摘要
将一维时间序列采用相空间重构算法将其扩展到高维相空间中,通过研究时间序列在高维空间中的邻近点分布规律和运动特点,提出采用递归图和近似熵获取原始时间序列的动力学行为的方法。对递归图受噪声的影响,近似熵的数值稳定性进行研究,结果表明两种算法在较短的数据集下就能有效地对信号的复杂度进行描述。同时将近似熵和递归图用于设备振动信号的复杂度描述,对两组不同的信号的递归图和近似熵进行比较,得出的结论与实际相符。
One dimension time series is extended to high dimension phase space by using phase space reconstruction algorithm. By studying the distribution rule and movement characteristic of neighborhood points of time series in high dimension. The method of recurrence plot and approximate entropy used for getting the dynamics behavior of original time series were proposed. The influence of noise on recurrence plot and numerical calculation stability of approximate entropy method are studied. The results show that the two methods can depict the complexity of signal effectively. Approximate entropy and recurrence plot are used for characterizing complexity analysis of machinery fault signal, the approximate entropy and recurrence plot of two different signals are compared, The calculation results are consistent with the fact.
出处
《机械强度》
EI
CAS
CSCD
北大核心
2006年第3期317-321,共5页
Journal of Mechanical Strength
基金
国家博士点基金20020008019)
机械传动与制造工程湖北省重点实验室基金2003A04)
湖北省自然科学基金2005ABA287)资助项目~~
关键词
递归图
近似熵
相空间重构
故障诊断
Recurrence plot
Approximate entropy
Phase space reconstruction
Fault diagnosis
