基于FTA-BN的食品安全风险评估与控制模型  

Modeling for Food Safety Risk Assessment and Control Based on FTA-BN

作　　者：李柯[1] 唐小艳[1] 

机构地区：[1]武汉理工大学管理学院,湖北武汉430070

出　　处：《武汉理工大学学报（信息与管理工程版）》2017年第3期243-249,共7页Journal of Wuhan University of Technology：Information & Management Engineering

基　　金：国家社会科学基金重大项目(15&ZDB168)

摘　　要：为探究食品链中的薄弱环节,控制食品安全风险,提出基于FTA-BN模型的风险分析方法。首先,构建风险因素、风险事件的故障树,并将故障树映射为贝叶斯网络;然后利用贝叶斯正向推理预测食品安全风险发生概率,并结合反向推理对风险事件进行故障排查;最后利用互信息进行敏感性分析,辨识薄弱环节;提出以环节控制为核心,结合最大后验假设问题和最大后验概率的方法逆向识别诱发风险事件的关键线路。以2009—2014年猪肉食品安全为例进行实证研究,结果表明,猪肉食品安全风险概率为2.41%,其中加工环节是引发食品安全问题的高危环节,食品安全监测能力不足、使用不合格原料、销售不合格产品、滥用添加剂等是导致风险事件发生的关键因素。In order to explore weak link in food chain, and control safety risk of food, a new risk safety analysis method is worked out that based on fault tree analysis and Bayesian network. Firstly, a model was built for relationship between risk factors and risk events based on Fault tree, and FT was mapped into BN according to the mapping algorithm. Then, the probability of food safety risk has been calculated through deductive reasoning of BN and troubleshooting of food safety risk was analyzed with abductive reasoning. Finally, a sensitivity analysis was used to find sensitive risk factors based on MI index, link control was taken as core,and combine maximum a posteriori estimation and maximum posterior probability to retroactively identify the critical line of risk events. Meat food safety from 2009 to 2014 in China was taken as example, the results show that probability of food safety risk is 2. 41% , manufacture processing is the uppermost step of food safety problems and poor regulation, using unqualified raw materials, sales of unqualified products, and abuse of additives were key factors.

关 键 词：食品安全 故障树分析法 贝叶斯网络 互信息 关键路径识别 风险评估与控制 

分 类 号：C93[经济管理—管理学]