基于Bagging集成算法的产品质量等级智能判定研究  

Research on Intelligent Judgment of Product Quality Level Based on Bagging Integrated Algorithm

作　　者：柏雪 李剑锋 BAI Xue;LI Jianfeng(College of Economic and Management,China Jiliang University,Hangzhou,Zhejiang 310018,China)

机构地区：[1]中国计量大学经济与管理学院,浙江杭州310018

出　　处：《工业工程与管理》2022年第4期58-66,共9页Industrial Engineering and Management

基　　金：国家自然科学基金资助面上项目(71972172)。

摘　　要：工业大数据驱动人工智能赋能企业创新发展,正在对制造业生产方式、运行模式、生态体系产生重大而深远的影响。为实现产品质量等级的智能判定,提出一种基于Bagging集成的产品质量等级判定模型。模型将梯度提升决策树引入Bagging集成框架,权衡偏差和方差以减弱合格品与不合格品数量不平衡带来的影响。将该方法应用于某化纤企业涤纶长丝的真实生产中,结果表明模型能够实现长丝质量等级的准确判定,帮助企业提高产品质量检验效率,降低人工成本并有效提升产品的质量管理水平。Industrial big data drives artificial intelligence to empower enterprise innovation and development,which has a major and far-reaching impact on manufacturing production methods,operating modes,and ecosystems.In order to realize the intelligent determination of quality level,a product quality level determination model based on Bagging integration was proposed.Gradient lifting decision tree was introduced into the Bagging integration framework in the model,and the bias and variance were both weighed to reduce the impact of the quantity imbalance of qualified products and unqualified products.The method was applied into the actual production of polyester filament in a chemical fiber enterprise.The results show that the model can accurately determine the quality grade of filament,help the enterprise improve the inspection efficiency of product quality,reduce labor cost and improve the level of product quality management effectually.

关 键 词：涤纶长丝 质量等级判定 Bagging集成 GBDT 最大信息系数 

分 类 号：F273[经济管理—企业管理] TP399[经济管理—国民经济]