实时数据流聚类等方法在钢铁质量监控中的应用  

THE APPLICATION OF REAL-TIME DATA STREAM CLUSTERING METHOD IN STEEL QUALITY MONITORING

作　　者：黄娇龙 张晓龙[1,2] 边小勇 余海[1,2] 

机构地区：[1]武汉科技大学计算机科学与技术学院,湖北武汉430065 [2]智能信息处理与实时工业系统湖北省重点实验室,湖北武汉430065

出　　处：《计算机应用与软件》2015年第5期297-301,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(60975031;61273225);国家十二五支撑计划项目(2012BAC22B01);湖北省自然科学基金重点项目(2009CDA034);武汉市学科带头人计划项目(201150530152)

摘　　要：实时数据流独特的特点,给传统数据的处理方法带来了很大的挑战,利用聚类方法挖掘其中的有用知识则显得非常有优势。针对某钢铁企业在生产过程中出现的信息不通、产品质量无法跟踪的问题,设计开发一个钢铁产品生产过程实时监控系统。同时结合在仿真数据集上具有良好表现的基于密度维度树的增量式网格聚类算法(IGDDT),对产品工艺实时数据流进行分析。实现了企业对产品质量的实时监控,以及对关键工艺的改进,优化了钢铁产品生产过程和产品管理,有利于提高钢铁产品质量。The characteristics of real-time data stream bring great challenges to traditional data processing methods, but using clustering method to mind useful knowledge in the stream seems to have a large advantage. In this paper, aiming at the problems in a certain steel enterprise that the information is blocked in production process and the product quality cannot be tracked, we design and develop a real-time monitoring system for production process of steel products. At the same time, in combination with IGDDT （ incremental grid clustering algorithm based on density-dimension tree） clustering algorithm, which has good performance in simulation datasets, we carry out the analysis on real-time data streams of product technology. The real-time monitoring on products quality and the improvement on key technologies are realised by the enterprise, which optimise the production process and products management of the steel products and benefit the improvement in steel products.

关 键 词：实时数据流 实时数据库 聚类算法 实时监控 质量分析 

分 类 号：TP316.2[自动化与计算机技术—计算机软件与理论]