Agricultural intelligent analysis and decision-making method based on multi-modal data

Abstract

The invention relates to the technical field of agricultural intelligent decision making, and discloses an agricultural intelligent analysis and decision making method based on multi-modal data, which comprises a multi-modal agent, an agent collaborative decision making technology, an intelligent decision making method and a crop identification and diagnosis method. The multi-modal intelligent agent comprises an intelligent agent framework and multi-modal capability construction; the intelligent agent collaborative decision-making technology comprises intelligent agent ecological construction, a land parcel analysis method and a land parcel economy evaluation model; the intelligent decision-making method comprises the steps of agricultural knowledge graph construction, understanding and decision making. The crop identification and diagnosis method comprises vision-language model preprocessing and intelligent identification and diagnosis. According to the agricultural intelligent analysis and decision-making method based on the multi-modal data, the technical problem in traditional agricultural decision-making is solved through processing, analysis and technical innovation of the decision-making process of agricultural multi-source data, and accurate analysis and scientific decision-making support is provided for all stages of agricultural production before production, during production and after production.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural intelligent decision-making, and specifically to an agricultural intelligent analysis and decision-making method based on multimodal data. Background Art

[0002] With population growth and limited resources, agricultural development faces numerous challenges, such as increasing yields, reducing costs, and protecting the environment. Artificial intelligence (AI) technology, with its powerful data processing and pattern recognition capabilities, provides strong support for smart agricultural decision-making. Smart agriculture leverages modern information technology to achieve intelligent and refined agricultural production, improving agricultural production efficiency and resource utilization. Research on the application of AI technology in smart agricultural decision-making can address practical production issues, improve agricultural production efficiency, promote technological advancement and innovation, promote sustainable agricultural development, and safeguard food security and the ecological environment. Intelligent agricultural analysis and decision-making methods are systematic solutions developed by integrating technologies such as the Internet of Things, big data, and AI.

[0003] The existing technology has the following deficiencies:

[0004] 1. The land parcel analysis method is single and fails to comprehensively consider multi-dimensional factors.

[0005] 2. Lack of intelligent agricultural knowledge question-answering and decision-making reasoning capabilities.

[0006] 3. Crop identification and pest and disease diagnosis have limited accuracy.

[0007] 4. Decision recommendations lack scenario adaptability and explainability. Summary of the Invention

[0008] (1) Technical problems solved

[0009] In response to the shortcomings of existing technologies, the present invention provides an agricultural intelligent analysis and decision-making method based on multimodal data, which has the advantage of effectively improving the intelligence and precision of agricultural production decisions, and solves the problem that agriculture fails to comprehensively consider multi-dimensional factors.

[0010] (2) Technical solution

[0011] To achieve the above objectives, the present invention provides the following technical solutions: an agricultural intelligent analysis and decision-making method based on multimodal data, comprising a multimodal intelligent agent, intelligent agent collaborative decision-making technology, an intelligent decision-making method, and a crop identification and diagnosis method;

[0012] Multimodal agents include agent architecture and multimodal capability building:

[0013] Intelligent agent collaborative decision-making technology includes intelligent agent ecosystem construction, land parcel analysis methods, and land parcel economic evaluation models;

[0014] Intelligent decision-making methods include agricultural knowledge graph construction, understanding and decision-making;

[0015] Crop recognition and diagnosis methods include vision-language model preprocessing and intelligent recognition and diagnosis.

[0016] Optimally, the agent architecture is:

[0017] Build a data processing technology framework;

[0018] Realize domain knowledge transformation mechanism;

[0019] Establishing a decision analysis process;

[0020] Multimodal capacity building:

[0021] Multimodal data processing: Fusion of heterogeneous data such as images, text, and time series data enables comprehensive analysis of agricultural scenarios;

[0022] Cross-modal feature processing: Through feature extraction and association analysis technology, cross-modal association and mapping of images, texts, and data can be achieved;

[0023] Knowledge-enhanced analysis: Applying knowledge graph technology to enhance professional analytical capabilities in the agricultural field;

[0024] Agricultural semantic processing: Optimize the parsing process of agricultural professional terms and expressions, and improve the accuracy of processing agricultural contexts.

[0025] Optimally, the intelligent body ecosystem is constructed:

[0026] Land parcel analysis engine agent: responsible for adaptability analysis of soil, terrain, etc.;

[0027] Price analysis engine agent: responsible for market supply and demand, and price trend analysis;

[0028] Meteorological data engine agent: responsible for analyzing climate conditions and crop growth adaptability;

[0029] Intelligent agent collaborative decision-making system: realizes multi-agent information sharing and decision-making collaboration;

[0030] Land parcel analysis method:

[0031] Time series analysis of meteorological data: collecting climate change trends in recent years;

[0032] Soil data feature extraction: analysis of soil physical and chemical properties;

[0033] Market data collection: historical prices, supply and demand trend analysis;

[0034] Land economic evaluation model:

[0035] Planting adaptability analysis: based on natural conditions such as climate, soil, and topography;

[0036] Economic benefit forecast: based on market supply and demand, and price trend analysis;

[0037] Investment risk assessment: Consider natural risks and market risks;

[0038] Comprehensive decision-making recommendations: Generate planting investment return analysis report.

[0039] Optimally, agricultural knowledge graph construction:

[0040] Structuring of expert knowledge;

[0041] Knowledge entity relationship extraction;

[0042] Knowledge graph is dynamically updated;

[0043] Understanding and Decision-Making:

[0044] Speech recognition preprocessing;

[0045] Intent understanding and slot filling;

[0046] Personalized recommendation generation.

[0047] Preferably, vision-language model preprocessing:

[0048] Image standardization processing;

[0049] Scene text extraction;

[0050] Multimodal feature alignment;

[0051] Contextual information fusion;

[0052] Intelligent identification and diagnosis include crop identification, growth status assessment, intelligent diagnosis of pests and diseases, and generation of prevention and control plans based on VL models;

[0053] Crop recognition based on VL model: utilizing visual-linguistic bimodal features;

[0054] Growth status assessment: multi-dimensional analysis combined with expert knowledge base;

[0055] Intelligent diagnosis of pests and diseases: semantic matching based on symptom descriptions;

[0056] Generation of prevention and control plans: personalized suggestions based on regional characteristics.

[0057] Preferably, it includes a multimodal feature processing method, an intelligent agent collaborative decision-making framework implementation method, a plot analysis implementation method, an intelligent interaction implementation method, and a crop identification and diagnosis implementation method;

[0058] Multimodal feature processing methods include data preprocessing technology, feature extraction methods, cross-modal feature fusion technology and knowledge enhancement analysis methods;

[0059] The implementation method of the intelligent agent collaborative decision-making framework includes intelligent agent module initialization, task allocation mechanism, parallel computing processing, result integration and collaboration, and feedback optimization loop;

[0060] The implementation methods of plot analysis include data collection and preprocessing, multi-factor adaptability analysis, economic benefit prediction, risk assessment and decision-making recommendation generation;

[0061] Intelligent interaction implementation methods include multimodal input recognition, intent understanding and slot filling, knowledge graph query, context-aware reasoning, and personalized suggestion generation;

[0062] Crop identification and diagnosis implementation methods include image acquisition and enhancement, multimodal information fusion, feature extraction and matching, pest and disease identification and diagnosis, and prevention and control plan generation.

[0063] Preferably, data preprocessing technology: establish standardized processing procedures for agricultural data of different modalities to eliminate noise and outliers.

[0064] Feature extraction method:

[0065] Image feature extraction: Apply algorithms to extract key features of images of crops, pests, and diseases;

[0066] Text feature extraction: Use natural language processing technology to analyze text information such as user descriptions and expert suggestions;

[0067] Time series data feature extraction: Apply time series analysis methods to process weather, output, price and other data to identify changing trends;

[0068] Cross-modal feature fusion technology: Design feature alignment algorithms and fusion mechanisms to address heterogeneity issues with data from different modalities.

[0069] Technical processing flow: data preprocessing → feature extraction → cross-modal fusion → knowledge enhancement → feature representation;

[0070] Knowledge-enhanced analysis method: Establish an agricultural knowledge graph query and reasoning mechanism to enhance the professionalism of feature understanding.

[0071] Preferably, the intelligent agent module is initialized: according to different decision-making areas, professional intelligent agents are initialized, including land parcel analysis engine, price analysis engine and meteorological data engine, etc.

[0072] Task allocation mechanism: According to the user's query requirements, tasks are assigned to relevant agents for processing.

[0073] Parallel computing processing: Each agent processes the assigned tasks in parallel and conducts professional field analysis.

[0074] Result integration and collaboration: Integrate the analysis results of each intelligent agent and generate final recommendations through a collaborative decision-making mechanism.

[0075] Feedback optimization loop: collect user feedback and continuously optimize the analytical capabilities and collaborative effects of each agent;

[0076] Collaborative decision-making process: task allocation → parallel processing of intelligent agents → result collection → collaborative analysis → comprehensive decision-making → feedback optimization.

[0077] Preferably, data collection and preprocessing: collect multi-dimensional data such as meteorological, soil, historical yield, etc. related to the plot and perform standardized processing.

[0078] Multi-factor adaptability analysis:

[0079] Climate adaptability: Analyze the matching degree between climate factors such as temperature, precipitation, and light and crop growth requirements;

[0080] Soil adaptability: assessing the suitability of soil texture, fertility, pH and other parameters with crop requirements;

[0081] Terrain adaptability: Consider the impact of terrain factors such as altitude, slope, and orientation on crop growth;

[0082] Economic benefit forecast: Combine market supply and demand, price trends, cost inputs and other factors to predict the economic benefits of different crops;

[0083] Risk assessment: Comprehensively consider natural risks (extreme weather, pests and diseases, etc.) and market risks (price fluctuations, changes in supply and demand, etc.);

[0084] Decision-making recommendation generation: Based on the comprehensive assessment results, planting decision recommendations are generated, including crop selection, planting time, management measures, etc.

[0085] Evaluation dimensions of land parcel analysis: adaptability to natural conditions + expected economic benefits + risk factors = comprehensive decision-making recommendations.

[0086] Preferably, multimodal input recognition: supports multiple input methods such as voice, text, and image, and performs unified processing;

[0087] Intent understanding and slot filling: Identify user query intent and extract key information (crop type, geographic location, growth stage, etc.);

[0088] Knowledge graph query: Based on intent and slot information, query the agricultural knowledge graph to obtain relevant professional knowledge;

[0089] Context-aware reasoning: Combines multiple rounds of interaction history to perform context-aware reasoning and generate coherent responses;

[0090] Personalized recommendation generation: Generate personalized decision-making recommendations based on user-specific circumstances (region, planting scale, technical level, etc.);

[0091] Interaction process: user input → intent understanding → knowledge retrieval → contextual reasoning → response generation → user feedback;

[0092] Image acquisition and enhancement: Capture crop images through mobile devices and perform pre-processing such as denoising and enhancement.

[0093] Multimodal information fusion: Combine image information and user text descriptions (symptoms, growth environment, etc.) to perform multimodal information fusion;

[0094] Feature extraction and matching: Extract crop morphology, color, texture and other features and match them with standard features in the knowledge base;

[0095] Pest and disease identification and diagnosis:

[0096] Symptom identification: Identify abnormal symptoms such as leaf spots, wilting, and discoloration;

[0097] Cause analysis: Analyze the possible causes of diseases and pests based on symptom combinations and environmental conditions;

[0098] Development prediction: predict the possible development trend and degree of harm of the disease;

[0099] Generation of prevention and control plans: Based on the diagnosis results, combined with regional characteristics and crop growth stages, targeted prevention and control suggestions are generated;

[0100] Diagnostic decision-making process: image acquisition → symptom identification → etiology analysis → severity assessment → prevention and treatment plan generation.

[0101] Compared with the existing technology, the present invention provides an agricultural intelligent analysis and decision-making method based on multimodal data, which has the following beneficial effects:

[0102] 1. This agricultural intelligent analysis and decision-making method based on multimodal data solves the technical difficulties in traditional agricultural decision-making through technological innovation in the processing, analysis and decision-making process of multi-source agricultural data, and provides accurate analysis and scientific decision-making support for the pre-production, production and post-production stages of agricultural production.

[0103] 2. This agricultural intelligent analysis and decision-making method based on multimodal data realizes the intelligent integration and analysis of multi-source agricultural data, establishes an accurate land adaptability assessment method, builds an intelligent agricultural knowledge question-and-answer system, provides high-precision crop identification and diagnosis methods, and generates explainable decision-making recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1 This is the overall technical flow chart of the present invention;

[0105] Figure 2 This is a flow chart of the plot analysis of the present invention;

[0106] Figure 3 This is a flow chart of crop identification and diagnosis according to the present invention. DETAILED DESCRIPTION

[0107] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0108] See also Figure 1-3 , an agricultural intelligent analysis and decision-making method based on multimodal data, including multimodal intelligent agents, intelligent agent collaborative decision-making technology, intelligent decision-making methods and crop identification and diagnosis methods;

[0109] Multimodal agents include agent architecture and multimodal capability building:

[0110] Agent Architecture:

[0111] Build a data processing technology framework;

[0112] Realize domain knowledge transformation mechanism;

[0113] Establishing a decision analysis process;

[0114] Multimodal capacity building:

[0115] Multimodal data processing: Fusion of heterogeneous data such as images, text, and time series data enables comprehensive analysis of agricultural scenarios;

[0116] Cross-modal feature processing: Through feature extraction and association analysis technology, cross-modal association and mapping of images, texts, and data can be achieved;

[0117] Knowledge-enhanced analysis: Applying knowledge graph technology to enhance professional analytical capabilities in the agricultural field;

[0118] Agricultural semantic processing: Optimize the parsing process of agricultural terminology and expressions, and improve the accuracy of processing agricultural context;

[0119] Intelligent agent collaborative decision-making technology includes intelligent agent ecosystem construction, land parcel analysis methods, and land parcel economic evaluation models;

[0120] Intelligent body ecosystem construction:

[0121] Land parcel analysis engine agent: responsible for adaptability analysis of soil, terrain, etc.;

[0122] Price analysis engine agent: responsible for market supply and demand, and price trend analysis;

[0123] Meteorological data engine agent: responsible for analyzing climate conditions and crop growth adaptability;

[0124] Intelligent agent collaborative decision-making system: realizes multi-agent information sharing and decision-making collaboration;

[0125] Land parcel analysis method:

[0126] Time series analysis of meteorological data: collecting climate change trends in recent years;

[0127] Soil data feature extraction: analysis of soil physical and chemical properties;

[0128] Market data collection: historical prices, supply and demand trend analysis;

[0129] Land economic evaluation model:

[0130] Planting adaptability analysis: based on natural conditions such as climate, soil, and topography;

[0131] Economic benefit forecast: based on market supply and demand, and price trend analysis;

[0132] Investment risk assessment: Consider natural risks and market risks;

[0133] Comprehensive decision-making recommendations: Generate planting investment return analysis report.

[0134] Intelligent decision-making methods include agricultural knowledge graph construction, understanding and decision-making;

[0135] Agricultural knowledge graph construction:

[0136] Structuring of expert knowledge;

[0137] Knowledge entity relationship extraction;

[0138] Knowledge graph is dynamically updated;

[0139] Understanding and Decision-Making:

[0140] Speech recognition preprocessing;

[0141] Intent understanding and slot filling;

[0142] Personalized recommendation generation.

[0143] Crop recognition and diagnosis methods include vision-language model preprocessing and intelligent recognition and diagnosis.

[0144] Vision-language model preprocessing:

[0145] Image standardization processing;

[0146] Scene text extraction;

[0147] Multimodal feature alignment;

[0148] Contextual information fusion;

[0149] Intelligent identification and diagnosis include crop identification, growth status assessment, intelligent diagnosis of pests and diseases, and generation of prevention and control plans based on VL models;

[0150] Crop recognition based on VL model: utilizing visual-linguistic bimodal features;

[0151] Growth status assessment: multi-dimensional analysis combined with expert knowledge base;

[0152] Intelligent diagnosis of pests and diseases: semantic matching based on symptom descriptions;

[0153] Generation of prevention and control plans: personalized suggestions based on regional characteristics.

[0154] Agricultural decision-making methods for multimodal data, including multimodal feature processing methods, intelligent agent collaborative decision-making framework implementation methods, plot analysis implementation methods, intelligent interaction implementation methods, and crop identification and diagnosis implementation methods;

[0155] Multimodal feature processing methods include data preprocessing technology, feature extraction methods, cross-modal feature fusion technology and knowledge enhancement analysis methods;

[0156] Data preprocessing technology: Establish standardized processing procedures for agricultural data of different modalities to eliminate noise and outliers.

[0157] Feature extraction method:

[0158] Image feature extraction: Apply algorithms to extract key features of images of crops, pests, and diseases;

[0159] Text feature extraction: Use natural language processing technology to analyze text information such as user descriptions and expert suggestions;

[0160] Time series data feature extraction: Apply time series analysis methods to process weather, output, price and other data to identify changing trends;

[0161] Cross-modal feature fusion technology: Design feature alignment algorithms and fusion mechanisms to address heterogeneity issues with data from different modalities.

[0162] Technical processing flow: data preprocessing → feature extraction → cross-modal fusion → knowledge enhancement → feature representation;

[0163] Knowledge-enhanced analysis method: Establish an agricultural knowledge graph query and reasoning mechanism to enhance the professionalism of feature understanding.

[0164] The implementation method of the intelligent agent collaborative decision-making framework includes intelligent agent module initialization, task allocation mechanism, parallel computing processing, result integration and collaboration, and feedback optimization loop;

[0165] analysis engine, price analysis engine, and weather data engine, etc.

[0166] Task allocation mechanism: According to the user's query requirements, tasks are assigned to relevant agents for processing.

[0167] Parallel computing processing: Each agent processes the assigned tasks in parallel and conducts professional field analysis.

[0168] Result integration and collaboration: Integrate the analysis results of each intelligent agent and generate final recommendations through a collaborative decision-making mechanism.

[0169] Feedback optimization loop: collect user feedback and continuously optimize the analytical capabilities and collaborative effects of each agent;

[0170] Collaborative decision-making process: task allocation → parallel processing of intelligent agents → result collection → collaborative analysis → comprehensive decision-making → feedback optimization.

[0171] The implementation methods of plot analysis include data collection and preprocessing, multi-factor adaptability analysis, economic benefit prediction, risk assessment and decision-making recommendation generation;

[0172] Intelligent interaction implementation methods include multimodal input recognition, intent understanding and slot filling, knowledge graph query, context-aware reasoning, and personalized suggestion generation;

[0173] Data collection and preprocessing: Collect multi-dimensional data related to the plot, such as meteorological, soil, and historical yield, and perform standardized processing.

[0174] Multi-factor adaptability analysis:

[0175] Climate adaptability: Analyze the matching degree between climate factors such as temperature, precipitation, and light and crop growth requirements;

[0176] Soil adaptability: assessing the suitability of soil texture, fertility, pH and other parameters with crop requirements;

[0177] Terrain adaptability: Consider the impact of terrain factors such as altitude, slope, and orientation on crop growth;

[0178] Economic benefit forecast: Combine market supply and demand, price trends, cost inputs and other factors to predict the economic benefits of different crops;

[0179] Risk assessment: Comprehensively consider natural risks (extreme weather, pests and diseases, etc.) and market risks (price fluctuations, changes in supply and demand, etc.);

[0180] Decision-making recommendation generation: Based on the comprehensive assessment results, planting decision recommendations are generated, including crop selection, planting time, management measures, etc.

[0181] Evaluation dimensions of land parcel analysis: adaptability to natural conditions + expected economic benefits + risk factors = comprehensive decision-making recommendations.

[0182] Crop identification and diagnosis implementation methods include image acquisition and enhancement, multimodal information fusion, feature extraction and matching, pest and disease identification and diagnosis, and prevention and control plan generation.

[0183] Multimodal input recognition: supports multiple input methods such as voice, text, and images, and performs unified processing;

[0184] Intent understanding and slot filling: Identify user query intent and extract key information (crop type, geographic location, growth stage, etc.);

[0185] Knowledge graph query: Based on intent and slot information, query the agricultural knowledge graph to obtain relevant professional knowledge;

[0186] Context-aware reasoning: Combines multiple rounds of interaction history to perform context-aware reasoning and generate coherent responses;

[0187] Personalized recommendation generation: Generate personalized decision-making recommendations based on user-specific circumstances (region, planting scale, technical level, etc.);

[0188] Interaction process: user input → intent understanding → knowledge retrieval → contextual reasoning → response generation → user feedback;

[0189] Image acquisition and enhancement: Capture crop images through mobile devices and perform pre-processing such as denoising and enhancement.

[0190] Multimodal information fusion: Combine image information and user text descriptions (symptoms, growth environment, etc.) to perform multimodal information fusion;

[0191] Feature extraction and matching: Extract crop morphology, color, texture and other features and match them with standard features in the knowledge base;

[0192] Pest and disease identification and diagnosis:

[0193] Symptom identification: Identify abnormal symptoms such as leaf spots, wilting, and discoloration;

[0194] Cause analysis: Analyze the possible causes of diseases and pests based on symptom combinations and environmental conditions;

[0195] Development prediction: predict the possible development trend and degree of harm of the disease;

[0196] Generation of prevention and control plans: Based on the diagnosis results, combined with regional characteristics and crop growth stages, targeted prevention and control suggestions are generated;

[0197] Diagnostic decision-making process: image acquisition → symptom identification → etiology analysis → severity assessment → prevention and treatment plan generation.

[0198] Example 1

[0199] Agricultural intelligent analysis and decision-making methods based on multimodal data have been applied and verified in Tudi AI, achieving good results:

[0200] Multimodal Agent Applications:

[0201] Improved cross-modal comprehension capabilities;

[0202] Enhanced the level of agricultural semantic understanding;

[0203] Improved farmers’ satisfaction;

[0204] Agent collaborative decision-making framework:

[0205] Accelerated decision-making response speed;

[0206] Improved multi-agent coordination effects;

[0207] Optimized the efficiency of agricultural inputs;

[0208] Land parcel analysis:

[0209] Improved the level of planting adaptability assessment;

[0210] Enhanced economic benefit forecasting capabilities;

[0211] Improved the adoption rate of investment recommendations;

[0212] Crop Identification:

[0213] Improved variety recognition accuracy;

[0214] Improved the ability to assess growth status;

[0215] Enhanced accuracy of disease and insect pest diagnosis;

[0216] Improved the effectiveness of prevention and control programs;

[0217] Through the integrated application of multimodal intelligent agents and intelligent collaborative decision-making frameworks, this invention effectively improves the intelligence and precision of agricultural production decisions, and has significant practical value and promotion prospects.

[0218] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An agricultural intelligent analysis based on multimodal data, characterized in that: Including multimodal intelligent agents, intelligent agent collaborative decision-making technology, intelligent decision-making methods and crop identification and diagnosis methods; Multimodal agents include agent architecture and multimodal capability building: Intelligent agent collaborative decision-making technology includes intelligent agent ecosystem construction, land parcel analysis methods, and land parcel economic evaluation models; Intelligent decision-making methods include agricultural knowledge graph construction, understanding and decision-making; Crop recognition and diagnosis methods include vision-language model preprocessing and intelligent recognition and diagnosis.

2. The agricultural intelligent analysis based on multimodal data according to claim 1, characterized in that: Agent Architecture: Build a data processing technology framework; Realize domain knowledge transformation mechanism; Establishing a decision analysis process; Multimodal capacity building: Multimodal data processing: Fusion of heterogeneous data such as images, text, and time series data enables comprehensive analysis of agricultural scenarios; Cross-modal feature processing: Through feature extraction and association analysis technology, cross-modal association and mapping of images, texts, and data can be achieved; Knowledge-enhanced analysis: Applying knowledge graph technology to enhance professional analytical capabilities in the agricultural field; Agricultural semantic processing: Optimize the parsing process of agricultural professional terms and expressions, and improve the accuracy of processing agricultural contexts.

3. The agricultural intelligent analysis based on multimodal data according to claim 1, characterized in that: Intelligent body ecosystem construction: Land parcel analysis engine agent: responsible for adaptability analysis of soil, terrain, etc.; Price analysis engine agent: responsible for market supply and demand, and price trend analysis; Meteorological data engine agent: responsible for analyzing climate conditions and crop growth adaptability; Intelligent agent collaborative decision-making system: realizes multi-agent information sharing and decision-making collaboration; Land parcel analysis method: Time series analysis of meteorological data: collecting climate change trends in recent years; Soil data feature extraction: analysis of soil physical and chemical properties; Market data collection: historical prices, supply and demand trend analysis; Land economic evaluation model: Planting adaptability analysis: based on natural conditions such as climate, soil, and topography; Economic benefit forecast: based on market supply and demand, and price trend analysis; Investment risk assessment: Consider natural risks and market risks; Comprehensive decision-making recommendations: Generate planting investment return analysis report.

4. The agricultural intelligent analysis based on multimodal data according to claim 1, characterized in that: Agricultural knowledge graph construction: Structuring of expert knowledge; Knowledge entity relationship extraction; Knowledge graph is dynamically updated; Understanding and Decision-Making: Speech recognition preprocessing; Intent understanding and slot filling; Personalized recommendation generation.

5. The agricultural intelligent analysis based on multimodal data according to claim 1, characterized in that: Vision-language model preprocessing: Image standardization processing; Scene text extraction; Multimodal feature alignment; Contextual information fusion; Intelligent identification and diagnosis include crop identification, growth status assessment, intelligent diagnosis of pests and diseases, and generation of prevention and control plans based on VL models; Crop recognition based on VL model: utilizing visual-linguistic bimodal features; Growth status assessment: multi-dimensional analysis combined with expert knowledge base; Intelligent diagnosis of pests and diseases: semantic matching based on symptom descriptions; Generation of prevention and control plans: personalized suggestions based on regional characteristics.

6. An agricultural decision-making method based on multimodal data, characterized in that: Including multimodal feature processing methods, intelligent collaborative decision-making framework implementation methods, plot analysis implementation methods, intelligent interaction implementation methods and crop identification and diagnosis implementation methods; Multimodal feature processing methods include data preprocessing technology, feature extraction methods, cross-modal feature fusion technology and knowledge enhancement analysis methods; The implementation method of the intelligent agent collaborative decision-making framework includes intelligent agent module initialization, task allocation mechanism, parallel computing processing, result integration and collaboration, and feedback optimization loop; The implementation methods of plot analysis include data collection and preprocessing, multi-factor adaptability analysis, economic benefit prediction, risk assessment and decision-making recommendation generation; Intelligent interaction implementation methods include multimodal input recognition, intent understanding and slot filling, knowledge graph query, context-aware reasoning, and personalized suggestion generation; Crop identification and diagnosis implementation methods include image acquisition and enhancement, multimodal information fusion, feature extraction and matching, pest and disease identification and diagnosis, and prevention and control plan generation.

7. The agricultural decision-making method based on multimodal data according to claim 6, characterized in that: Data preprocessing technology: Establish standardized processing procedures for agricultural data of different modalities to eliminate noise and outliers; Feature extraction method: Image feature extraction: Apply algorithms to extract key features of images of crops, pests, and diseases; Text feature extraction: Use natural language processing technology to analyze text information such as user descriptions and expert suggestions; Time series data feature extraction: Apply time series analysis methods to process weather, output, price and other data to identify changing trends; Cross-modal feature fusion technology: Design feature alignment algorithms and fusion mechanisms to address heterogeneity issues with data from different modalities. Technical processing flow: data preprocessing → feature extraction → cross-modal fusion → knowledge enhancement → feature representation; Knowledge-enhanced analysis method: Establish an agricultural knowledge graph query and reasoning mechanism to enhance the professionalism of feature understanding.

8. The agricultural decision-making method based on multimodal data according to claim 6, characterized in that: Agent module initialization: Initialize specialized agents based on different decision-making areas, including the land parcel analysis engine, price analysis engine, and weather data engine; Task allocation mechanism: assign tasks to relevant agents for processing based on user query requirements; Parallel computing: Each agent processes assigned tasks in parallel and performs specialized field analysis. Result integration and collaboration: Integrate the analysis results of each agent and generate final recommendations through a collaborative decision-making mechanism; Feedback optimization loop: collect user feedback and continuously optimize the analytical capabilities and collaborative effects of each agent; Collaborative decision-making process: task allocation → parallel processing of intelligent agents → result collection → collaborative analysis → comprehensive decision-making → feedback optimization.

9. The agricultural decision-making method based on multimodal data according to claim 6, characterized in that: Data collection and preprocessing: Collect multi-dimensional data related to the plot, such as meteorological, soil, and historical yield, and perform standardized processing; Multi-factor adaptability analysis: Climate adaptability: Analyze the matching degree between climate factors such as temperature, precipitation, and light and crop growth requirements; Soil adaptability: assessing the suitability of soil texture, fertility, pH and other parameters with crop requirements; Terrain adaptability: Consider the impact of terrain factors such as altitude, slope, and orientation on crop growth; Economic benefit forecast: Combine market supply and demand, price trends, cost inputs and other factors to predict the economic benefits of different crops; Risk assessment: Comprehensively consider natural risks (extreme weather, pests and diseases, etc.) and market risks (price fluctuations, changes in supply and demand, etc.); Decision-making recommendation generation: Based on the comprehensive assessment results, planting decision recommendations are generated, including crop selection, planting time, management measures, etc. Evaluation dimensions of land parcel analysis: adaptability to natural conditions + expected economic benefits + risk factors = comprehensive decision-making recommendations.

10. The agricultural decision-making method based on multimodal data according to claim 6, characterized in that: Multimodal input recognition: supports multiple input methods such as voice, text, and images, and performs unified processing; Intent understanding and slot filling: Identify user query intent and extract key information (crop type, geographic location, growth stage, etc.); Knowledge graph query: Based on intent and slot information, query the agricultural knowledge graph to obtain relevant professional knowledge; Context-aware reasoning: Combines multiple rounds of interaction history to perform context-aware reasoning and generate coherent responses; Personalized recommendation generation: Generate personalized decision-making recommendations based on user-specific circumstances (region, planting scale, technical level, etc.); Interaction process: user input → intent understanding → knowledge retrieval → contextual reasoning → response generation → user feedback; Image acquisition and enhancement: Capture crop images through mobile devices and perform pre-processing such as denoising and enhancement; Multimodal information fusion: Combine image information and user text descriptions (symptoms, growth environment, etc.) to perform multimodal information fusion; Feature extraction and matching: Extract crop morphology, color, texture and other features and match them with standard features in the knowledge base; Pest and disease identification and diagnosis: Symptom identification: Identify abnormal symptoms such as leaf spots, wilting, and discoloration; Cause analysis: Analyze the possible causes of diseases and pests based on symptom combinations and environmental conditions; Development prediction: predict the possible development trend and degree of harm of the disease; Generation of prevention and control plans: Based on the diagnosis results, combined with regional characteristics and crop growth stages, targeted prevention and control suggestions are generated; Diagnostic decision-making process: image acquisition → symptom identification → etiology analysis → severity assessment → prevention and treatment plan generation.

Images