【作者】 王纪章；

【导师】 李萍萍；

【作者基本信息】 江苏大学 ， 农业生物环境与能源工程， 2013， 博士

【摘要】 物联网技术是通过各种信息传感设备,按约定的协议,将任何物品与互联网相连接,进行信息交换和通讯,以实现智能化识别、定位、追踪、监控和管理的一种网络技术。近年来物联网技术在各领域的应用得到迅速发展,如何将该技术很好地应用到温室环境智能化调控管理中,对于提高设施农业的现代化水平具有重要的意义。本文通过分析国内外研究现状,针对物联网环境下的温室环境智能管理系统中服务层和决策层研究缺乏的现状,开展了温室环境无线测控网络、网络中多传感器信息融合、温室环境信息处理及调控效果预测模型、作物模型库、多尺度的环境调控优化、基于物联网的温室智能管理系统开发等进行了系统地研究。主要研究结论包括：(1)构建了基于Zigbee/3G的温室环境无线测控网络,并针对无线传感器网络的三层结构特点,提出了传感节点+汇聚节点两层的融合方法,利用卡尔曼滤波方法进行传感节点层融合,利用加权最小二乘方法进行汇聚节点层融合。对所建立的模型进行了试验验证,结果表明通过两层的融合处理可以提高温室环境无线测控网络的测量精度和系统的稳定性。在无线测控网络基础上,建立了基于多Agent的无线测控网络资源管理系统,实现对温室环境无线传感器节点的管理。(2)针对温室内外环境信息特点,依据温室环境调控规则,提出了基于支持向量机预测和多模型切换的温室环境调控预测模型。采用在线式支持向量机算法建立了室外气象预测模型,采用增量式支持向量机算法建立了环境调控预测模型库,采用多模型切换控制器实现子模型的自适应切换。实验结果表明建立的调控效果预测模型具有较好的预测精度,并可实现模型的自适应切换。(3)为实现在定期上市目标下的温室蔬菜作物的生产规划,以作物积温模型为依据,利用历史气象数据和市场价格信息,建立基于积温模型的温室蔬菜生产长尺度环境规划决策模型,模型能实现在定植时间确定条件下预计上市期及逐日环境优化决策、在作物计划定植日期和上市期确定条件下的逐日环境优化决策、温室运行过程中的日环境优化决策。(4)针对温室蔬菜作物生长的长尺度和温室环境变化的短尺度之间的协调困难的问题,提出了基于多模型融合的温室环境调控参数优化方法,对基于积温、设定值和光照的温室环境优化调控利用D-S证据理论进行多模型的融合,实现温室环境参数的优化调控,试验温室环境调控结果表明,采用多模型融合的温室环境优化调控方法与设定值调控方法相比增加了积温,实现了温室的高产高效生产。(5)针对作物模拟模型的通用性和可重用性差的问题,建立了基于多智能体的温室作物模拟模型库系统框架,系统采用访问层、模型集成层和数据层三层结构,设计了模型Agent、管理Agent、目录服务器、通信系统和访问Agent。基于JADE软件平台,开发了基于多Agent的温室作物模拟模型库系统,并利用黄瓜的生长模拟模型对系统进行了测试。(6)建立了基于XML的温室作物环境信息交换标准化接口。开发了基于Agent的温室环境调控智能决策支持系统,并将温室环境无线测控网络与资源管理系统、温室环境调控智能决策支持系统集成,建立了基于物联网的温室环境智能管理系统。系统在镇江市京口区瑞京农业科技示范园智能温室中运行,实现了温室环境智能管理。更多还原

【Abstract】 The technology for the Internet of Things (IOT) is to connect everything to the Internet for information exchange and telecommunications through a variety of information sensing devices according to the given protocols, so as to realize the intelligentized identification, location, tracking, monitoring and management. IOT technology has been applied to various fields in recent years. In this paper, after analyzing the research status of the technology of Internet of things for intelligent measurement and controlling technology in the greenhouse environment, the studies have made for the wireless measurement and control system of greenhouse climate based on multi-sensor information fusion, climate information processing, crop model base, and the optimization of scale environmental regulation. The details of the sudies for the greenhouse intelligent management system based on Internet of things main include as followings:(1) The measurement and controlling system of wireless sensing networks for greenhouse climate based on Zigbee/3G has been built, and the multi-sensor information fusion models of node/cluster layers were proposed according to the characteristics of the three-layer structure of wireless sensing network. The node level fusion has been conducted by using the kalman filtering model, and the cluster level fusion by using weighted least square model. The experimental results showed that the two-level fusion models can improve the measurement accuracy and stability of the wireless sensing networks.(2) According the climate information measured with WSN system of the greenhouse environment and in conformity to climate control rules, the controlling-effect model of greenhouse climate has been proposed on the base of support vector machine (SVM) and multi-model switching models. The outdoor meteorological forecast models have been established based on least squares support vector machine (LSSVM) algorithms with the online learning. By using the incremental learning LSSVM algorithms, the greenhouse climate effect forecast models have been establish, and the multi-model switch controllers have been used to realize the adaptive switch of control effect models. The established climate control effect models were verified through the field experiments, and the results showed that the climate control effect models can obtain the satisfactory prediction precision, thus realizing the adaptive switch of climate control effect models. The management system of wireless sensor network resource based on multi-Agent has been established to realize the management of wireless sensors.(3) In order to accomplish the production planning of the vegetables in the greenhouse targeting for the sale of certain date, the environmental plan decision support models for long-scale greenhouse have been established on the basis of the temperature-accumulated models of crops and by using the history climate data and market price information etc. This model can achieve the decision support for the forecast of sale date and price with the given plant establishment, and determine the day-by-day optimized decision of environment, as well as to make the day-by-day optimized decision of the environment in the conditions of predicted plant establishment date and the certain sale date, while making the daily optimized decision of the environment of the greenhouse operation etc.(4) In order to realize the coordination between the long-scale changes of crop growth and short-scale of climate changes, the realtime greenhouse climate control parameters optimal model was proposed on the basis the multi-model fusion. The model was integrated for the fusion of the control model of accumulation temperature model, settings of model and the light-based models, where the mutil-model fusion is realized by using the D-S evidence theory. Meanwhile, the experimental model with the multi-model fusion was verified in the testing greenhouse. The results show that the multi-model fusion controlling models can increase the accumulation temperatures more than those of the setting values for model, thus realizing the yields of greenhouse.(5) In order to resolve the issues of poor generality and repeatability of the crop simulation models, the systematic framework has been established for the greenhouse crop growth simulation model database on the basis of multi-agent. The model database consists of the three layers of access layer, model integration layer and data layer. And it is designed as the model agent, management agent, catalog server agent, communication system and access agent. The Agent-based software of the crop growth simulation model in the greenhouse was developed on the JADE platform. And the growth simulation model for the cucumber was used to test this system.(6) The standardized information exchange interface has been established for crop growth environment information in the greenhouse on XML. The Agent-based greenhouse climate controlling decision support system was developed. The software of greenhouse intelligent management system for IOT was developed, so the system can be integrated with WSN resources management, the greenhouse climate controlling decision support system and information exchange interface. This system has succeeded to be applied in the greenhouse of Ruijin Agricultural Scientific and Technologic demonstration park at Jinkou district of Zhenjiang, which can have realized the intelligentized management for greenhouse.更多还原