一种基于K-means聚类的软件测试数据异常检测方法  

An Anomaly Detection Method of Software Test Data Based on K-means Clustering

作　　者：陈自力[1] CHEN Zili(Fujian Chuanzheng Communications College,Fuzhou 350007,China)

机构地区：[1]福建船政交通职业学院,福建福州350007

出　　处：《太原师范学院学报（自然科学版）》2021年第3期38-42,共5页Journal of Taiyuan Normal University:Natural Science Edition

基　　金：福建省社会科学规划项目“基于案例驱动的事故灾害应急方案生成方法研究”(FJ2019B079).

摘　　要：为了提高可移植多维控制软件测试数据的异常检测能力,提出一种基于K-means聚类的软件测试数据异常检测方法.构建可移植多维控制软件测试的异常数据分布结构模型,采用模糊性语义特征重建方法,识别可移植多维控制软件模糊参数,提取关联信息特征量,根据关联特征演化分布,采用联合组合特征分析方法,实现可移植多维控制软件测试异常数据的模糊聚类中心检测,进行异常特征分布融合,完成联合多维特征检测,采用K-means聚类方法进行可移植多维控制软件中的有效数据组合控制,实现可移植多维控制软件测试数据的异常特征提取和检测.测试结果表明,采用该方法进行可移植多维控制软件测试异常数据检测的精度较高,检测性能较好.In order to improve the ability of anomaly detection for portable multidimensional control software test data,a method of anomaly detection for software test data based on K-means clustering is proposed.The structural model of abnormal data distribution for portable multidimensional control software testing is constructed.The fuzzy semantic feature reconstruction method is used to identify the fuzzy parameters of portable multidimensional control software and extract the feature quantity of association information.According to the evolution and distribution of association features,the combined feature analysis method is used to realize the fuzzy clustering center detection of abnormal data for portable multidimensional control software testing,the K-means clustering method is used for the effective data combination control in the portable multidimensional control software to realize the abnormal feature extraction and detection of the test data of the portable multidimensional control software.The test results show that the method has high accuracy and good performance in the detection of abnormal data in the portable multidimensional control software test.

关 键 词：K-MEANS聚类 软件 测试数据 异常 检测 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]