基于微服务和GRU的卷烟工艺质量预警系统设计  

Design of the Cigarette Process Quality Monitoring System Based on Micro services and GRU

作　　者：杨俊杰 徐志强 张军 万宇超 欧阳敏 范安平 许冰洋 YANG Junjie;XU Zhiqiang;ZHANG Jun;WAN Yuchao;OUYANG Min;FAN Anping;XU Bingyang(Technology Center,Jiangxi China Tobacco Industry Co.,Ltd.,Nanchang 330000,China;Guangfeng Cigarette Factory,Jiangxi China Tobacco Industry Co.,Ltd.,Shangrao 334600,China)

机构地区：[1]江西中烟工业有限责任公司技术中心,南昌330000 [2]江西中烟工业有限责任公司广丰卷烟厂,江西上饶334600

出　　处：《计算机测量与控制》2024年第12期153-158,177,共7页Computer Measurement &Control

基　　金：江西中烟工业有限责任公司科技项目(2020-09)。

摘　　要：针对传统的卷烟工艺质量预警系统自动化程度低,提出一种基于微服务和GRU的卷烟工艺质量预警系统;该系统采用Spring Cloud微服务架构,分为数据资源层、微服务组件、业务层和网关层,可以满足不同场景的集成需求;搭建了卷烟生产过程监控数据库,结合注意力机制改进GRU建立卷烟加工工艺质量预警模型,并与传统方法逻辑回归、支持向量机、GRU进行对比;经实验测试证明,该预警模型的平均绝对误差和均方根误差分别为0.381和0.570,模型预测效果和系统性能均优于其它方法,更加有效地实现了工艺质量预警,提高了卷烟加工过程自动化程度。In response to the low automation level in traditional cigarette manufacturing process quality warning systems,this paper proposes a cigarette process quality warning system based on microservices and Gated Recurrent Unit(GRU).The system adopts the Spring Cloud microservice architecture,which is composed of the data resource layer,microservice components,business layer,and gateway layer,meeting the integration requirements in various scenarios.A monitoring database for cigarette production processes is presented,and an attention mechanism is integrated to improve the GRU and construct the cigarette process quality warning model,which makes a comparative analysis with traditional methods such as logistic regression,support vector machine,and the GRU.Experimental results demonstrate that the proposed warning model achieves an average absolute error of 0.381 and a root mean square error of 0.570.The predictive performance and efficiency of the method are superior to those of other ones,effectively realizing process quality warning and enhancing the automation level of cigarette processing.

关 键 词：微服务架构 质量预警 Spring Cloud 注意力机制 门控循环单元 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]