摘要
提出一种基于有效信息重构的故障旋转机械振动加速度信号积分方法。该方法首先利用集总经验模态分解(EEMD)对积分信号进行分解,然后将峭度指标、均方差值和奇异值组合成特征向量对分解信号进行有效特征提取,并通过欧氏距离搜索提取分解信号中隐含的有效特征分量,最后通过阈值限定将有效特征分量进行叠加重构得到所需有效积分信号。仿真分析和实测数据结果表明,该方法在去除噪声和保留有效特征信息方面,比传统积分方法具有更高的精度和优越性。
A faulty rotation machinery vibration acceleration signal integration method was pro- posed based on useful information reconstruction. Firstly, ensemble empirical mode decomposition （EEMD） was utilized to decompose the integral signals. Secondly,the values of kurtosis,mean square error and singular value were proposed to be combined into a feature vector, and further used to ex- tract the effective feature for the decomposed signals. Then, the effective feature components were ex- tracted by Euclidean distance search. Finally, the effective feature components were reconstructed by threshold limit, and the desired integral signals were obtained. The effectiveness and practicability of the method were testified by simulation analysis and measured data. Results indicate that this method is excellent at removing noise and reserving effective feature information,and possesses high precision and superiority.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2015年第18期2511-2517,共7页
China Mechanical Engineering
基金
国家自然科学基金资助项目(51475405)
国家重点基础研究发展计划(973计划)资助项目(2014CB046405)
河北省自然科学基金资助项目(E2013203161
E2013203114)
河北省研究生创新资助项目(00302-6370002)
关键词
振动加速度
时域积分
频域积分
频域-时域混合积分
有效信息重构
vibration acceleration
time domain integral
frequency domain integral
frequency-time domain mixed integral
useful information reconstruction
