关键词: 农业, 大灾, 风险, 智能, 大模型, 评估, 决策

Abstract: To meet the full-chain demands of agricultural disaster monitoring and identification, dynamic early warning, yield prediction, loss assessment and insurance claims, and to effectively support multi-scenario, multi-user agricultural catastrophe risk analysis and intelligent decision-making services, an integrated platform for agricultural catastrophe risk analysis and intelligent decision-making has been designed and developed. Guided by the design principles of integration, modularization, intelligence and platformization, based on the data layer, platform layer, application layer, presentation layer composed of the overall architecture, the platform employs key technologies such as data integration based on the meteorological big data cloud platform, multi-source heterogeneous algorithm integration based on cloud-native, low-code development via geospatial processing automation, micro-frontend and microservices architecture, intelligent product generation based on artificial intelligence generated content, and visualization of major agricultural disaster risks based on large artificial intelligence models. By building a data middle platform, a technology middle platform and a business middle platform, it enables functions including meteorological element monitoring, identification of agricultural meteorological disasters, disaster impact assessment, and risk-informed decision support. Preliminary applications demonstrate that the platform exhibits strong operational capabilities and promising potential for development. It contributes to enhancing cross-departmental, multi-role collaborative analysis, intelligent reasoning and decision-making throughout the entire workflow, thereby effectively addressing climate change challenges, safeguarding food security and advancing the modernization of agricultural governance.

Key words: agriculture, great disaster, risk, intelligence, large model, evaluation, decision-making