摘要
在大规模时序数据分析中,传统数理统计与分析技术耗时较多,精度不高,抗干扰能力不强。针对这些问题,文中基于波动向量分级技术,提出一种病变时序数据快速分析方法。该方法在TSTKS突变点检测算法与滑动窗口理论的基础上,采用多阈值分割技术实现了波动向量的多层分级策略,进而实现了对大规模病变时序数据的状态分析与快速诊断。仿真实验与脑癫痫病变信号分析实验结果表明,文中所提出的新方法速度较快,效率较高,可以为大规模时序数据快速分析提供参考。
In large-scale time series data analysis,traditional mathematical statistics and analysis techniques have problems such as time-consuming,low accuracy,and weak anti-interference ability.In view of the deficiency,based on the wave vector grading technology,this study presents a fast analysis method for the time series data of lesions.Based on the TSTKS mutation point detection algorithm and sliding window theory,this method uses multi-threshold segmentation technology to realize the multi-level classification strategy of fluctuation vectors,and then realizes the state analysis and rapid diagnosis of large-scale lesion time series data.The result of simulation experiments and brain epilepsy lesion signal analysis show that the proposed method has the advantages of faster speed and higher efficiency,and can provide a new method for the rapid analysis and research of large-scale time series data.
作者
孔凡书
齐金鹏
龚汉鑫
朱俊俊
曹一彤
KONG Fanshu;QI Jinpeng;GONG Hanxin;ZHU Junjun;CAO Yitong(College of Information Science & Technology,Donghua University,Shanghai 201620,China)
出处
《电子科技》
2022年第7期1-6,共6页
Electronic Science and Technology
基金
国家自然科学基金(61305081,61104154)
上海市自然科学基金(16ZR1401300,16ZR1401200)。
关键词
时序数据
数据分析
TSTKS算法
突变点检测
滑动窗口理论
波动向量
阈值分割
多层分级
time series data
data analysis
TSTKS algorithm
mutation point detection
sliding window theory
fluctuation vector
threshold segmentation
multi-level classification
