时间序列相似性与预测算法研究及其应用
发布时间：2018-09-08 09:22
【摘要】：时间序列分析广泛应用于各个领域,相似性研究是时间序列分析的基础,预测是时间序列分析中的重要问题。论文针对医学中的时间序列数据和轨道交通客流时间序列数据的相似性与预测问题,对四个问题进行了研究和讨论：脑卒中(脑中风)症状证候布尔时间序列的动态分析及预后判断；脑卒中脑电数据的特征分析及预后判断；脑卒中病灶位置及其脑电信号的特征分析；轨道交通客流数据的相似性及预测分析。论文的主要工作及创新点总结如下： 一、针对脑卒中症状及证候布尔时间序列数据进行了分析,从中医的角度讨论了脑中风病症状及证候的布尔时间序列监测数据的特点,提出了面向布尔时间序列数据的关联规则挖掘算法,利用症状及证候数据的动态变化信息对患者的预后作出判断。 二、针对脑卒中脑电时间序列数据进行了分析,从西方医学的角度讨论了正常脑电数据与异常脑中风病脑电数据的特点,提出了脑电时间序列信息的双侧对称度量指标,并以此提出了正常脑电与异常中风脑电的判别方法和脑中风病预后判别算法。 三、针对量化的脑卒中脑电时间序列的病灶位置进行了进一步分析,讨论并对比了脑中风病脑电序列的多种特征,提出了脑中风病脑电时间序列病灶位置分区分析方法。 四、针对轨道交通客流时间序列数据进行了分析,主要讨论了轨道交通客流数据的类周期性特点,对已有的城市道路交通流时间序列相似性与预测方法进行改进,并应用于轨道交通客流数据分析,提出了基于相似模式的轨道交通客流长期预测算法。
[Abstract]:Time series analysis is widely used in various fields. Similarity analysis is the basis of time series analysis and prediction is an important problem in time series analysis. This paper aims at the similarity and prediction of time series data and passenger flow time series data in medical science. Four problems were studied and discussed in this paper: dynamic analysis and prognostic judgement of Boolean time series of stroke symptoms and syndromes, characteristic analysis and prognosis judgement of EEG data of stroke; The characteristic analysis of focal location and EEG signal of stroke, the similarity and prediction analysis of passenger flow data of rail transit. The main work and innovations of this paper are summarized as follows: firstly, the Boolean time series data of stroke symptoms and syndromes are analyzed. This paper discusses the characteristics of Boolean time series monitoring data of cerebral apoplexy symptoms and syndromes from the perspective of traditional Chinese medicine, and proposes an association rule mining algorithm for Boolean time series data. The prognosis of patients was judged by dynamic change information of symptoms and syndromes. Secondly, this paper analyzes the EEG time series data of stroke, discusses the characteristics of normal EEG data and abnormal cerebral apoplexy EEG data from the point of view of western medicine, and puts forward the bilateral symmetry metric index of EEG time series information. The method of distinguishing normal EEG from abnormal apoplexy and the algorithm of predicting the prognosis of stroke were put forward. Thirdly, the location of the focus of the time series of stroke is further analyzed, and various features of the EEG sequence of stroke are discussed and compared, and the method of location analysis of the lesion in the time series of stroke is put forward. Fourthly, the paper analyzes the time series data of rail transit passenger flow, mainly discusses the quasi-periodicity of the passenger flow data, and improves the similarity and prediction method of the existing urban road traffic flow time series. Based on the analysis of passenger flow data of rail transit, an algorithm of long-term passenger flow prediction based on similarity model is proposed.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：R743.3;O211.61
本文编号：2230114
[Abstract]:Time series analysis is widely used in various fields. Similarity analysis is the basis of time series analysis and prediction is an important problem in time series analysis. This paper aims at the similarity and prediction of time series data and passenger flow time series data in medical science. Four problems were studied and discussed in this paper: dynamic analysis and prognostic judgement of Boolean time series of stroke symptoms and syndromes, characteristic analysis and prognosis judgement of EEG data of stroke; The characteristic analysis of focal location and EEG signal of stroke, the similarity and prediction analysis of passenger flow data of rail transit. The main work and innovations of this paper are summarized as follows: firstly, the Boolean time series data of stroke symptoms and syndromes are analyzed. This paper discusses the characteristics of Boolean time series monitoring data of cerebral apoplexy symptoms and syndromes from the perspective of traditional Chinese medicine, and proposes an association rule mining algorithm for Boolean time series data. The prognosis of patients was judged by dynamic change information of symptoms and syndromes. Secondly, this paper analyzes the EEG time series data of stroke, discusses the characteristics of normal EEG data and abnormal cerebral apoplexy EEG data from the point of view of western medicine, and puts forward the bilateral symmetry metric index of EEG time series information. The method of distinguishing normal EEG from abnormal apoplexy and the algorithm of predicting the prognosis of stroke were put forward. Thirdly, the location of the focus of the time series of stroke is further analyzed, and various features of the EEG sequence of stroke are discussed and compared, and the method of location analysis of the lesion in the time series of stroke is put forward. Fourthly, the paper analyzes the time series data of rail transit passenger flow, mainly discusses the quasi-periodicity of the passenger flow data, and improves the similarity and prediction method of the existing urban road traffic flow time series. Based on the analysis of passenger flow data of rail transit, an algorithm of long-term passenger flow prediction based on similarity model is proposed.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：R743.3;O211.61
【参考文献】
相关期刊论文 前10条
1 雷皇英;;动态脑电图对脑梗死急性期早期诊断的意义[J];湘南学院学报(医学版);2008年02期
2 彭军强;吴平东;殷罡;陈之龙;马树元;;Hilbert-Huang变换结合近似熵在疲劳驾驶时脑电分析中的应用[J];公路交通科技;2008年06期
3 白惠霞;;动态脑电图变化在急性进展型脑梗死中的应用[J];中国实用神经疾病杂志;2009年05期
4 沈斌;姚敏;;一种新的动态关联规则及其挖掘算法[J];控制与决策;2009年09期
5 曾海泉;胡勤友;周水庚;胡运发;;基于互关联后继树的时序模式挖掘[J];模式识别与人工智能;2003年03期
6 王金龙;徐从富;;启发式全局偏序挖掘算法[J];模式识别与人工智能;2008年02期
7 张志芳,俞丽华,贾莉娟,周敏杰,张捷,卢娜;37例心肺复苏后昏迷病人的脑电图分析[J];临床脑电学杂志;2000年04期
8 吴大嵘,梁伟雄,温泽淮,黄燕,赖世隆;建立中风病血瘀证宏观辨证量化标准的方法探讨[J];广州中医药大学学报;1999年04期
9 赵丽娜;王保强;尧德中;;基于信号处理的脑电相位同步性分析方法研究[J];生物医学工程学杂志;2008年02期
10 王化贤;王东生;邱佳陵;李英群;;动态脑电图监护对大面积脑梗死早期诊断的临床价值[J];中国医师进修杂志;2006年28期
相关博士学位论文 前1条
1 李星毅;基于相似性的交通流分析方法[D];北京交通大学;2010年
，本文编号：2230114
本文链接：https://www.wllwen.com/yixuelunwen/shenjingyixue/2230114.html
最近更新
教材专著
