulated temperature of crops (accumulated temperature when high yield is planted in normal years), and σ is the tolerance range (usually 10% of the length of the growth period of crops).

[0084] In this optional implementation, the crop stress index is an index calculated based on the relative value of the vegetation index, the water stress index, the nutrient stress index and the temperature stress index, and is used to comprehensively reflect the comprehensive influence of various stress factors on crops in the growth process.

[0085] The crop growth index (CGI) reflects the growth and development stages of crops from seed to germination, emergence, flowering and maturity. The crop growth index CGI is specifically represented by a vector as shown in formula (8).

[0086] CGI = f(NDVI, SWC, Ntotal, Tavg, P) (8)

[0087] In formula (8), SWC is the soil water content, Ntotal is the total nitrogen content of soil and fertilization, Tavg is the average temperature, and P is the rainfall. In actual calculation, the crop growth index CGI can be calculated by the function formula shown in formula (9).

[0088] CGI = F(a) x (k1 x f(NDVI) + k2 x f(SWC) + k3 x f(Ntotal) + k4 x f(Tavg) + k5 x f(P)) (9)

[0089] In formula (9), a is a phenology adjustment coefficient [0-1], k1-k5 are dynamic weight factors, with default values (0.15, 0.4, 0.1, 0.25, 0.1) and recalculated using past data to calibrate; F(a) represents an effective accumulated temperature function; SWC is soil water content, obtained by using a soil moisture instrument (or obtained according to soil moisture data inversed by remote sensing); Ntotal: obtained from the amount of fertilization and soil nutrients; P is rainfall; soil photosynthetic capacity value f(SWC), an index calculated based on soil water content, used to reflect the support capacity of soil to photosynthesis; fertility photosynthetic capacity value f(Ntotal), an index calculated based on total nitrogen of soil and fertilization, used to reflect the support capacity of soil fertility to photosynthesis; precipitation photosynthetic capacity value f(P), an index calculated based on precipitation, used to reflect the support capacity of precipitation to photosynthesis; temperature photosynthetic capacity value f(Tavg), an index calculated based on average temperature, used to reflect the support capacity of temperature to photosynthesis; crop growth index: an index calculated based on the relative value of comprehensive vegetation index, soil photosynthetic capacity value, fertility photosynthetic capacity value, precipitation photosynthetic capacity value and temperature photosynthetic capacity value, used to comprehensively reflect the growth status and potential of crops, the f() representation is explained below by using the effective accumulated temperature function shown in formula (10) and the photosynthetic capacity function shown in formula (11)

[0090] F(a) = 1 / (1+exp(-b(GDD-GDD50))) (10)

[0091] In formula (10), GDD is the effective accumulated temperature with a base temperature of Tbase (such as 8℃), GDD = ∑max(0, (Tmax+Tmin) / 2 - Tbase), b is the growth rate (0.005-0.01), and GDD50 is the accumulated temperature at 50% growth potential.

[0092] F(NDVI) = 1 / (1+exp(-b(NDVI-NDVIc))) (11)

[0093] In formula (11), NDVIc is a crop phenology dynamic threshold (for example, 0.15 for the germination period and 0.85 for the flowering period), and b is a slope parameter (5-10).

[0094] In this optional implementation, firstly, the normalized difference vegetation index (NDVI) and the maximum normalized vegetation index are calculated respectively by using the remote sensing unit data in the spatiotemporal feature map. The relative value of the vegetation index is obtained by dividing the normalized difference vegetation index by the maximum normalized vegetation index, which is used to measure the relative level of the current vegetation growth condition and the best state. Then, the water stress index is calculated according to the available water and the crop water requirement in the spatiotemporal feature map; the nutrient stress index is calculated according to the utilization rate of crop growth nitrogen and the theoretical nitrogen requirement of crops; and the temperature stress index is calculated according to the difference between the actual effective accumulated temperature and the most suitable accumulated temperature during the growth period. Finally, the relative value of the vegetation index, the water stress index, the nutrient stress index and the temperature stress index are comprehensively calculated to obtain the crop stress index, which is used to comprehensively evaluate the comprehensive influence of various stress factors on crops.

[0095] At the same time, based on the temperature data in the spatiotemporal feature map, the average temperature is calculated; the soil photosynthetic capacity value is calculated according to the soil water content; the fertility photosynthetic capacity value is calculated according to the total nitrogen content of soil and fertilization; the precipitation photosynthetic capacity value is calculated according to the precipitation; and the temperature photosynthetic capacity value is calculated according to the average temperature. The relative value of the vegetation index, the soil photosynthetic capacity value, the fertility photosynthetic capacity value, the precipitation photosynthetic capacity value and the temperature photosynthetic capacity value are comprehensively calculated to obtain the growth crop index, which is used to comprehensively reflect the growth condition and potential of crops.

[0096] In some optional implementations of the present disclosure, the above determining the target agent from the agent set based on the crop stress index and the crop growth index includes: detecting whether the crop stress index is within the non-stress range; in response to detecting that the crop stress index is within the non-stress range, determining the growth stage of the crop based on the crop growth index; and determining the target agent from the agent set based on the growth stage of the crop.

[0097] In this optional implementation, firstly, it is detected whether the crop stress index is within the non-stress range. If the detection result indicates that the crop is not stressed, the system will further determine the growth stage of the current crop based on the crop growth index. According to the determined growth stage, the system will automatically select and trigger the corresponding target agent to realize precise agricultural management operation, for example, to start irrigation or fertilization equipment at a specific stage of crop growth, thereby improving the efficiency of agricultural production and the yield of crops.

[0098] Figure 3 The flow 300 of one embodiment of the agricultural task execution method according to the present disclosure is shown, which includes the following steps:

[0099] Step 301, receiving the crop growth index and the crop stress index of the crop.

[0100] In this embodiment, the execution subject on which the agricultural task execution method runs can be any one of the agent set, and the execution subject can receive the crop growth index and the crop stress index from the central agent.

[0101] In this embodiment, the central agent can monitor the multi-source agricultural data of crops in real time through the sensor network installed in the farmland, calculate the crop growth index and stress index by analyzing and processing the multi-source agricultural data, and the specific calculation process can refer to the crop growth index and stress index calculation method shown in the embodiment. Figure 1 The agent cooperation method described in the embodiment.

[0102] Step 302, based on the crop growth index and the crop stress index, the agricultural task is executed on the crop.

[0103] In this embodiment, according to the numerical range of these indexes, the threshold value is set, when the growth index is lower than the normal range, it may need to fertilize or irrigate; when the stress index is higher than the warning value, it may need to take disease and pest control or drought resistance measures. The execution subject automatically triggers the corresponding agricultural task according to the real-time data of these indexes, such as starting the irrigation equipment, releasing pesticides or notifying the farmer to take manual intervention measures, so as to realize the precision agriculture management and improve the crop yield and quality.

[0104] Step 303, when executing the agricultural task, the environmental data is obtained;

[0105] In this embodiment, when executing the agricultural task (for example, irrigation), the environmental data is obtained through the sensor network installed in the farmland. These sensors can monitor the soil moisture, air temperature and humidity and other key indicators in real time. For example, the soil moisture sensor can accurately measure the water content in the soil, when the soil moisture is lower than the preset threshold value, the execution subject will automatically trigger the irrigation operation to ensure that the crops obtain appropriate amount of water. At the same time, combined with the weather data (such as the rainfall forecast in the next few days) provided by the weather station, the execution subject can further optimize the irrigation plan to avoid excessive irrigation. In this way, not only the utilization efficiency of water resources is improved, but also the frequency of manual intervention is reduced, realizing the intelligent and precise management of agricultural production.

[0106] Step 304, based on the environmental data, the related agents in the agent set are linked.

[0107] In this embodiment, the agent set is the agent set controlled by the execution subject on which the agent cooperation method runs, and the agent set includes a plurality of agents, each agent has a corresponding function, when the execution subject on which the agricultural task execution method runs executes the corresponding agricultural task, if it is found that the environmental data is related to the related agent in the agent set, the related agent is linked. The related agent is one or more agents in the agent set, and the related agent is related to the environmental data.

[0108] In this embodiment, the environmental data of the farmland (such as soil humidity, temperature, light intensity, carbon dioxide concentration, etc.) is collected by sensors and other devices, and then transmitted to the execution subject through Internet of Things technology. The execution subject analyzes the data according to the preset algorithm and model to determine the growth state and needs of the crops. Subsequently, the execution subject will automatically send control signals to the relevant agents (such as irrigation agents, fertilization agents, greenhouse agents, etc.) in the agent set to perform precise operations on the relevant agents. For example, when the soil humidity is below the set threshold, the irrigation agent is automatically started; when the light is insufficient, the light supplement agent in the greenhouse is adjusted. This intelligent linkage mode can improve the efficiency of agricultural production, reduce labor costs, and at the same time optimize the growth environment of crops, improve yield and quality.

[0109] In this embodiment, the agent set can include plant protection agents, irrigation agents, nutrition agents, and yield agents. When the agricultural task execution method runs on the execution subject of the plant protection agent, the irrigation agent can be linked. Irrigation will affect the field humidity and increase the risk of disease occurrence. Irrigation is suspended within 24 hours before pesticide application to avoid dilution of the pesticide and ensure leaf drying. After pesticide application, trace irrigation can be started to promote the penetration of the pesticide to the root system and enhance the effect of systemic pesticides. The nutrition agent is linked, which provides nutrient balance. Good nutrition can thicken the cell wall of crops, thereby reducing the probability of insect infestation, and the plant protection agent adjusts the pesticide proportion accordingly. The yield agent is linked, which predicts the yield level based on the growth, agricultural data, and weather data. The plant protection agent patrols and prevents in advance according to the predicted yield level to avoid risks.

[0110] When the agricultural task execution method runs on the execution subject of the irrigation agent, the plant protection agent can be linked, such as spraying herbicides, which requires a delay of irrigation for more than 12 hours, otherwise the efficacy will be lost. The nutrition agent is linked, and the water requirement for nitrogen fertilizer dissolution is directly related to the irrigation depth. For example, urea requires soil moisture content > 18% at 30 cm soil layer for dissolution, otherwise ammonia volatilization loss will occur. The irrigation agent needs to adjust the drip irrigation time according to the fertilization plan of the nutrition agent. The yield agent is linked, which marks the water-sensitive period (such as the corn tasseling and silk shedding period) based on the crop growth model. The irrigation module accordingly increases the water allocation priority by 50% to ensure grain formation.

[0111] When the execution subject of the agricultural task execution method is a nutrition agent, the nutrition agent needs to suspend the application of magnesium fertilizer for 3 days after the fungicide is applied by the plant protection agent to avoid the light decomposition of the fungicide. When the plant protection agent detects an aphid outbreak, the nutrition agent can increase the proportion of silicon fertilizer to enhance the mechanical resistance of the plant. The nutrition agent is linked to the irrigation agent. When the drip irrigation belt is buried 5 cm deep, the nitrogen moves 40 cm in the wetting front, while when it is buried 10 cm deep, it only moves 25 cm. The nutrition module needs to optimize the fertilization position accordingly to avoid deep seepage. The nutrition agent is linked to the yield agent. According to the yield prediction value, the nutrition agent is instructed to increase the fertilizer to compensate for the yield.

[0112] When the execution subject of the agricultural task execution method is a yield agent, the yield agent is linked to the plant protection agent. The live data provided by the plant protection agent, such as a 15% decrease in the rust leaf area index, is directly input into the yield model to correct the prediction error. If the plant protection stress decreases or there is no stress 7 days after the plant protection agent applies the pesticide, the yield model will return to normal prediction level. The yield agent is linked to the irrigation agent. Soil water potential data, such as <-60 kPa for 5 days, triggers the yield model to start the drought yield reduction coefficient, affecting the yield. When the actual irrigation amount is lower than the planned value, the model automatically adjusts the expected yield. The yield agent is linked to the nutrition agent. When the nitrogen content of the crop is low, the nutrition module marks it as a "potential yield reduction area", and the yield model increases the estimation error range accordingly. The yield model responds to the top dressing after 7 days.

[0113] The agricultural task execution method provided by the embodiments of the present disclosure first receives a crop growth index and a crop stress index of a crop; second, executes an agricultural task on the crop based on the crop growth index and the crop stress index; then, acquires environmental data while executing the agricultural task; and finally, links related agents in the agent set based on the environmental data, executes the agricultural task on the crop based on the crop growth index and the crop stress index, and links the related agents based on the environmental data, thereby improving the reliability of the collaborative operation between multiple agents.

[0114] In some optional implementations of the present disclosure, the above-mentioned agricultural task is a fertilization task, and executing the agricultural task on the crop based on the crop growth index and the crop stress index includes: determining a fertilization index based on the crop growth index and the crop stress index; determining a top dressing amount based on the fertilization index; calculating a standard value of nitrogen content based on historical crop leaf nitrogen content; and determining fertilization recommendation information in the fertilization task based on the standard value of nitrogen content and the top dressing amount.

[0115] In the optional implementation, the fertilization index is an index for guiding fertilization, which is calculated according to the crop growth index and the crop stress index and reflects the urgency of the current fertilizer demand of the crop; the additional fertilizer amount is the amount of fertilizer that needs to be additionally applied and is used to supplement the nutrients required during the growth of the crop; the nitrogen content standard value is a reference value of the nitrogen content of the crop leaf calculated based on historical data and is used to evaluate whether the current nutritional status of the crop reaches the ideal level; and the fertilization recommendation information is specific recommendations about fertilization generated after comprehensively considering the fertilization index, the additional fertilizer amount and the nitrogen content standard value, including the fertilization time, the fertilization amount, the fertilization method and the like.

[0116] In the optional implementation, the fertilization index is calculated according to the crop growth index and the crop stress index, which can reflect the current fertilizer demand of the crop. Then, the standard value of the nitrogen content of the crop leaf is calculated based on historical data, which is used as a reference for evaluating the nutritional status of the crop. Finally, the additional fertilizer amount is determined according to the fertilization index, and specific fertilization recommendation information is generated in combination with the nitrogen content standard value, which provides precise fertilization guidance for agricultural producers, thereby optimizing the fertilization effect and improving the yield and quality of crops.

[0117] Further reference Figure 4 , as an implementation of the method shown in the above figures, the present disclosure provides an embodiment of an agent cooperation device, which corresponds to the method embodiment shown in Figure 1 , and the device can be applied to various electronic devices.

[0118] As shown in Figure 4 , the agent cooperation device 400 provided by the embodiment includes a collection unit 401, a graph construction unit 402, an index construction unit 403, a determination unit 404 and a sending unit 405. The collection unit 401 can be configured to collect multi-source agricultural data of a crop in a historical period and a current growth stage. The graph construction unit 402 can be configured to construct a spatio-temporal feature graph based on the multi-source agricultural data. The index construction unit 403 can be configured to calculate a crop stress index and a crop growth index based on the spatio-temporal feature graph. The determination unit 404 can be configured to determine a target agent from a set of agents based on the crop stress index and the crop growth index. The sending unit 405 can be configured to send the crop growth index and the crop stress index to the target agent, so that the target agent performs a corresponding agricultural task on the crop based on the crop growth index and the crop stress index, and links relevant agents in the set of agents based on environmental data.

[0119] In the embodiment, the specific processing of the acquisition unit 401, the atlas construction unit 402, the index construction unit 403, the determination unit 404, and the sending unit 405 in the intelligent agent cooperation device 400 and the technical effects brought by the specific processing can be referred to the specific processing of the acquisition unit 401, the atlas construction unit 402, the index construction unit 403, the determination unit 404, and the sending unit 405 respectively. Figure 1 The related description of steps 101, 102, 103, 104, and 105 in the corresponding embodiment will not be repeated here.

[0120] In some embodiments of the present disclosure, the multi-source agricultural data includes remote sensing data, Internet of Things data, field management data, meteorological data, and pest monitoring data of historical periods and current growth stages, and the atlas construction unit 402 is configured to: perform data preprocessing on the remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest monitoring data; adopt a spatio-temporal attention mechanism to align the frequency and spatial resolution of the preprocessed remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the pest monitoring data to obtain aligned data; and construct a spatio-temporal feature atlas based on the aligned data.

[0121] In some embodiments of the present disclosure, the index construction unit 403 is further configured to: based on the remote sensing unit in the spatio-temporal feature atlas, obtain a normalized difference vegetation index and a maximum normalized difference vegetation index; based on the normalized difference vegetation index and the maximum normalized difference vegetation index, calculate a vegetation index relative value; based on the available water and the crop water requirement in the spatio-temporal feature atlas, calculate a water stress index; based on the utilization rate of crop growth nitrogen and the theoretical nitrogen requirement of crops in the spatio-temporal feature atlas, calculate a nutrient stress index; based on the actual effective accumulated temperature in the growth period and the most suitable accumulated temperature of crops in the spatio-temporal feature atlas, calculate a temperature stress index; based on the vegetation index relative value, the water stress index, the nutrient stress index, and the temperature stress index, calculate a crop stress index; based on the temperature in the spatio-temporal feature atlas, calculate an average temperature; based on the soil water content in the spatio-temporal feature atlas, calculate a soil photosynthetic capacity value; based on the total nitrogen content of soil and fertilization in the spatio-temporal feature atlas, calculate a fertilizer photosynthetic capacity value; based on the precipitation in the spatio-temporal feature atlas, calculate a precipitation photosynthetic capacity value; based on the average temperature, calculate a temperature photosynthetic capacity value; and based on the vegetation index relative value, the soil photosynthetic capacity value, the fertilizer photosynthetic capacity value, the precipitation photosynthetic capacity value, and the temperature photosynthetic capacity value, calculate a crop growth index.

[0122] In some embodiments of the present disclosure, the determination unit 404 is further configured to: detect whether the crop stress index is within a non-stress range; in response to detecting that the crop stress index is within the non-stress range, determine a crop growth stage based on the crop growth index; and determine a target intelligent agent from the set of intelligent agents based on the crop growth stage.

[0123] The intelligent agent collaborative device provided in the embodiments of this disclosure firstly involves a data acquisition unit 401 acquiring multi-source agricultural data on the historical period and current growth stage of crops; a map construction unit 402 constructing a spatiotemporal feature map based on the multi-source agricultural data; an index construction unit 403 calculating a crop stress index and a crop growth index based on the spatiotemporal feature map; a determination unit 404 determining a target intelligent agent from the intelligent agent set based on the crop stress index and the crop growth index; and a sending unit 405 sending the crop growth index and the crop stress index to the target intelligent agent, so that the target intelligent agent can perform corresponding agricultural tasks on the crops based on the crop growth index and the crop stress index, and coordinate with relevant intelligent agents in the intelligent agent set based on environmental data.

[0124] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides an embodiment of an agricultural task execution device, which is similar to... Figure 3 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0125] like Figure 5 As shown, the agricultural task execution device 500 provided in this embodiment includes: a receiving unit 501, an execution unit 502, an acquisition unit 503, and a linkage unit 504. The receiving unit 501 can be configured to receive crop growth index and crop stress index. The execution unit 502 can be configured to perform agricultural tasks on crops based on the crop growth index and crop stress index. The acquisition unit 503 can be configured to acquire environmental data while performing agricultural tasks. The linkage unit 504 can be configured to link relevant agents in the agent set based on the environmental data.

[0126] In this embodiment, the specific processing of the receiving unit 501, the execution unit 502, the acquisition unit 503, and the linkage unit 504 in the agricultural task execution device 500, and the resulting technical effects, can be found in references to [reference needed]. Figure 3 The relevant descriptions of steps 301, 302, 303, and 304 in the corresponding embodiments will not be repeated here.

[0127] In some embodiments of this disclosure, the execution unit 502 is further configured to: determine a fertilization index based on the crop growth index and the crop stress index; determine the amount of topdressing based on the fertilization index; calculate a standard value of nitrogen content based on the historical nitrogen content of crop leaves; and determine fertilization recommendation information in the fertilization task based on the standard value of nitrogen content and the amount of topdressing.

[0128] The embodiment of the present disclosure provides an agricultural task execution device. First, the receiving unit 501 receives a crop growth index and a crop stress index of a crop; the execution unit 502 executes an agricultural task on the crop based on the crop growth index and the crop stress index; the acquisition unit 503 acquires environmental data when the agricultural task is executed; and the linkage unit 504 links related agents in the agent set based on the environmental data. Thus, this cooperative working mode can ensure that the agricultural production process is more efficient, accurate and adaptable, improve crop yield and quality, reduce resource waste and environmental impact, and realize intelligent and sustainable development of agricultural production.

[0129] Further referring to Figure 6 The present disclosure also provides an agent coordination system, which comprises a central agent 601 and an agent set 602.

[0130] The central agent 601 is configured to collect multi-source agricultural data of a crop a in a historical period and a current growth stage, construct a spatio-temporal feature map based on the multi-source agricultural data, calculate a crop stress index and a crop growth index based on the spatio-temporal feature map, determine a target agent 6021 from the agent set 602 based on the crop stress index and the crop growth index, and send the crop growth index and the crop stress index to the target agent 6021.

[0131] The target agent 6021 executes a corresponding agricultural task on the crop a based on the crop growth index and the crop stress index, and links related agents 6022 in the agent set 602 based on environmental data.

[0132] The agent coordination system provided by the present disclosure improves the globality and cooperativeness of agricultural decision-making through the cooperation of the central agent and the agents in the agent set. Through the task scheduling mechanism of the multi-agent system, the linkage optimization of plant protection, irrigation, nutrition, yield prediction and other links is realized, and the local optimal trap caused by a single model is avoided.

[0133] According to the embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0134] Figure 7A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their patterns are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0135] like Figure 7 As shown, the electronic device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 702 or a computer program loaded into a random access memory (RAM) 703 from a storage unit 708. The RAM 703 may also store various programs and data required for the operation of the electronic device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0136] Multiple components in electronic device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of displays, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows electronic device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0137] The computing unit 701 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs various methods and processes described above, such as the agent coordination method. For example, in some embodiments, the agent coordination method can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded onto the RAM 703 and executed by the computing unit 701, one or more steps of the agent coordination method described above can be performed. Alternatively, in other embodiments, the computing unit 701 can be configured to perform the agent coordination method by any other suitable means, such as by means of firmware.

[0138] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0139] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable intelligent agent coordination device to produce a machine, such that the program code, when executed by the processor or controller, causes the machine to perform the methods / operations specified in the flowcharts and / or block diagrams. The program code can execute entirely on a machine, partly on a machine, as a stand-alone software package, partly on a machine and partly on a remote machine or entirely on a remote machine or server.

[0140] In the context of this disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0141] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0142] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0143] It should be understood that various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the spirit of the present disclosure. For example, the steps recited in the present disclosure can be performed in parallel, in series, or in a different order, without departing from the desired results of the technology disclosed in the present disclosure, which are not limited herein.

[0144] The foregoing description of specific exemplary embodiments of the disclosure has been presented for the purposes of illustration and explanation. It is not intended to be exhaustive or to limit the disclosure to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching. It is intended that the embodiments be chosen and / or described such that the disclosure is practical and sufficient for one skilled in the art to practice and utilize the disclosure, as well as to customize the disclosure for various uses and conditions. The scope of the disclosure is intended to be limited only by the claims and their equivalents.

Claims

1. An agent coordination method, characterized by, The method comprises: Collecting multi-source agricultural data of a crop in a historical period and a current growth stage; Based on the multi-source agricultural data, constructing a spatio-temporal feature atlas; Based on the spatio-temporal feature atlas, calculating a crop stress index and a crop growth index; Based on the crop stress index and the crop growth index, determining a target agent from a set of agents, including: based on the crop stress index and the crop growth index, triggering a plant protection agent in the set of agents, and performing an agricultural task of the corresponding target agent at different stages, while running a disease and pest monitoring and early warning program according to the crop stress index and monitoring equipment, meteorological factors; Sending the crop growth index and the crop stress index to the target agent, so that the target agent performs corresponding agricultural tasks on the crop based on the crop growth index and the crop stress index, and links related agents in the set of agents based on environmental data; The agricultural tasks of the target agent include the agricultural tasks of an irrigation agent, the agricultural tasks of a nutrition agent, and the agricultural tasks of a yield estimation agent, the agricultural tasks of the irrigation agent include: real-time analysis of soil water content, meteorological data, and crop water requirement curve; analyzing the relationship between crop water requirement and soil water content in combination with the crop stress index and the growth index, dynamically adjusting the irrigation plan; outputting irrigation recommendations to the control system to realize automatic irrigation scheduling; the agricultural tasks of the nutrition agent include: analyzing soil nutrient content, crop growth stage, and target yield; analyzing the relationship between crop nutrients and soil nutrient supply in combination with the crop stress index and the growth index, dynamically recommending fertilizer type, dosage, and application time; fusing remote sensing NDVI data to judge leaf nutrient status and assist in diagnosing nutrient deficiency symptoms; triggering the nutrition agent according to the crop stress index and the growth index, and the nitrogen content of crop leaves is dynamically changed at different growth stages of the crop; calculating the normal nitrogen content standard of each crop growth index stage based on the historical crop leaf nitrogen content level, then comparing the current crop leaf nitrogen content level to give fertilizer recommendations, and increasing or decreasing the normal fertilizer amount; the agricultural tasks of the yield estimation agent include: dynamically predicting yield in combination with current year climate conditions, management measures, stress index, etc.; fusing remote sensing NDVI data to support multi-scale prediction, starting the dynamic yield estimation agent during the grain formation period according to the change value of the crop stress index and the growth index, and evaluating yield changes using NDVI, meteorological, CSI, and management data.

2. The method of claim 1, wherein, The multi-source agricultural data includes remote sensing data, Internet of Things data, field management data, meteorological data, and disease and pest monitoring data in the historical period and the current growth stage, and the construction of the spatio-temporal feature atlas based on the multi-source agricultural data comprises: Data preprocessing is performed on the remote sensing data, the Internet of Things data, the field management data, the meteorological data, and the disease and pest monitoring data; aligning the frequency and spatial resolution of the preprocessed remote sensing data, internet of things data, field management data, meteorological data, and pest and disease monitoring data using a spatio-temporal attention mechanism to obtain aligned data; constructing a spatio-temporal feature graph based on the aligned data.

3. The method of claim 1, wherein, calculating a crop stress index and a crop growth index based on the spatio-temporal feature graph includes: obtaining a normalized difference vegetation index and a maximum normalized difference vegetation index based on a remote sensing unit in the spatio-temporal feature graph; calculating a vegetation index relative value based on the normalized difference vegetation index and the maximum normalized difference vegetation index; calculating a water stress index based on available water and crop water requirement in the spatio-temporal feature graph; calculating a nutrient stress index based on crop growth nitrogen utilization rate and crop theoretical nitrogen requirement in the spatio-temporal feature graph; calculating a temperature stress index based on actual effective accumulated temperature and crop optimum accumulated temperature in the spatio-temporal feature graph during the growth period; calculating a crop stress index based on the vegetation index relative value, the water stress index, the nutrient stress index, and the temperature stress index; calculating an average temperature based on temperature in the spatio-temporal feature graph; calculating a soil photosynthetic capacity value based on soil water content in the spatio-temporal feature graph; calculating a fertilizer photosynthetic capacity value based on total nitrogen amount of soil and fertilization in the spatio-temporal feature graph; calculating a precipitation photosynthetic capacity value based on precipitation in the spatio-temporal feature graph; calculating a temperature photosynthetic capacity value based on the average temperature; calculating a crop growth index based on the vegetation index relative value, the soil photosynthetic capacity value, the fertilizer photosynthetic capacity value, the precipitation photosynthetic capacity value, and the temperature photosynthetic capacity value.

4. The method of claim 1, wherein, determining a target agent from a set of agents based on the crop stress index and the crop growth index includes: detecting whether the crop stress index is within a non-stress range; in response to detecting that the crop stress index is within the non-stress range, determining a crop growth stage based on the crop growth index; determining a target agent from the set of agents based on the crop growth stage.

5. An agricultural task execution method applied to a target agent, wherein the target agent is determined by the agent cooperation method of any one of claims 1-4, characterized in that, The method includes: receiving a crop growth index and a crop stress index of a crop; performing an agricultural task on the crop based on the crop growth index and the crop stress index; acquiring environmental data while performing the agricultural task; linking relevant agents in a set of agents based on the environmental data.

6. The method of claim 5, wherein the agricultural task is a fertilization task, and performing an agricultural task on the crop based on the crop growth index and the crop stress index includes: determining a fertilization index based on the crop growth index and the crop stress index; determining a topdressing amount based on the fertilization index; calculating a standard value of nitrogen content based on historical crop leaf nitrogen content; determining fertilization recommendation information in the fertilization task based on the standard value of nitrogen content and the topdressing amount.

7. An agent coordination apparatus, the apparatus comprising: an acquisition unit configured to acquire multi-source agricultural data of a crop in a historical period and a current growth stage; a graph construction unit configured to construct a spatio-temporal feature graph based on the multi-source agricultural data; an index construction unit configured to calculate a crop stress index and a crop growth index based on the spatio-temporal feature graph; a determination unit configured to determine a target agent from a set of agents based on the crop stress index and the crop growth index; the determination unit is configured to trigger a plant protection agent in the set of agents to perform an agricultural task of the corresponding target agent at different stages, and simultaneously run a disease and pest monitoring and early warning program according to the crop stress index and monitoring equipment and meteorological factors; a sending unit configured to send the crop growth index and the crop stress index to the target agent, so that the target agent performs corresponding agricultural tasks on the crop based on the crop growth index and the crop stress index, and links related agents in the set of agents based on environmental data; the agricultural tasks of the target agent include agricultural tasks of an irrigation agent, agricultural tasks of a nutrition agent, and agricultural tasks of a yield estimation agent; the agricultural tasks of the irrigation agent include real-time analysis of soil water content, meteorological data, and crop water requirement curve; analysis of the relationship between crop water requirement and soil water content in combination with the crop stress index and the growth index to dynamically adjust irrigation plan; and output of irrigation recommendations to a control system to realize automatic irrigation scheduling; the agricultural tasks of the nutrition agent include analysis of soil nutrient content, crop growth stage, and target yield; analysis of the relationship between crop nutrients and soil nutrient supply in combination with the crop stress index and the growth index to dynamically recommend fertilizer type, dosage, and application time; fusion of remote sensing NDVI data to determine leaf nutrition status and assist in diagnosing nutrient deficiency; triggering of the nutrition agent according to the crop stress index and the growth index, with dynamic changes in nitrogen content of crop leaves at different growth stages of the crop; calculation of normal nitrogen content standard for each crop growth index stage based on historical crop leaf nitrogen content level, comparison of current crop leaf nitrogen content level, and giving of fertilizer recommendations with increase or decrease on the basis of normal fertilizer amount; and the agricultural tasks of the yield estimation agent include dynamic yield prediction in combination with factors such as current year climate conditions, management measures, and stress index; multi-scale prediction supported by fusion of remote sensing NDVI data; starting of a dynamic yield estimation agent during grain formation period according to change values of the crop stress index and the growth index; and evaluation of yield change by using NDVI, meteorological, CSI, and management data.

8. An agricultural task execution device applied to a target agent determined by the agent coordination device of claim 7, the device comprising: a receiving unit configured to receive a crop growth index and a crop stress index of a crop; an execution unit configured to perform an agricultural task on the crop based on the crop growth index and the crop stress index; an acquisition unit configured to acquire environmental data when the agricultural task is performed. A linkage unit configured to link relevant agents in the agent set based on the environmental data.

9. An agent coordination system, the system comprising: a central agent, an agent set; a central agent configured to collect multi-source agricultural data of a crop in a historical period and a current growth stage; constructing a spatiotemporal feature atlas based on the multi-source agricultural data; calculating a crop stress index and a crop growth index based on the spatiotemporal feature atlas, triggering a plant protection agent in the agent set to perform an agricultural task of a target agent at different stages, and simultaneously running a disease and pest monitoring and early warning program according to the crop stress index and monitoring equipment and meteorological factors, determining the target agent from the agent set based on the crop stress index and the crop growth index, and sending the crop growth index and the crop stress index to the target agent; the target agent performing corresponding agricultural tasks on the crop based on the crop growth index and the crop stress index, and linking relevant agents in the agent set based on environmental data; the agricultural tasks of the target agent include an agricultural task of an irrigation agent, an agricultural task of a nutrition agent, and an agricultural task of a yield estimation agent, the agricultural task of the irrigation agent includes real-time analysis of soil water content, meteorological data, and crop water requirement curve, analysis of the relationship between crop water requirement and soil water content based on the crop stress index and the growth index, dynamic adjustment of irrigation plan, output of irrigation recommendations to a control system, and realization of automatic irrigation scheduling, the agricultural task of the nutrition agent includes analysis of soil nutrient content, crop growth stage, and target yield, analysis of the relationship between crop nutrients and soil nutrient supply based on the crop stress index and the growth index, dynamic recommendation of fertilizer type, dosage, and application time, fusion of remote sensing NDVI data to determine leaf nutrient status and assist in diagnosing nutrient deficiency symptoms, triggering of the nutrition agent based on the crop stress index and the growth index, dynamic change of nitrogen content in crop leaves at different growth stages of the crop, calculation of normal nitrogen content standards for each crop growth index stage based on historical crop leaf nitrogen content levels, comparison of current crop leaf nitrogen content levels, and giving of fertilizer recommendations based on normal fertilizer application amount, and the agricultural task of the yield estimation agent includes dynamic prediction of yield based on factors such as current year climate conditions, management measures, and stress index, multi-scale prediction supported by fusion of remote sensing NDVI data, starting of a dynamic yield estimation agent during grain formation period according to change values of the crop stress index and the growth index, and evaluation of yield changes by using NDVI, meteorological data, CSI, and management data.

10. An electronic device, comprising: comprise: at least one processor; and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.

11. A non-transitory computer-readable storage medium having stored thereon computer instructions, wherein, The computer instructions are for causing the computer to perform the method of any one of claims 1-6.

Images