[1]王俐莉,刘力维,熊艳晔.基于厚尾分布的异常信号检测方法[J].南京理工大学学报(自然科学版),2011,(01):19-20.
 WANG Li-li,LIU Li-wei,XIONG Yan-ye.Abnormal Signal Detection Method Based on Heavy-tailed Distribution[J].Journal of Nanjing University of Science and Technology,2011,(01):19-20.
点击复制

基于厚尾分布的异常信号检测方法()
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
期数:
2011年01期
页码:
19-20
栏目:
出版日期:
2011-02-28

文章信息/Info

Title:
Abnormal Signal Detection Method Based on Heavy-tailed Distribution
作者:
王俐莉12刘力维3熊艳晔1
1. 海军指挥学院科研部,江苏南京210016; 2. 解放军理工大学指挥自动化学院,江苏南京210007; 3. 南京理工大学理学院,江苏南京210094
Author(s):
WANG Li-li12LIU Li-wei3XIONG Yan-ye1
1. Scientific Researching Department,Naval Command College,Nanjing 210016,China; 2. Institute of Command Automation,PLA University of Science and Technology,Nanjing 210007,China; 3. School of Sciences,NUST,Nanjing 210094,China
关键词:
期望最大化算法 极大似然估计 Cook 距离 Q 函数 异常信号
Keywords:
expectation maximization algorithm maximum likelihood estimation Cook distance Q function abnormal signals
分类号:
TP274
摘要:
为了检测具有厚尾分布信号的异常值,提出将基于EM 算法的统计诊断方法应用于正态逆高斯信号的异常信号检测。为了避免计算贝塞尔函数的困难,考虑将正态逆高斯信号看作缺失变量,通过EM 算法求解参数的极大似然估计,简化了计算过程; 通过Q 函数代替对数似然函数,提出了在正态逆高斯分布的信号中基于Q 函数的影响度量方法,分别给出了数据删除模型与局部影响分析的诊断统计量,理论和实例计算结果表明: 该文所提出的检测方法对于正态逆高斯信号的检测效果明显。
Abstract:
In order to detect the abnormal signals with a heavy tailed distribution,the statistical diagnosis method based on expectation maximization( EM) algorithm is applied to abnormal signal detection for normal inverse Gaussian signal. To avoid the difficulties of calculating Bessel functions, the normal inverse Gaussian signal is treated as missing data. The maximum likelihood estimation of parameters is obtained using the EM algorithm,simplifying the calculation. The Q function takes place of the log-likelihood function and a Q-based measurement for the diagnosis method of normal inverse Gaussian distribution is proposed. The results of the calculation example and computer simulation demonstrate the superiority of the methods proposed here.

参考文献/References:

[1] Markou M,Singh S. Novelty detection: A review. Part 1. : Statistical approaches[J]. Signal Processing,2003, 83( 12) : 2481-2497.
[2] Cook R D. Assessment of local influence[J]. J R Statist Soc B,1986,48( 2) : 133-169.
[3] Zhu H T,Lee S Y. Local influence for incomplete-data models[J]. J R Statist Soc B ( Part1 ) ,2001 ( 63 ) : 111-126.
[4] Zhu H T,Lee S Y,Wei B C,et al. Case-deletion measures for models with incomplete data [J]. Biometrika,2001,88( 3) : 727-737.
[5] 李爱萍,解锋昌. 基于数据删除的Poisson-Gamma 模型的影响评价[J]. 南京理工大学学报( 自然科学版) ,2006,30( 6) : 797-800.
[6] Karlis D. An EM type algorithm for maximum likelihood estimation of the normal-inverse Gaussian distribution [J]. Statistics & Probability Letters,2002,57( 1) : 43-52.
[7] Tsihrintzis G A,Tsakalides P,Nikias C L. Signal detection in severely heavy-tailed radar clutter[A]. Twenty- Ninth Asilomar Conference on Signal,Systems and Computers[C]. CA: Pacific Grove,1995: 865-869.

备注/Memo

备注/Memo:
基金项目:国家自然科学基金( 60874118) 作者简介:王俐莉( 1981-) ,女,博士生,讲师,主要研究方向:计算机应用技术,E-mail: liliwang_1981@ yahoo. com. cn。
更新日期/Last Update: 2012-02-28