基于属性分类的装备保障数据特征挖掘法  

Feature mining method of equipment support data based on attribute classification

作　　者：杨华[1] YANG Hua(School of Electronic Information Engineering,Nanchong Vocational and Technical College,Nanchong 637000,China)

机构地区：[1]南充职业技术学院电子信息工程系,四川南充637000

出　　处：《兵器材料科学与工程》2020年第6期124-128,共5页Ordnance Material Science and Engineering

基　　金：教育部职业院校信息化教学指导委员会信息化教学研究课题(2018LXB0084);南充市应用技术研究与开发专项项目(18YFZJ0021)。

摘　　要：为提高装备保障数据特征挖掘准确率和召回率,降低其均方根误差及平均绝对百分比误差,提出属性分类法。选取装备保障数据综合管理与分析系统,采集特征数据,集成处理,对集成后数据进行清理、转换及规约预处理。用属性分类法,遵循类内距离平方和最小、类间距离平方和最大的"高类聚、低耦合"原则,获取最优分类即为装备保障数据特征挖掘结果。结果表明:采用本文方法挖掘装备保障数据特征准确率、召回率均高于99%,均方根误差、平均绝对百分比误差均低于0.2。In order to improve the accuracy and recall rate of equipment support data feature mining and reduce its root mean square error and average absolute percentage error,an attribute classification method was proposed.The method included selecting the equipment support data comprehensive management and analysis system,collecting characteristic data,integrating processing,and implementing data cleaning,data conversion and data protocol preprocessing on the integrated data.Using the attribute classification method,following the principle of"high clustering and low coupling"with the smallest sum of squared distances within classes and the largest sum of squared distances between classes,the optimal classification was the result of equipment support data feature mining.The results show that the accuracy and recall rate of mining equipment support data features using this method are higher than 99%,and the root mean square error and average absolute percentage error are both lower than 0.2.

关 键 词：属性分类 装备 保障数据 特征 挖掘方法 

分 类 号：TJ81[兵器科学与技术—武器系统与运用工程]