一种改进型C4.5算法在SMT焊接质量中的应用研究  

Application of Improved C4.5 Algorithm in SMT Welding Quality

作　　者：唐露新[1] 张强武[1] 

机构地区：[1]广东工业大学信息工程学院,广州510006

出　　处：《舰船电子工程》2014年第7期184-187,202,共5页Ship Electronic Engineering

基　　金：广东省教育部产学研合作专项资金(编号:2012B091100056)资助

摘　　要：C4.5算法是目前应用最为广泛的一种决策树算法,针对其分支策略和属性选取的不足,提出与分裂属性和类属性相关的平衡因子,以修正属性的分裂信息,进而选取更有意义的属性作为分裂节点,协调各属性的信息增益率。采用经典天气实例与原C4.5算法对比,改进的决策树分类更为合理、准确。改进C4.5算法用来预测不同参数下SMT焊接质量以选择最佳参数,可提高SMT生产效率。C4.5algorithm is currently the most widely used decision tree algorithm.In view of the decision tree C4.5algorithm branching strategy and attribute selection,the balance factors associated with division and class attribute is put forward.The factors can be applied to modify the division of property information,to choose a more meaningful attribute as the split node and to coordinate information gain ratio of various properties.The classic example weather can be compared to the improved C4.5algorithm and the original C4.5algorithm,the improved decision tree classification is more reasonable and accurate.The improved C4.5algorithm is used to predict the SMT welding quality under different parameters to choose the best parameters,can improve the efficiency of the SMT production line.

关 键 词：信息增益率 平衡因子 决策树 C4 5算法 焊接质量 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]