摘要
在机电设备的故障诊断过程中,故障特征辨识通常会受到背景噪声的干扰,如何剔除干扰成分、有效捕捉隐含的故障特征是诊断成败的关键。为了实现这一目的,提出了一种基于改进阈值策略的多小波循环平移降噪算法。针对传统阈值降噪算法的不足,同时结合软、硬阈值函数的特点,对已有的阈值函数进行改进。考虑到降噪过程中产生的PesudoGibbs现象,将循环平移算法引入多小波的降噪过程中。通过滚动轴承的模拟故障实验和轧机齿轮箱的现场诊断,表明循环平移多小波改进阈值降噪方法能有效地滤除背景噪声;运用峭度指标和均方根误差作为量化参照,结果表明该方法处理后的信号信噪比明显提高。
In the fault diognosis of electromechanical equipment, the feature identification is usually influenced by the complex surroundings. As a result, how to reduce the noise disturbance and extract the hidden fault feature effectively has become the key issue in this field. Based on above requirements, a multiwavelet cycle-spinning denosing method based on improved threshold strategy was introduced. Due to the deficiency of traditional threshold function, the existing algorithm was improved by considering the characteristics of soft and hard threshold functions. Owing to the Pesudo-Gibbs phenomenon of singular points in denoising process, the cycle -spinning method was combined with multiwavelet decomposition and reconstruction Being validated by seeded fault experiment and in situ diagnosis of rolling mill gearbox, the above mentioned method is suitable for noise reduction in complicated industrial field. On the other hand, the SNR increase was also demonstrated by kurtosis index and root mean sauare error comparison,
出处
《机械设计与制造》
北大核心
2015年第6期201-205,共5页
Machinery Design & Manufacture
基金
北京市教委科技计划项目(KM201410005027)
