关联规则挖掘算法及其在冷轧生产中的应用  

Association rules mining algorithm for cold-rolling processes

作　　者：王金城[1] 王晓琳[2] 庞古风 

机构地区：[1]大连理工大学自动化系,大连116024 [2]沈阳化工学院数理系,沈阳110142

出　　处：《清华大学学报（自然科学版）》2007年第z2期1761-1765,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(60474058)

摘　　要：针对Apriori算法在实际应用中无法发现关联规则变化趋势的问题,该文根据增量挖掘算法的优点对Apriori算法进行了改进。改进的Apriori算法能够在原算法的基础上,通过关联规则统计量的变化确定强规则与候选规则之间的转换,从而进一步发现关联规则的变化趋势,提高了依靠Apriori算法得到的关联规则对决策分析支持的可靠性。将改进算法应用于冷轧生产过程预测中,试验结果表明,改进算法相对于传统的Apriori算法对产量预测的精度提高了30%。An incremental mining algorithm was developed to overcome limitations in the Apriori algorithm which can not find trends in association rules.The improved Apriori algorithm is able to identify information inside the rules through the transformation between strong rules and alternate rules by variations in the association rule statistical data.The decision-making reliability is enhanced by the association rules obtained from the improved algorithm.The algorithm was used to forecast the output of a cold-rolling process with test results showing that the prediction precision of the algorithm was 30% better than that of the traditional Apriori algorithm.

关 键 词：关联规则 APRIORI算法 增量挖掘 冷轧生产过程 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]