摘要
现有的异常数据检测方法存在限制条件过多,导致检测准确度低,检测所需时间长等问题。为解决现有方法存在的问题,提出一种基于状态转移矩阵的便携式医疗设备通信异常数据检测方法。使用OPTICS算法进行原始数据聚类,将数据集划分为不同密度的数据簇;利用信息熵提取数据离群特征,计算不同数据簇的数据离群度;通过聚类数据簇与数据离群度数值更新状态转移矩阵,并结合状态与观测值之间的统计关系来判断并筛选异常数据。仿真证明,与现有的异常数据检测方法相比,所提方法检测准确度更高,检测所需时间更短。
The current method for detecting abnormal data has too many constraints,which leads to low detection accuracy and long time consuming detection.Therefore,an abnormal data detection method in communication of portable medical device based on state transfer matrix was proposed.This method used OPTICS algorithm to cluster initial data and divided the data set into data clusters with different densities.Then,the method used information en- tropy to extract data outlier characteristic and calculate the degree of outlier of different data clusters.Moreover,our method updated state transition matrix through clustering data cluster and value of data outlier degree.Based on sta- tistical relationship between state and observed value,we judged and filtered abnormal data.Simulation shows that the proposed abnormal data detection method has higher detection accuracy and the shorter detection time than the current method.
作者
丁可
DING Ke(Wuxi 9th People's Hospital,Wuxi Handsurgery Hospital,Jiangsu Wuxi 214000,China)
出处
《计算机仿真》
北大核心
2018年第12期313-316,共4页
Computer Simulation
关键词
便携式医疗设备
设备通信异常
异常数据检测
Portable medical device
Equipment communication abnormality
Abnormal data detection
