摘要
给出了一种驱动桥故障特征提取的方法 ,即无论驱动桥处于工作时的动态 ,还是非工作时的静态 (采用锤击制造源信号 ) ,所提取的信号都经过离散小波消噪处理 ,和小波包分解。对工作时的动态 ,需再用倒谱变换方法进行特征提取。此方法成功地解决了特征提取环境与工作环境不一致及动、静态故障特征提取方法差异过大的矛盾。用此方法提取的神经网络训练样本 ,会提高系统辨识的精确性。
A kind method of rare-shaft trouble character collected was given, that is whether rear axle is working or not, all signals were handled by discrete wavelets and decomposed by wavelet packets. When rare-shaft is working, character will be gotten by cepstrum. This method is succeed to resolve a environmental contradiction of character collection and working no-fitting. It also resolve a contradiction of difficulty to get trouble character of moving and stationary. The training models of getting by this method can rise up distinguishable accuracy of system.
出处
《计算机测量与控制》
CSCD
2003年第8期580-582,共3页
Computer Measurement &Control
基金
国家自然科学基金资助项目 (5 9835 0 5 0 )
关键词
驱动桥
故障特征提取
倒谱
小波变换
时频域处理
rear axle
wavelet analysis
cepstrum
trouble character
character collection
