基于小波变换的泄露故障特征提取研究  

Leak Fault Feature Extraction Based on Wavelet Transform

作　　者：黄建招[1] 谢建[1] 李良[1] 

机构地区：[1]第二炮兵工程学院202教研室,陕西西安710025

出　　处：《计算机测量与控制》2012年第3期586-589,共4页Computer Measurement &Control

摘　　要：为研究液压系统管路泄漏对压力脉动信号的影响,提出利用改进的小波消噪算法和小波包对压力脉动信号进行消噪和特征提取;针对传统小波变换阈值函数在去噪处理中存在的恒定偏差、不连续等缺点,提出一种改进阈值函数和新阈值相结合的新方法,将该方法与平移不变量方法相结合,避免了传统小波变换去噪时在不连续点存在的Pseudo-Gibbs现象;通过利用改进小波消噪方法和小波包对FESTO试验系统采集的正常和故障压力脉动信号进行分析比较,研究结果表明,不同工况下压力脉动信号3个主要能量频带的分布特性,可以作为泄露检测和识别的故障特征。In order to study the influence on the pressure fluctuation caused by pipe leak, an improved wavelet denosing algorithm and wavelet packet analysis was used to do denosing and feature extraction of the pressure fluctuation. Aiming at the disadvantages of the constant deviation, the discontinuous and so on, by the method of the classic soft and hard threshold function with the generalized threshold, an new method composed of improved threshold function and threshold was presented. In order to avoiding the Pseudo--Gibbs phenomenon in discontinuous points by using the classic wavelet transform, the translation invariance wavelet and the new method were comb;ned together. The pressure fluctuation udder normal and fault working conditions by FESTO experiment were collected and analyzed, the result shows that under different conditions, the distribution characteristic of the three main energy of the pressure fluctuation are difference, it can be used as a judgeable characteristic parameter for leak fault diagnosis.

关 键 词：小波变换 压力脉动 特征提取 故障诊断 

分 类 号：TP271.31[自动化与计算机技术—检测技术与自动化装置] TH137.74[自动化与计算机技术—控制科学与工程]