食品安全风险过程控制的贝叶斯统计与知识发现
|
摘要
食品安全已成为世界范围内广泛关注的问题,理想的食品质量控制模式是“从农田到餐桌”的全过程质量控制。食品安全追溯系统提供了“从农田到餐桌”的追溯模式,建立了食品安全信息数据库。应用食品安全追溯系统,一旦发现问题,能够通过溯源进行有效的控制和召回,从源头上保障消费者的合法权益。但这依然是一种事后控制,并不能进行食品安全风险预警与控制。
本文利用食品安全追溯系统各管理过程的抽样、检验以及有关的监控数据,提取影响食品安全风险因素,获取因素变化取值,建立基于贝叶斯网络的知识表示与知识推理模型,以达到食品安全风险的预测、预警与控制。由于食品链中涉及的食品原料、加工、包装、储藏、运输、销售、消费等环节都会对最终食品的安全性造成影响,本文选取各环节的安全状况作为节点变量,变量的安全状况依赖于不同的危害后果严重程度与可能性。通过由危害后果严重程度与可能性定义的食品安全潜在风险,赋予各环节随机变量的取值,获取相关样本,在此基础上,建立贝叶斯网络结构,并依据网络中各节点条件概率分布的先验信息及获取的样本信息,应用贝叶斯估计方法进行网络参数的估计与更新学习,在食品安全各环节潜在风险程度的影响下,实现原因推理原因、原因推理结果、结果诊断原因的风险知识推理。进一步地,对已建立的食品安全风险推理的贝叶斯网络,针对贝叶斯网络知识推理的计算过程复杂与耗时问题,进行简化知识推理研究,在软件Matlab辅助推理下,论证了具有更高效率的简化知识推理的实现过程。
本文建立的基于贝叶斯网络食品安全风险知识推理模型,可以通过食品安全追溯系统各环节的原始数据,预测相应环节的潜在风险,通过环节与环节的潜在风险,实现任意食品生产环节潜在风险与食品安全风险的推理与诊断,当食品安全风险推理结果达到某阈值时,进行预警并对影响环节与原因进行逆推理,达到控制或避免风险的出现。经数据检验表明:模型的正确推理识别率达到93%。
Food Safety has become a world-wide concerned problem. The ideal mode of food quality control is the 'from farm to fork' whole process quality control. Food Safety Traceability System provided the 'from farm to fork' traceability model and established a food safety information database. If we apply the Food Safety Traceability System to food quality control, we can control effectively and recall the trouble as soon as we find it,so as to protect the legitimate rights and interests of consumers from the source. But this mode is still a risk control after we find the trouble,it can not warn and control the risk of food safety.
In this paper, we select the factors which affect the risk of food safety and assess which value by extracting relational sampling, testing and monitoring data from the Food Safety Traceability System. In order to achieve the prediction warning and control of the risk of the Food Safety, we establish knowledge representation and reasoning model based on Bayesian network.
The sectors involved in food chain of food ingredients, processing, packing, storage, transportation, distribution, consumption and so on which will impact on the final food safety.
We select the security situation of all sectors as the node variable, and the variable’s security situation depends on the likelihood of harm and severity of harmful consequences. In order to assess values of variables and obtain the relevant samples, we define potential risk of food safety by likelihood of harm and severity of harmful consequences. After establisted the structure of Bayesian network and got samles for study, we can according conditional probability distribution’s priori information of each node and sample information to update learning network parameters applying Bayesian estimation method.Under the influence of the potential risk level involved in food safety’s all sectors, we achieve the reasoning of risk knowledge which contains cause to reason, cause to cause and result to cause. Further, aiming at reduce the calculation’s complexity and time-consuming of the Bayesian network knowledge reasoning, we research on the simplified knowledge reasoning of Bayesian network. Under the help of Matlab, we demonstrate the implementation process of more efficient simplified knowledge reasoning.
We can predict relevant sector’s potential risk through origin data of this sectors in Food Safety Traceability System. Through the potential risk between sector and sector, the model establist in this paper can reason and diagnose any potential risk in food production process and any risk of food safety. Once the result of food safety risk reasoning reaching the threshold, the model will be warning and doing inverse reference to the involved sectors and the cause for controlling or avoiding the risk. The test depending on data show that the correct rate of model’s reasoning reach 93%.
本文利用食品安全追溯系统各管理过程的抽样、检验以及有关的监控数据,提取影响食品安全风险因素,获取因素变化取值,建立基于贝叶斯网络的知识表示与知识推理模型,以达到食品安全风险的预测、预警与控制。由于食品链中涉及的食品原料、加工、包装、储藏、运输、销售、消费等环节都会对最终食品的安全性造成影响,本文选取各环节的安全状况作为节点变量,变量的安全状况依赖于不同的危害后果严重程度与可能性。通过由危害后果严重程度与可能性定义的食品安全潜在风险,赋予各环节随机变量的取值,获取相关样本,在此基础上,建立贝叶斯网络结构,并依据网络中各节点条件概率分布的先验信息及获取的样本信息,应用贝叶斯估计方法进行网络参数的估计与更新学习,在食品安全各环节潜在风险程度的影响下,实现原因推理原因、原因推理结果、结果诊断原因的风险知识推理。进一步地,对已建立的食品安全风险推理的贝叶斯网络,针对贝叶斯网络知识推理的计算过程复杂与耗时问题,进行简化知识推理研究,在软件Matlab辅助推理下,论证了具有更高效率的简化知识推理的实现过程。
本文建立的基于贝叶斯网络食品安全风险知识推理模型,可以通过食品安全追溯系统各环节的原始数据,预测相应环节的潜在风险,通过环节与环节的潜在风险,实现任意食品生产环节潜在风险与食品安全风险的推理与诊断,当食品安全风险推理结果达到某阈值时,进行预警并对影响环节与原因进行逆推理,达到控制或避免风险的出现。经数据检验表明:模型的正确推理识别率达到93%。
Food Safety has become a world-wide concerned problem. The ideal mode of food quality control is the 'from farm to fork' whole process quality control. Food Safety Traceability System provided the 'from farm to fork' traceability model and established a food safety information database. If we apply the Food Safety Traceability System to food quality control, we can control effectively and recall the trouble as soon as we find it,so as to protect the legitimate rights and interests of consumers from the source. But this mode is still a risk control after we find the trouble,it can not warn and control the risk of food safety.
In this paper, we select the factors which affect the risk of food safety and assess which value by extracting relational sampling, testing and monitoring data from the Food Safety Traceability System. In order to achieve the prediction warning and control of the risk of the Food Safety, we establish knowledge representation and reasoning model based on Bayesian network.
The sectors involved in food chain of food ingredients, processing, packing, storage, transportation, distribution, consumption and so on which will impact on the final food safety.
We select the security situation of all sectors as the node variable, and the variable’s security situation depends on the likelihood of harm and severity of harmful consequences. In order to assess values of variables and obtain the relevant samples, we define potential risk of food safety by likelihood of harm and severity of harmful consequences. After establisted the structure of Bayesian network and got samles for study, we can according conditional probability distribution’s priori information of each node and sample information to update learning network parameters applying Bayesian estimation method.Under the influence of the potential risk level involved in food safety’s all sectors, we achieve the reasoning of risk knowledge which contains cause to reason, cause to cause and result to cause. Further, aiming at reduce the calculation’s complexity and time-consuming of the Bayesian network knowledge reasoning, we research on the simplified knowledge reasoning of Bayesian network. Under the help of Matlab, we demonstrate the implementation process of more efficient simplified knowledge reasoning.
We can predict relevant sector’s potential risk through origin data of this sectors in Food Safety Traceability System. Through the potential risk between sector and sector, the model establist in this paper can reason and diagnose any potential risk in food production process and any risk of food safety. Once the result of food safety risk reasoning reaching the threshold, the model will be warning and doing inverse reference to the involved sectors and the cause for controlling or avoiding the risk. The test depending on data show that the correct rate of model’s reasoning reach 93%.
引文
[1]朱明,王林祥,邓立.食品安全与质量控制[M].北京:化学工业出版社, 2008: 10-11
[2]国家食品安全管理技术标委会食品追溯技术分会.海量数据处理[EB/OL]. http://www.safefood.gov.cn/Scientific/n1328.html/, 2008-05-21
[3]王硕.食品安全溯源与预警[EB/OL]. http://wenku.baidu.com/view/a92b0b1d59eef8c75fbf b34c.html/, 2008-05
[4] Fayyad U.M..Data Mining and Knowledge Discovery: Making Sense out of Data. IEEE Expert 1996, 11(5): 20-25
[5]史忠植.知识发现[M].北京:清华大学出版社, 2002:Ι, 2-17
[6]张贵金,徐卫亚.不确定性知识表示及其度量方法[J].计算机工程与应用, 2003: 80-83
[7] Cristina C., Abigail G.., Kurt V., Using Bayesian Networks to Manage Uncertainty in Student Modeling[J]. User Modeling And User-Adapted Interaction, 2002, 12: 371-417
[8] Yoon J.P., Kerschberg L.. A Framework for Knowledge Discovery and Evolution in Databases[J]. In IEEE Transaction on Knowledge and Data Engineering, 1993, 5(10): 979-984
[9] Faydd U.M., Piatesky-Shapiro G., Smyth P.. From Data Mining to Knowledge Discovery in Databases[J]. AI.Magazine. 1996. 17: 37-54
[10]张连文,郭海鹏.贝叶斯网引论[M].北京:科学出版社, 2006: 9, 65-98
[11]孙鹏程,陈吉宁.基于贝叶斯网络的河流突发性水质污染事故风险评估[J],环境科学, 2009, 30: 47-51
[12]沈静.基于贝叶斯网络模型的我国商业银行操作风险管理研究[D].哈尔滨:哈尔滨工业大学, 2006
[13] Cristina C., Abigail G.., Kurt V., Using Bayesian Networks to Manage Uncertainty in Student Modeling[J]. User Modeling And User-Adapted Interaction, 2002, 12: 371-417
[14] Koski T., Noble J.. Graphical models and probabilistic reasoning [EB/OL]. http:// media. wiley.com/product_data/excerpt/42/04707430/0470743042.pdf, 2009
[15] Heckerman D.. A Bayesian Approach to Causal Discovery[R]. Technical Report MSR-TR-97-05, Microsoft Research, 1997
[16] Cooper G. F., Herskovits E.. A Bayesian Method for the Induction of Probabilistic Networks from Data[J]. Machine Leaming,1992, 9(4): 309-347
[17]朱惠明,韩玉启.贝叶斯多元统计推断理论[M].北京:科学出版社, 2006: 2-3
[18] Berger J.O..Bayesian Analysis:A look at Today and Thoughts of Tomorrow[J]. Journal of the American Statistical Association, 2000, 95 (452): 1269-1274
[19]黄友平.贝叶斯网络研究[D].沈阳:中国科学院研究生院(计算技术研究所), 2005
[20]冀俊忠,刘椿年,沙志强.贝叶斯网模型的学习、推理和应用[J].计算机工程与应用, 2003, 7(5): 24-27
[21]梁循.数据挖掘算法与应用[M].北京:北京大学出版社, 2006: 164-165
[22] Heckerman D., Geiger D., Chickering D.M..Learning Bayesian networks:The combination of knowledge and statistical data[J]. Machine Learning, 1995, 20: 197-243
[23]胡玉胜,涂序彦,崔晓瑜等.基于贝叶斯网的不确定性知识推理方法[J].计算机集成制造系统—CIMS. 2001, 7(12): 65-68
[24] Cooper G..F.. The Computational Complexity of Probabilistic Inference Using Bayesian Belief Networks. Artificial Intelligence. 1990. 42(2-3): 395-405
[25] Heckerman D.. Bayesian networks for data mining[J]. Data Mining and Knowledge Discovery, 1997, 1(1): 79-119
[26] WANG wei, CHEN En-hong, WANG Xu-fa. Knowledge Discovery Based on Bayesian Approach[J]. Journal of China University of Science and Technology 2000, 30(4): 467- 472
[27]张少中.基于贝叶斯网络的知识发现与决策应用研究[D].大连:大连理工大学, 2003
[28]赵培山.食品安全风险评估的现状及发展趋势[EB/OL]. http://www. china value.net/ Article/Archive/2010/1/22/190173.html/, 2010-01-22
[29]陈君石.食品安全风险评估概述[J].中国食品卫生杂志, 2011, 23(1): 4-7
[30]朱坚,张晓岚,张东平等.食品安全与控制导论[M].北京:化学工业出版社, 2009: 102-117, 164, 217
[31] FAO. Food safety risk analysis - A guide for national food safety authorities[R]. Rome: FAO Food and Nutrition, 2006
[32] Risk Management and Food Safety. FAO food and nutrition paper Number 65[R]. Rome: Joint FAO /WHO Consultation, 1997
[33]张竹青,靳东明.应用HACCP原理控制食品安全风险[J].中国质量技术监督2011, 1: 56-57
[34]姜南,张欣,贺国铭等.危害分析和关键控制点(HACCP)及在食品生产中的应用[M].北京:化学工业出版社, 2003: 175-200
[35]百度百科.生产过程质量检验[EB/OL]. http://baike.baidu.com/ view/4207719. html#sub4207719, 2011-04-15
[36]卜庆婧.食品安全追溯的现状[J].食品安全导刊, 2010, 3: 76-78
[37]黄东平.基于模糊贝叶斯网络的食品安全控制知识推理模型的研究[D].广州:华南理工大学, 2010
[38]百度百科.风险[EB/OL] . http://baike.baidu.com/view/156901.htm#sub156901, 2011-04-11
[39]李卫红,杨练根.质量统计技术[M].北京:中国计量出版社, 2006: 22-23, 235-309
[40]茆诗松,程依明,濮晓龙.概率论与数理统计教程[M].北京:高等教育出版社, 2004: 330-339
[41]周金萍,王冉,吴斌. MATLAB6实践与提高[M].北京:中国电力出版社, 2001: 1-30
[42] MATLAB. BNT软件包[EB/OL]. http://www.ai.nit. edu/murphyk/Software/ BNT/ bnt. Html, 2010-04-20
[43] MATLAB. How to use the Bayes Net Toolbox [EB/OL]. http://down.cenet.org.cn/ view.asp?id=85140, 2010-04-20
[44]蒋望东,林士敏.基于贝叶斯网络工具箱的贝叶斯学习和推理[J].信息技术, 2007, (2): 5-8
[2]国家食品安全管理技术标委会食品追溯技术分会.海量数据处理[EB/OL]. http://www.safefood.gov.cn/Scientific/n1328.html/, 2008-05-21
[3]王硕.食品安全溯源与预警[EB/OL]. http://wenku.baidu.com/view/a92b0b1d59eef8c75fbf b34c.html/, 2008-05
[4] Fayyad U.M..Data Mining and Knowledge Discovery: Making Sense out of Data. IEEE Expert 1996, 11(5): 20-25
[5]史忠植.知识发现[M].北京:清华大学出版社, 2002:Ι, 2-17
[6]张贵金,徐卫亚.不确定性知识表示及其度量方法[J].计算机工程与应用, 2003: 80-83
[7] Cristina C., Abigail G.., Kurt V., Using Bayesian Networks to Manage Uncertainty in Student Modeling[J]. User Modeling And User-Adapted Interaction, 2002, 12: 371-417
[8] Yoon J.P., Kerschberg L.. A Framework for Knowledge Discovery and Evolution in Databases[J]. In IEEE Transaction on Knowledge and Data Engineering, 1993, 5(10): 979-984
[9] Faydd U.M., Piatesky-Shapiro G., Smyth P.. From Data Mining to Knowledge Discovery in Databases[J]. AI.Magazine. 1996. 17: 37-54
[10]张连文,郭海鹏.贝叶斯网引论[M].北京:科学出版社, 2006: 9, 65-98
[11]孙鹏程,陈吉宁.基于贝叶斯网络的河流突发性水质污染事故风险评估[J],环境科学, 2009, 30: 47-51
[12]沈静.基于贝叶斯网络模型的我国商业银行操作风险管理研究[D].哈尔滨:哈尔滨工业大学, 2006
[13] Cristina C., Abigail G.., Kurt V., Using Bayesian Networks to Manage Uncertainty in Student Modeling[J]. User Modeling And User-Adapted Interaction, 2002, 12: 371-417
[14] Koski T., Noble J.. Graphical models and probabilistic reasoning [EB/OL]. http:// media. wiley.com/product_data/excerpt/42/04707430/0470743042.pdf, 2009
[15] Heckerman D.. A Bayesian Approach to Causal Discovery[R]. Technical Report MSR-TR-97-05, Microsoft Research, 1997
[16] Cooper G. F., Herskovits E.. A Bayesian Method for the Induction of Probabilistic Networks from Data[J]. Machine Leaming,1992, 9(4): 309-347
[17]朱惠明,韩玉启.贝叶斯多元统计推断理论[M].北京:科学出版社, 2006: 2-3
[18] Berger J.O..Bayesian Analysis:A look at Today and Thoughts of Tomorrow[J]. Journal of the American Statistical Association, 2000, 95 (452): 1269-1274
[19]黄友平.贝叶斯网络研究[D].沈阳:中国科学院研究生院(计算技术研究所), 2005
[20]冀俊忠,刘椿年,沙志强.贝叶斯网模型的学习、推理和应用[J].计算机工程与应用, 2003, 7(5): 24-27
[21]梁循.数据挖掘算法与应用[M].北京:北京大学出版社, 2006: 164-165
[22] Heckerman D., Geiger D., Chickering D.M..Learning Bayesian networks:The combination of knowledge and statistical data[J]. Machine Learning, 1995, 20: 197-243
[23]胡玉胜,涂序彦,崔晓瑜等.基于贝叶斯网的不确定性知识推理方法[J].计算机集成制造系统—CIMS. 2001, 7(12): 65-68
[24] Cooper G..F.. The Computational Complexity of Probabilistic Inference Using Bayesian Belief Networks. Artificial Intelligence. 1990. 42(2-3): 395-405
[25] Heckerman D.. Bayesian networks for data mining[J]. Data Mining and Knowledge Discovery, 1997, 1(1): 79-119
[26] WANG wei, CHEN En-hong, WANG Xu-fa. Knowledge Discovery Based on Bayesian Approach[J]. Journal of China University of Science and Technology 2000, 30(4): 467- 472
[27]张少中.基于贝叶斯网络的知识发现与决策应用研究[D].大连:大连理工大学, 2003
[28]赵培山.食品安全风险评估的现状及发展趋势[EB/OL]. http://www. china value.net/ Article/Archive/2010/1/22/190173.html/, 2010-01-22
[29]陈君石.食品安全风险评估概述[J].中国食品卫生杂志, 2011, 23(1): 4-7
[30]朱坚,张晓岚,张东平等.食品安全与控制导论[M].北京:化学工业出版社, 2009: 102-117, 164, 217
[31] FAO. Food safety risk analysis - A guide for national food safety authorities[R]. Rome: FAO Food and Nutrition, 2006
[32] Risk Management and Food Safety. FAO food and nutrition paper Number 65[R]. Rome: Joint FAO /WHO Consultation, 1997
[33]张竹青,靳东明.应用HACCP原理控制食品安全风险[J].中国质量技术监督2011, 1: 56-57
[34]姜南,张欣,贺国铭等.危害分析和关键控制点(HACCP)及在食品生产中的应用[M].北京:化学工业出版社, 2003: 175-200
[35]百度百科.生产过程质量检验[EB/OL]. http://baike.baidu.com/ view/4207719. html#sub4207719, 2011-04-15
[36]卜庆婧.食品安全追溯的现状[J].食品安全导刊, 2010, 3: 76-78
[37]黄东平.基于模糊贝叶斯网络的食品安全控制知识推理模型的研究[D].广州:华南理工大学, 2010
[38]百度百科.风险[EB/OL] . http://baike.baidu.com/view/156901.htm#sub156901, 2011-04-11
[39]李卫红,杨练根.质量统计技术[M].北京:中国计量出版社, 2006: 22-23, 235-309
[40]茆诗松,程依明,濮晓龙.概率论与数理统计教程[M].北京:高等教育出版社, 2004: 330-339
[41]周金萍,王冉,吴斌. MATLAB6实践与提高[M].北京:中国电力出版社, 2001: 1-30
[42] MATLAB. BNT软件包[EB/OL]. http://www.ai.nit. edu/murphyk/Software/ BNT/ bnt. Html, 2010-04-20
[43] MATLAB. How to use the Bayes Net Toolbox [EB/OL]. http://down.cenet.org.cn/ view.asp?id=85140, 2010-04-20
[44]蒋望东,林士敏.基于贝叶斯网络工具箱的贝叶斯学习和推理[J].信息技术, 2007, (2): 5-8
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |