Abstract: Plant intelligence is driven by the real-time sensing data of Things Internet and provide intelligent decision by knowledge model and multi-source heterogeneous data fusion calculation to realize the intelligent process of connectivity, whole perception, real-time feedback between plant and environment, plant and people, plant and plant. This paper discussed the concept, connotation and technical system of intelligent plant, and the construction method of intelligent plant system based on Multi-Agent. Aiming at the complex isomerism of intelligent plant behavior, the Multi-Agent technology was applied in plant growth count and behavior simulation. From service-oriented application in intelligent plant, the paper comprehensively analyzed intelligent property, concept and connotation including plant autonomy, reactiveness, sociality, and initiative, and then the intelligent plant system was constructed. The Kernel intelligent plant system consists of sensor agent, data processing agent, knowledge model agent, intelligent calculation agent, and simulation interaction agent. This paper described in detail the system structure and its key model. In the Multi-Agent system, plant could perceive its environment context, and independently make behavior decision. The intelligent plant system was constructed based on Multi-Agent, and took a fruit tree pruning virtual training service case as an example. This system showed strong interactivity, high intelligence degree, good experience effect. So, this research might provide useful help for the development of Intelligent Plant.

Key words: intelligent plant, Multi-Agent, technical system, Ontology Modeling, Service agent

摘要： 植物的智能化是在物联网传感数据实时驱动下，由知识模型和多源异构数据融合计算提供智能决策，从而实现植物与环境、植物与人、植物与植物之间互联互通、全程感知与实时反馈的智能化过程，探讨了智能植物的概念、内涵、技术体系及基于Multi-Agent的智能植物系统构建方法。针对植物智能行为复杂异构性,将Multi-Agent技术应用于植物的生长计算和行为模拟，从面向服务的智能植物农业应用出发，综合分析了植物的自治性、反应性、社会性和主动性等智能属性、概念和内涵，构建了多智能体智能植物系统。多智能体智能植物系统的内核由感知器Agent、数据处理Agent、知识模型Agent、智能计算Agent以及虚拟交互Agent构成，并详细描述了系统结构及其关键模型，在多Agent系统中,植物能够感知其所处的环境上下文,自主地决策其行为。构建了基于Multi-Agent的智能植物系统，并以果树修剪虚拟培训智能服务为例验证，结果表明，该系统交互性强、智能程度高、体验效果好，以期为智能植物的发展提供有益的帮助。

关键词: 智能植物, Multi-Agent, 技术体系, 本体建模, 服务代理