【作者】 高嵘；

【导师】 秦志光；

【作者基本信息】 电子科技大学 ， 信息管理与电子商务， 2011， 博士

【摘要】 食品安全不仅关系到人民群众身体健康和生命安全,还关系到经济发展和社会稳定。党中央、国务院历来把保护人民群众身体健康和生命安全放在第一位,十分重视食品安全监管工作,全面提高农产品质量安全水平已成为一项全局性的战略任务。本文立足民生产业领域,以猪肉产品安全为应用蓝本,对基于物联网的猪肉溯源模型及生猪价格预警模型进行研究。研究内容主要有：溯源过程控制和溯源体系；溯源数据的采集、传输和存储；高性能计算和大规模数据备份；猪肉价格风险预警等。研发了猪肉食品安全溯源系统和多部门协同监管系统两套核心软件,最终建立一个基于物联网的猪肉食品安全溯源应用示范体系。本文的主要创新点有：1、建立了基于支持向量机的生猪价格预警模型。该模型以支持向量机理论为基础,猪粮价格比率为指标,在小样本量的条件下对猪肉价格风险等级进行分析预测。实证分析表明,该模型适用于小样本量,并能以较高的准确度对生猪价格的风险等级进行预测,从而实现为政府部门提供生猪价格预警和抉择信息。为样本量不足的条件下进行生猪价格风险预测找到了新的方法。2、提出了基于Panel Data模型的猪肉销售价格影响因素和区域特征分析的方法。通过对商务部畜禽屠宰统计监测系统所收集的2008-2009年我国32个地区的生猪平均收购价、屠宰量、生产量、销售量、白条肉出厂价等指标进行实证分析,表明我国猪肉销售价格具有显著的正相关性,主要受到生猪收购价格因素的影响；除贵州、兵团地区外,其余地区的销售价格对屠宰量、生产量和销售量三个因素的变化均不敏感。3、提出了基于物联网的食品溯源应用模型、食品溯源有限状态机模型和食品溯源时序数据的函数聚类分析。基于物联网的食品溯源应用模型融合了多种信息技术,并对众多的异构信息进行转换、融合和挖掘,实现食品安全追溯信息管理,以及信息化时代下政府食品安全高效监管新模式。食品溯源有限状态机模型对溯源过程中的状态转换进行了分析和模拟,为进一步研究食品溯源系统的行为描述和结构设计提供了理论基础,使得溯源流程更加合理和有效。猪肉供应链跟踪与溯源信息是大量的时序数据,食品溯源时序数据的函数型聚类分析把时序数据看成一个完整的关于时间的函数对象,对溯源数据进行分析,为决择者提供依据。4、建立了基于物联网的猪肉溯源应用示范,创新了食品安全监管模式。针对中国政府对食品安全监管的工作特点,利用物联网技术对生猪产业链信息进行整合,创立了政府多部门联合监管模式和生猪供应链业务管理体系,形成了完整的物联网技术在食品安全行业的应用示范体系和应用模式。变食品的不可溯源为可溯源,创新了食品安全监管模式,实现了生猪供应链局部独立监管向生猪供应链全流程协同监管的转变；实现了部门监管向社会参与的转变；实现了单向监管向双向监管的转变；实现了人力监管向技术监管的转变；实现了粗放式监管向精细化监管的转变。详细、准确、实时的溯源信息保证了一旦出现食品安全问题,可第一时间追溯到源头,极大提升了政府的监管能力和水平。更多还原

【Abstract】 Food safety is not only critical to public health and safety, but also closely related to economic development and social stability. Our government has always been dedicating herself to the protection of people’s health and life safety, makes it a basic national policy, and attaches great importance to the supervision and management of the food safety. Comprehensively improve the quality and safety of agricultural products has become a global strategic mission. This thesis considers the pork products safety problem which comes from the livelihood business that is very closely related and extremely important to our people’s daily life, we study on the backward source-tracing model and the price precaution model of the pork products from the perspective of the Internet of Things. The major research topics of this thesis include: food source-tracing process control scheme and management system; food source-tracing data collection, transport and storage; high-performance computation and large-scale data backup; precaution of the pork price risks, etc. We developed two core software systems, one for pork products safety backward source-tracing, another for multi-authorities cooperating supervision of the food safety problems. Based on these two systems, we finally set up a demonstrating application system for pork products backward source-tracing, which can be adapted to many practical scenarios of the Internet of Things. In summary, the major achievements of this thesis are outlined as follows:Firstly, we proposed a novel SVM-based pork price precaution model. This model was built upon the support vector machine theory, by making use of the pig feed price ratio as predictors, it can provide analysis and precaution support on risk levels of the pork prices under small sample conditions. Empirical studies show that the proposed model is adapted to small sample analysis, and can perform risk level precaution on pork prices with high accuracy, so as to provide necessary information for government decision-making. These research results offer us a new way to predict price risks of the pork products under limited sample conditions--which was typical in practice.Secondly, by using of the Panel Data Model, we proposed a novel approach for analyzing on the impact factors and regional characteristics of the sale prices of the pork products. We then perform empirical study on the public dataset issued by the Department of Commerce of our country, it was collected from thirty two regions of our country through a statistical monitoring system for livestock and poultry slaughter, which consists of many characteristics including the purchase price, the slaughter number, the production and sale numbers, and the ex-factory price of carcass meat, etc. According to our analysis, the sales price of the pork products of our country has significant positive correlation with the purchase price of live pigs. Except for Gui-Zhou province and construction corps areas, the sales price changes are not sensitive to the changes of the slaughter numbers, the production and the sale numbers in other areas.Thirdly, we proposed a practical backward food source-tracing model based on the Internet of Things application scenario, a finite state automaton model for food backward source-tracing, and a clustering analysis approach for time series analysis of the food trace data. The RFID based food source-tracing application model was integrated with a variety of information technology, through conversion, fusion and data mining techniques on a large number heterogeneous information sources, make it practical to manage the whole food trace system. This model also provides a new scheme for our government to efficiently supervise the food safety. The finite state automaton model for food traceability was responsible for analyzing and simulation of the state transition process, so as to make the process more rational and efficient. The data generated by the supply chain and backward source trace procedure were typically a huge amount of time series data, our clustering analysis approach treated it as a complete functional object with respect to time, and then perform data analysis on it to provide the basis for decision making.Finally, we build up a demonstrating application system for pork products backward source-tracing based on the Internet of Things, and this is an innovation for food safety supervision and management. According to the characteristics of the food safety supervision of our government, by using of the internet of things techniques to combine the information from pork products supply chains, we set up a joint supervision scheme between government bodies and a pork products supply chain management system, thus formed a demonstration system and application mode of the internet of things techniques for the food safety industry. This approach makes the food products traceable, brings us an innovative approach for food safety supervision and regulation, changes the way monitoring supply chain of the pork products from local independent monitoring to collaborative monitoring during the whole process; shares the supervision responsibility with communal participation; changes the supervision work mode from single way to duplex way; changes the work mode from human supervision to technical supervision; changes the manner of the government management from extensive to fine-grit supervision. Detailed, accurate, and real-time source-tracing information makes sure that once occurring a food safety event, it can be traced back to the source in the first moment, and this will greatly promote the supervisory capacity and level of our government in public health and safety area.更多还原