摘要
为更好地分析复杂装备事故非线性耦合特性及成因,利用小波包消噪(WPD)改进多重分形(MF)去趋势波动分析法(MFDFA),并结合q阶矩配分函数分析MF谱;通过对比随机重排和相位随机化2种变换事故序列MF谱,探讨非线性时间序列的长程相关性与胖尾分布对MF特性成因的贡献程度;最后,以美空军1946-1996年A级飞行事故为例分析其非线性耦合特征。结果表明:该事故率序列广义hurst指数和分形谱宽度显著不为常数,分布范围分别为[1.2138,3.4314]和[0.2937,1.7025],具有明显MF特性。该MF特性是由长程相关性为主因,胖尾分布为次因共同作用所致。
In order to better analyze non-linear coupling characteristics and causes of complex materiel accidents,MF detrended fluctuation analysis(MFDFA)was improved by using WPD,and MF spectrum was analyzed with qthmoment partition function.Secondly,influence on multifractality of long-range correlation and fated-tail distribution of nonlinear time series was discussed through comparing MF spectrum of two kinds of transformed time series,random shuffling and phase randomization.Finally,with class A flight accidents data of United States air force(USAF)from 1946 to 1996 as an example,nonlinear coupling characteristics of those accidents were analyzed.The results show that it has obvious multifractality with its distribution range of general hurst index and fractal spectrum width at[1.2138,3.4314]and[0.2937,1.7025]respectively.The major cause for its multifractality is long range correlation and minor one is fated-tail distribution.
作者
李超
LI Chao(Equipment Management&UAV Engineering College,Air Force Engineering University,Xi'an Shaanxi 710051,China)
出处
《中国安全科学学报》
CAS
CSCD
北大核心
2019年第12期97-102,共6页
China Safety Science Journal
基金
国家自然科学基金资助(71601183)。
作者简介
李超(1984-),男,河南信阳人,博士,讲师,主要从事装备系统工程与管理决策、装备安全性分析与控制等方面的研究。E-mail:leecharle@sina.com。
