【作者】 齐林；

【导师】 傅泽田；

【作者基本信息】 中国农业大学 ， 农业工程， 2014， 博士

【摘要】 可追溯系统是保证生鲜农产品质量安全的重要手段,当前技术构架下可追溯系统面临的感知数据采集能力欠缺、追溯数据粒度输出单一、追溯平台体系结构薄弱等瓶颈,阻碍了系统的规模化应用。物联网技术的发展,使生鲜农产品质量安全可追溯系统突破技术与应用的瓶颈成为可能。本研究从可追溯系统的三个技术瓶颈出发,紧紧围绕物联网“无处不在的数据采集、可靠的数据传输与信息处理、智能化的信息应用”三个核心内涵,以动、植物源性农产品可追溯供应链为研究对象,构建了物联网环境下可追溯系统数据采集与建模方法,研发了基于WSN的生鲜农产品质量安全可追溯感知数据采集硬件与嵌入式软件,设计了基于SPC的可追溯感知数据时域压缩方法,针对用户数据粒度需求的差异,进行了可追溯数据兼容建模和面向粒度分级的规约,设计和实现了基于云计算的可追溯综合服务平台。研究的主要贡献与创新之处是：(1)提出了基于WSN的可追溯感知数据采集方法和基于SPC的时域压缩方法,提高了感知数据采集效率,并延长了监测网络寿命。基于WSN所研发的可追溯感知数据采集方法软、硬件原型,测试结果表明通信链路可靠,感知节点对生鲜农产品供应链保鲜工艺环境的兼容性、传感器硬件兼容性好：基于SPC所设计的改进X-Rs感知数据时域压缩算法与阈值、K-滑动均值算法对比,能耗在同一数量级,平稳时间序列的Se为最优,2种时间序列平稳性上tc值均接近最优,算法的平衡性和适应性好。(2)提出了面向粒度分级的可追溯系统建模方法,满足了不同用户的数据粒度需求。基于结构模式识别,构造了描述追溯单元转化的12种模式基元；基于关系代数,设计了模式基元的数据存储结构与数据采集算法；构建了基于2型文法的可追溯数据形式化描述文法和文法句子生成算法；基于改进下推自动机建立了粒度分级规约方法；以冻罗非鱼片加工、半滑舌鳎养殖、肉牛养殖与屠宰加工业务流程为实例进行了方法验证,结果表明在以上供应链,数据分级规约强度为44.8-99.4%,在供应链结构信息少的初级农产品生产流程中,规约强度最高。(3)设计了基于云计算的可追溯综合服务平台,实现了平台级可追溯服务。识别了可追溯数据在生鲜农产品供应链上各阶段的潜在价值,包括文档标准化、危害溯源、精确召回、物流监控、关键点预警、质量预测、货架期管理和库存优化；基于Hadoop设计了平台的技术架构、服务引擎、体系结构,基于Map/Reduce实现了决策模型并行化；在Ubuntu10.0.4操作系统和Hadoop0.20.0并行计算环境上进行了平台实现；以工厂化水产养殖、水产品冷链物流为例的系统评价表明平台在数据采集、信息追溯和智能决策等方面改善了生鲜农产品供应链管理水平。更多还原

【Abstract】 Traceability system is an important measure of ensuring the quality and safety of agro-product. Lack of sensing data acquisition capability, monotonous of output traceability data granularity and weakness in system service architecture are current obstacles stands on the way of the traceability system scaled applications. The development of Internet of Things (IoT) made it possible for traceability system to breakthrough its bottlenecks.This study started with the analysis of traceability system’s technical bottlenecks, based on3core substances of IoT which are ubiquitous data collection, reliable data transmission with information processing and intelligent information applications, took the animal and plant derived agro-product’s traceable supply chain as research object, built a traceability conception model in IoT context, developed the embanded software and hardware prototype for sensing data collection, designed a SPC based sensing data compression algorithm, established a tracability data compatible modeling and granularity classification and statute algorithm, designed and implemented a cloud computing based agro-product quality and safety traceability service platform. The contributions and innovations of this study are as follows:(1) Developed a WSN-based traceability sensing data acquisition method to improve data collection efficiency and a SPC-based traceability sensing data compressing method in time domain to extend the monitoring network lifecycle. System test on the WSN-based traceability sensing data acquisition software and hardware prototype showed that the data communication link was stable and sensor node could endure the agro-product preservation environment and compatible with digital and analog sensors. Cooperation with the threshold and K-means algorithms, the SPC based improved X-Rs data compression algorithm was in the same order of magnitude in energy cost, with best Se in stationary time series and closed to the optimal tc in2kinds of time series. The X-Rs algorithm showed balance and adaptability.(2) Designed a granularity classification oriented traceability data modeling method to meet information requirement of different user groups. Identified12pattern primitives in traceability supply chain.Constructed the type2grammar based traceability supply chain data structure formal description, recursive based grammar sentence generation and pushdown automata based data granularity statute.4agro-product supply chains’internal traceability work flows were taken as test beds to verify the model. Results showed that the data granularity statute strength were44.8-99.4%, the primary agro-product process with less structure information gained highest statute strength.(3) Implemented an integrated service platform for traceability based on cloud computing and achieved platform-level traceability service. Document standardization, harzard traceability, accurate recall et al. functions were identified as key fuctional requirements of the service platform. Platform technical architechture and service engine were designed based on Hadoop framework, decision support model are parallelized using Map/Reduce model. Platform was implemented based on Ubuntu10.0.4OS and Hadoop0.20.0. System evaluation showed that agro-product supply chian management was improved in3aspects.更多还原