连续闭环加工中智能柔性追溯模型构建及系统验证：以小麦粉加工为例

Traceability system is an important measure to ensure the quality and safety of products. The existed traceability model was established in the condition of undivided traceability resource unit and uniform quality feature. Due to the features of fuzzy batches boundaries and crossed quality safety information during the continuous and closed-loop agricultural products processing, the traceability model in this process with intelligent and flexibility characters is bottleneck to establish a comprehensive full traceability system. This project takes the wheat flour processing as the research object, analyzes the processing static structure and the dynamic change of the material, and makes the modeling and simulation of the continuous and closed-loop process. According to the probability distribution function of each batch, the batch fusion coefficients of different trace elements were studied, and the dynamic spectrum is established to define the change characteristics of trace elements in batch fusion; Quality features are introduced into the error propagation network by analyzing the key quality control points, , the multiple control quality coupling model is constructed through the merger of single control point quality predictive value to reveal links between different quality and safety transmission mechanism; The intelligent flexible traceability model with multiple constraints is constructed, and the improved particle swarm optimization model is used to validate the model. Improved wheat flour processing intelligent traceability system based on the developed model will lay a foundation for a "from field to table" full traceability system. In addition, this model will provide an important reference for the construction other similar continuous production and processing, such as dairy production and fruit grading and packaging.

已有追溯模型以单元不可分割和质量均一为前提，而连续闭环农产品加工具有批次边界模糊、质量安全信息交叉的特征，因此开发具有智能柔性特征的模型已成为加工过程追溯的瓶颈。本项目以小麦粉加工为研究对象，分析加工节点静态结构和物料动态变化，对连续闭环加工过程进行建模仿真；结合时间特征下各批次概率分布函数，研究不同追溯单元粒度下的批次融合系数，建立追溯单元动态谱系，明确批次融合下追溯单元变化特征；分析关键质量控制点，将质量特征引入误差传递网络，通过合并单控制点质量预测值，构建多控制点质量耦合模型，揭示不同环节之间质量安全传递机制；构建多重约束下的智能柔性追溯模型，采用改进的粒子群算法优化模型，并基于召回效率进行模型验证评价。利用本模型改进开发的小麦粉加工智能追溯系统将为建立“从麦田到餐桌”的全程追溯体系奠定基础；同时，模型可为乳品生产、果品分级包装等类似加工中追溯模型的构建提供重要参考。

连续闭环农产品加工过程具有原辅料来源多样、批次边界模糊、质量安全信息交叉等特点，已有追溯模型不能有效解决此类问题。本项目以小麦粉加工为研究对象，分析加工节点静态结构和物料动态变化，建立了小麦粉加工仿真模型；分析了先进先出和均匀混合条件下的批次混合差异，发现了追溯断链机制，奠定了追溯提升的理论基础。构建了以精度、宽度和深度为核心的追溯粒度评价模型，分析了小麦粉加工过程的追溯粒度变化特点；引入了时间戳概念，建立了基于有向图的追溯单元动态谱系；提出了一种适用于小麦粉的批次混合度实时计算方法与模拟装置，解决了批次混合中追溯单元难识别的问题。分析了影响小麦粉质量安全的主要风险，构建了以风险发生可能性的高低和风险危害程度的强弱为核心的风险二维矩阵图，通过识别得到了小麦粉加工中的单点风险；构建了小麦粉加工过程质量安全危害关系Petri 网模型，动态估测危害传播路径，解决了批次混合中风险难预测的问题。根据小麦粉加工特点，构建了包装产品的批次清单树结构，关联产品批次和原料批次、辅料批次以及检测节点，构建了智能追溯模型；系统验证结果表明传统方法相比，实现召回批次数量由7个减少到4个、危害因素查找数量由6个减少到5个，提高了追溯和召回效率；开发了小麦粉加工管理与智能追溯系统实现了原料入库信息采集、加工过程信息采集、检化验信息采集、批次清单生成、危害分析、产品追溯等功能。.项目申请发明专利6项、获得发明专利授权3项，获得实用新型专利授权1项，发表学术论文10篇、其中SCI论文6篇、EI论文2篇，获得软件著作权登记两项，制定地方标准两项，获得科技奖励1项。本研究构建的小麦粉加工智能追溯模型能有效对接小麦种植和粮油流通环节，使“从麦田到餐桌”的全程追溯体系建立成为可能，有利于提升质量安全保障水平；同时，本模型的构建将为果品分级包装等类似追溯问题的解决提供重要参考。