基于混沌理论的往复泵故障特征提取方法  

Study of fault characteristic extraction method of reciprocating pump based on chaotic theory

作　　者：李峰[1] 张来斌[1] 王朝晖[1] 段礼祥[1] 

机构地区：[1]中国石油大学(北京)机电工程学院

出　　处：《石油机械》2009年第2期56-59,91,共4页China Petroleum Machinery

基　　金：教育部新世纪优秀人才支持计划资助项目(NCET-05-0110);中国石油天然气集团公司创新基金项目(07E1005)

摘　　要：特征提取是往复泵状态监测与故障诊断的关键环节。往复泵的振动信号含有丰富的频率成分,大量的高频成分使得确定的时间延迟过小,导致重构的相空间难以充分恢复系统的动力学行为。分别利用小波包奇异值分解(WPSVD)降噪方法和小波包迭代奇异值分解(WPISVD)降噪方法对往复泵振动信号进行降噪以获得整体信号和低频部分信号;利用C-C方法合理地确定低频部分的相空间重构参数;利用得到的相空间参数提取整体信号的关联维数和最大Lyapunov指数。结果表明,关联维数和最大Lyapunov指数可作为往复泵故障诊断的特征量。The characteristic extraction is a pivotal stage in the condition monitoring and fault diagnosis of reciprocating pump. The vibration signal of reciprocating pump includes a plenty of frequency components. In the process of reconstructed phase space, a plenty of high-frequency components lead the delay time too small, the system dynamic behaviors can not be fully recovered by the reconstructed phase space. After analysis, this paper presents that whether the reconstructed phase space of the vibration signal of reciprocating pump is reasonable or not, it depends on the reasonability of the reconstructed phase space of the low-frequency components. Firstly, denoising signal and the multiple-frequency components of denoising signal can be obtained with the method of wavelet packet and singular value decomposition (WPSVD) and the method of wavelet packet and iterative singular value decomposition (WPISVD). Secondly, the phase space parameters of low-frequency components of denoising signal can be confirmed reasonable with the method of C-C. Finally, the correlation dimension and the largest Lyapunov exponent of denoising signal can be extracted with the phase space parameters. The result shows that the correlation dimension and the largest Lyapunov exponent can effectively distinguish different states of reciprocating pump valves.

关 键 词：往复泵 特征提取 关联维数 最大LYAPUNOV指数 

分 类 号：TE926[石油与天然气工程—石油机械设备] O415.5[理学—理论物理]