Abstract: With the advent of the big data era,the traditional data mining technology can no longer meet the requirements of massive data processing under the background of intelligent manufacturing.In view of the lack of computing capacity and low mining efficiency in massive data mining,a step-by-step clustering method was introduced,and then the classical K-means clustering algorithm was improved on the MapReduce framework of Hadoop platform under a cloud computing environment.Using this framework,a new algorithm for solving the massive data mining tasks was generated.The performance of T.K-means algorithm was examined in mining the virtual data of milling Ti-6Al-4V(TC4) titanium alloy.Then the relationship between operating parameters and performance indicators was produced and used to guide optimization operation of the machine tool.Mining results showed that the proposed algorithm could be effectively applied in the determination of the optimization targeted values to achieve the purpose of optimizing the surface roughness of workpiece.The obtained optimization targeted values were representative of optimum operation conditions of the surface roughness of TC4 titanium alloy milling by machine tool.

Key words: data mining, step by step clustering, T.K-means algorithm, MapReduce framework, milling parameters optimization

摘要： 随着大数据时代的来临,传统数据挖掘技术已不能满足智能制造背景下海量数据处理的要求。针对海量数据在挖掘过程中出现的计算能力不足、挖掘效率低下的现状,提出在云计算环境下,以分步聚类理念为基础,对经典K-means聚类算法进行改进,并将改进的算法与Hadoop平台的MapReduce计算架构相结合,实现算法计算并行化,从而形成能够应对海量数据挖掘任务的新算法T.K-means。为了验证新算法的实用性能,以虚拟铣削Ti-6Al-4V(TC4)钛合金加工运行数据为挖掘对象,利用T.K-means算法挖掘加工工艺参数与工件表面粗糙度之间的关系,得到可调控工艺参数的优化值以指导实际加工。挖掘结果表明,T.K-means算法可用于TC4钛合金铣削工艺参数优化目标值的确定,其所挖掘出的工艺参数能够反应机床铣削加工TC4钛合金表面粗糙度的最佳运行状态。

关键词: 数据挖掘, 分步聚类, T.K-means算法, MapReduce架构, 铣削参数优化