【作者】 王萍；

【导师】 单剑锋；

【作者基本信息】 南京邮电大学 ， 电子与通信工程（专业学位）， 2015， 硕士

【摘要】 小波变换作为一种数学理论和方法在处理信号和图像、智能语音判别、核探测邻域、机器状态的识别、超声波探测等工程应用范围内引起了越来越多的关注和重视。齿轮振动信号在机械故障诊断中起到关键性作用,通过采集齿轮运行过程中产生的振动信号来分析判断齿轮的故障原因。在采集振动信号的过程中,不可避免会受到大量背景噪声信号、零件摩擦等外界的影响,振动信号在传输过程中也会受到噪声的干扰。所以一般情况下采集到的齿轮振动信号是含有噪声的信号。因此,对齿轮振动信号进行去噪研究,提取原始信号从而为后期故障诊断提供准确信息是本文研究的重点。本文主要针对齿轮的振动信号进行去噪研究。由于齿轮振动信号本身的特点及噪声的影响使得振动信号具有非平稳的特性。然而傅里叶变换只在频域范围内有效,不能满足非平稳信号的分析。小波变换弥补了传统去噪的缺点,具有时频局部分析的优势,因此采用小波变换处理齿轮振动信号较为合理。由于小波函数的选取会直接影响去噪效果,从小波函数的正交性、对称性、紧支性等方面考虑,本文选用db5小波基较其他小波函数对分析齿轮振动信号更加有效。由于硬阈值函数法重构信号时小波系数的不连续性容易出现“伪吉布斯”现象,软阈值函数法小波系数的估计值与实际值之间存在偏差的情况。本文构造了一种可以介于硬、软阈值函数之间灵活改变的新阈值函数。阈值的选取对小波去噪效果起至关重要的影响,阈值选取大时,部分有用信号会当作噪声去掉,有用信号信息丢失太多;阈值过小又会残留过多的噪声成分,容易引起信号的失真,影响去噪效果。为了取得更好的去噪效果本文又对阈值作了进一步的改进。将小波分析应用于齿轮故障的分析诊断方面,在齿轮早期的故障诊断方面取得良好的效果,这在保证齿轮设备的安全可靠运行方面起着重要的作用。小波分析具有广阔的应用前景,是一项值得推广的新技术。更多还原

【Abstract】 Wavelet Transform as a mathematical theory and methods in signal and image processing, intelligent voice identification, nuclear detection neighborhood, machine status and fault diagnosis recognition, ultrasonic detection range of applications such as engineering attracted more and more concern and attention. Gear vibration signals in mechanical fault diagnosis plays a key role, gear vibration signal is used to analyze the state of the running gear in the process, in order to determine the gear failure. However, in the process of collecting the vibration signal, will Affected from a lot of background noise signal and components friction, vibration signal during transmission will also be noise. Therefore, under normal circumstances the collected gear vibration signal is noisy, therefore, gear vibration signal de-noising research, extract the original signal to provide accurate information for the post fault diagnosis is the focus of this study.This thesis studies on vibration signal de-noising gear. Due to their characteristics and noise of gear vibration signal so that the vibration signal with a non-stationary characteristics. However, the Fourier transform is only valid in the frequency domain, can not meet the analysis of non-stationary signals.Fourier transform is only valid in the frequency domain, can not meet the analysis of non-stationary signals. The wavelet transform makes up for the shortcomings of traditional denoising, has the analysis capability of time-frequency localization, so using wavelet transform processing gear vibration signal is more reasonable. consider the wavelet function orthogonality, symmetry, compactly supported other aspects, the thesis selection db5 wavelet function analysis wavelet more efficiently than other gear vibration signals. Because of the hard threshold function method discontinuity wavelet coefficients prone to "pseudo-Gibbs’ phenomenon reconstructed signal, the situation there is a deviation between the estimated value and the actual value of the soft threshold function method wavelet coefficients.In order to avoid the shortcomings of the " pseudo-Gibbs’ phenomenon and soft threshold of bias. In this paper, we construct a new threshold function, the new threshold function can change flexible between soft threshold function and hard threshold function. Appropriate wavelet function affects denoising effect directly, The threshold plays a key role in the wavelet denoising, the threshold is too Large,part of the useful signal will be removed as noise, useful information will lost too much; the threshold is too small, residual excessive noise component will easily lead signal distortion, affecting denoising effect. In order to achieve better denoising effect on the threshold of this article has been further improved.The wavelet analysis applied to gear fault diagnosis, and achieved good results in the diagnosis of early gear failure, which plays an important role in ensuring the safe and reliable operation of the gear device on. Wavelet analysis has broad application prospects, is a worthy new technologies.更多还原