摘要
针对往复压缩机活塞-缸套磨损故障微弱信号特征识别问题,提出一种识别该类信号微弱特征的自适应非抽样提升小波包方法(AULSP)。该方法以分解层信号所有样本的预测差值平方和最小为目标函数,算出与信号特征自适应匹配的初始算子,并构造非抽样算子算出下一层各频带信号。对各层细节信号进行阈值处理并重构,对降噪后的信号再进行小波包分解。各分解频带信号长度与原始信号的长度相同,无须重构即可识别时域故障微弱信号特征。用这种方法成功提取了某往复压缩机活塞与缸壁发生碰磨故障时产生的弱周期性冲击信号。
Aiming at characteristics identification problem of weak signal for piston-cylinder liner wear fault of reciprocating compressors,a novel method to design adaptive undecimated lifting scheme packet（AULSP） was developed and applied to identify successfully weak-signal fault features of a certain reciprocating compressor.The minimum square sum of prediction difference of all sample points was taken as object function,and the initial operators that adaptively match the weak-signal features were calculated,then undecimated operators were constructed and used to calculate each frequency band on the next level.Noise can be restrained via thresholding operation.The signal length of each frequency band was the same as that of the original signal,thus time-domain fault features could be recognized without reconstruction.AULSP was applied to identify weak periodic impact signals caused by piston-liner wear.
出处
《中国石油大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2011年第1期130-134,145,共6页
Journal of China University of Petroleum(Edition of Natural Science)
基金
国家'863'计划项目(2008AA06Z209)
中国石油天然气集团公司创新基金项目(07E1005)
关键词
提升小波包
往复压缩机
磨损
信号分解
诊断
lifting scheme packet
reciprocating compressor
wear
signal decomposition
diagnosis
