摘要
在信息熵模型的基础上 ,将多分辨率分析和奇异谱分析有效结合 ,提出了多分辨率奇异谱熵及其算法 ,该指标能够表征不同尺度上信号能量分布的奇异状况。进一步 ,针对奇异谱分析是基于信号二阶平稳的假设 ,本文提出了基于高阶累积量的高阶多分辨率奇异谱熵 ,它能够充分体现信号的内在信息 ,提高了信号形式变化的适应性和对于嵌入维数变化的稳定性。通过对机械振动信号的仿真分析 ,该信息熵能够作为设备运行状态识别的有效依据。
The multi resolution analysis and singular-spectrum analysis are combined effectively based on information entropy model, and a new method realized by multi resolution singular spectrum entropy is proposed. This guideline can appreciate the singularity condition of energy distributing of signal in different scale. Furthermore, since singular spectrum analysis is on the assumption of second order stability of signal, we introduce high order multi resolution singular spectrum entropy. The internal information of signal can be embodied and the adaptability of the algorithm to various signal forms can be increased. Finally, the new method is simulated in analysis of mechanical vibration signal and is proved that this information entropy can be effective index for the recognition of system running condition.
出处
《传感技术学报》
CAS
CSCD
2004年第4期547-550,共4页
Chinese Journal of Sensors and Actuators
基金
河北省自然科学基金资助项目 (6 0 3334)
国家十五重大技术装备研制项目 (科技攻关 )计划 (ZZ0 2 -13B -0 2 -0 3 -1)
关键词
信息熵
奇异谱分析
多分辨率奇异谱熵
高阶累积量
information entropy
singular-spectrum analysis
multi-resolution singular-spectrum entropy
high order cumulated quantity
