Based on the deformation characteristic of regular polygonal box stamped parts and the superfluous triangle material wrinkle model,the criterion of regular polygonal box stamped parts without wrinkle was deduced and u...Based on the deformation characteristic of regular polygonal box stamped parts and the superfluous triangle material wrinkle model,the criterion of regular polygonal box stamped parts without wrinkle was deduced and used to predict and control the wrinkle limit.According to the fracture model,the criterion of regular polygonal box stamped parts without fracture was deduced and used to predict and control the fracture limit.Combining the criterion for stamping without wrinkle with that without fracture,the stamping criterion of regular polygonal box stamped parts was obtained to predict and control the stamping limit.Taken the stainless steel0Cr18Ni9(SUS304)sheet and the square box stamped part as examples,the limit diagram was given to predict and control the wrinkle,fracture and stamping limits.It is suitable for the deep drawing without flange,the deep drawing and stretching combined forming with flange and the rigid punch stretching of plane blank.The limit deep-drawing coefficient and the minimum deep-drawing coefficient can be determined,and the appropriate BHF(blank holder force)and the deep-drawing force can be chosen.These provide a reference for the technology planning,the die and mold design and the equipment determination,and a new criterion evaluating sheet stamping formability,which predicts and controls the stamping process,can be applied to the deep drawing under constant or variable BHF conditions.展开更多
针对数据集标签缺失且类别分布极不平衡的信用卡欺诈检测问题,提出一种基于动态集成选择算法的信用卡审批异常检测模型DES-HBOS(Dynamic Ensemble Selection based on Histogram-based Outlier Score)。首先,利用无监督异常检测算法构...针对数据集标签缺失且类别分布极不平衡的信用卡欺诈检测问题,提出一种基于动态集成选择算法的信用卡审批异常检测模型DES-HBOS(Dynamic Ensemble Selection based on Histogram-based Outlier Score)。首先,利用无监督异常检测算法构造训练集客户的伪标签;然后,确定待测客户能力区域,根据Pearson相关系数评估分类器性能;最后,选择一组较优的分类器对待测客户进行集成。在真实信用卡客户数据集上的实验表明,与其他6种经典异常检测模型相比,DES-HBOS的Recall更高,能将更多欺诈客户识别出来。在4个不平衡数据集上进行对比实验,实验结果表明与HBOS相比,DES-HBOS检测异常能力更强。展开更多
文摘Based on the deformation characteristic of regular polygonal box stamped parts and the superfluous triangle material wrinkle model,the criterion of regular polygonal box stamped parts without wrinkle was deduced and used to predict and control the wrinkle limit.According to the fracture model,the criterion of regular polygonal box stamped parts without fracture was deduced and used to predict and control the fracture limit.Combining the criterion for stamping without wrinkle with that without fracture,the stamping criterion of regular polygonal box stamped parts was obtained to predict and control the stamping limit.Taken the stainless steel0Cr18Ni9(SUS304)sheet and the square box stamped part as examples,the limit diagram was given to predict and control the wrinkle,fracture and stamping limits.It is suitable for the deep drawing without flange,the deep drawing and stretching combined forming with flange and the rigid punch stretching of plane blank.The limit deep-drawing coefficient and the minimum deep-drawing coefficient can be determined,and the appropriate BHF(blank holder force)and the deep-drawing force can be chosen.These provide a reference for the technology planning,the die and mold design and the equipment determination,and a new criterion evaluating sheet stamping formability,which predicts and controls the stamping process,can be applied to the deep drawing under constant or variable BHF conditions.
文摘针对数据集标签缺失且类别分布极不平衡的信用卡欺诈检测问题,提出一种基于动态集成选择算法的信用卡审批异常检测模型DES-HBOS(Dynamic Ensemble Selection based on Histogram-based Outlier Score)。首先,利用无监督异常检测算法构造训练集客户的伪标签;然后,确定待测客户能力区域,根据Pearson相关系数评估分类器性能;最后,选择一组较优的分类器对待测客户进行集成。在真实信用卡客户数据集上的实验表明,与其他6种经典异常检测模型相比,DES-HBOS的Recall更高,能将更多欺诈客户识别出来。在4个不平衡数据集上进行对比实验,实验结果表明与HBOS相比,DES-HBOS检测异常能力更强。
