摘要
针对离散车间生产能耗数据多源异构,干扰因素间关联关系不易分析的问题,根据离散制造能耗机理分析结果构建多源能耗数据挖掘模型(Multi-source Energy-consumption Data-mining Model,MEDM),提出一种面向离散车间的频繁模式增长(Frequent Pattern Growth for Discrete Workshop,FP-Growth-DW)算法。首先,应用关联视图、分箱处理和多源异构统一编码方法,对离散车间多源异构海量数据进行预处理;然后,采用分区并行策略和3种剪枝技术优化候选项集以构建频繁模式树(FP-Tree),并提取强关联规则合并存储到规则库中。最后实例验证表明,该算法在离散车间多源异构数据挖掘分析上具有可行性和有效性,算法效率提升了68.8%。
Aiming at the problems of multi-source heterogeneity of discrete workshop production energy consumption data,and the correlation relationship between interference factors is not easy to analyze,based on the analysis result of discrete manufacturing energy consumption mechanism,a Multi-source Energy-consumption Data-mining Model(MEDM)was constructed,and an algorithm of Frequent Pattern Growth for Discrete Workshop(FP-Growth-DW)was proposed.Firstly,the associative view,binning processing and multi-source heterogeneous unified coding method were applied to preprocess multi-source heterogeneous massive data in discrete workshops.Secondly,the partition parallel strategy and three pruning techniques were used to optimize the candidate itemsets to build a Frequent Pattern Tree(FP-Tree),and the strong association rules were extracted and merged into the rule base file.Finally,the example verification showed that the algorithm is feasible and efficient in the multi-source heterogeneous data mining analysis of discrete workshop,and the algorithm efficiency is increased by 68.8%.
作者
崔志鹏
吉卫喜
曹桢淼
陈琛
周姝含
CUI Zhipeng;JI Weixi;CAO Zhenmiao;CHEN Chen;ZHOU Shuhan(School of Mechanical Engineering,Jiangnan University,Wuxi 214122,China;Jiangsu Provincial Key Laboratory of Food Manufacturing Equipment,Wuxi 214122,China;AVIC Leihua Electronic Technology Institute,Wuxi 214082,China)
出处
《现代制造工程》
CSCD
北大核心
2023年第3期45-54,44,共11页
Modern Manufacturing Engineering
基金
山东省重大科技创新工程基金项目(2019JZZY020111)。
关键词
离散制造
统一编码
剪枝优化
数据挖掘
discrete manufacturing
unified coding
pruning optimization
data mining