基于改进LCD的WVD算法及其在故障诊断中应用的试验研究

发布时间：2018-03-29 19:26

本文选题：Wigner-Ville分布　切入点：局部特征尺度分解　出处：《燕山大学》2015年硕士论文

【摘要】：旋转机械故障诊断能够为设备的安全提供必要保障。故障振动信号的时频分析,是故障诊断中的重要手段。振动信号特征提取是关键所在,在众多方法中,自适应时频分析的研究越来越受到关注。局部特征尺度分解(Local Characteristic-scale Decomposition,LCD)发展经历的时间很短,但该方法在计算效率及分量准确性上优势明显。另外,Wigner-Ville分布(WVD)优良特性突出,具有能量聚集性高、时频分辨率高及时变特性良好等优点,被广泛应用于信号分析与处理领域。但是,WVD分布计算过程不满足可加性,将产生交叉项,影响了该方法的工程实际应用价值。针对上述问题,本文提出了基于改进LCD与WVD结合抑制WVD分布交叉项的算法。首先,本文对旋转设备常见故障机理与频率特征做了研究,并用试验设备进行模拟,为以后的故障信号分析打下基础。深入研究LCD基本理论,并与经验模态分解(Empirical Mode Decomposition,EMD)、固有时间尺度分解(Intrinsic time-scale decomposition,ITD)做对比研究以证明LCD方法的优越性。针对LCD方法分解过程中信号失真、能量泄漏现象,本文提出了基于三次B样条插值(Cubic B-Spline Interpolation,CBI)改进的LCD方法。将改进后的LCD方法应用到模拟仿真信号、试验振动信号处理分析中,证明了此方法的有效性。其次,研究了WVD分布的基本理论及性质,对其交叉项的产生原因做了分析,有文献表明没有交叉项的WVD分布是不存在的。本文从抑制交叉项的思路出发提出了基于改进LCD的WVD分布,该方法将信号用改进LCD分解为近似单分量信号,再对各分量求WVD分布,最终求和得到原信号的WVD分布。模拟仿真信号与试验振动信号的分析结果表明,本文方法能够成功抑制WVD交叉项。最后,将此方法与传统WVD方法、传统LCD与WVD分布结合方法及EMD与WVD分布结合方法进行对比。结果证明本文提出的方法在抑制交叉项成分方面效果更好。燃气轮发电机故障实例分析表明了方法的有效性,该工程实例分析也证明了本文方法能够应用于工程实际的旋转设备故障诊断中。
[Abstract]:Fault diagnosis of rotating machinery can provide necessary guarantee for the safety of equipment. Time-frequency analysis of fault vibration signal is an important means in fault diagnosis. Feature extraction of vibration signal is the key, among many methods, The research of adaptive time-frequency analysis has attracted more and more attention. The development time of local characteristic scale decomposition (LCDs) is very short, but the advantages of this method in computing efficiency and component accuracy are obvious. In addition, Wigner-Ville distribution (WVD) has excellent characteristics. It has the advantages of high energy aggregation, high time-frequency resolution, high time-frequency resolution, etc. It is widely used in the field of signal analysis and processing. However, the calculation process of WVD distribution does not satisfy the additivity, which will result in crossover. The practical application value of this method is affected. In view of the above problems, an algorithm based on the combination of improved LCD and WVD to suppress the crossover term of WVD distribution is proposed. Firstly, the mechanism of common faults and frequency characteristics of rotating equipment are studied in this paper. The test equipment is used to simulate, which lays a foundation for future fault signal analysis. The basic theory of LCD is deeply studied. Compared with empirical Mode decomposition (EMD) and intrinsic time-scale decompostion (ITD), the advantages of LCD method are proved. The phenomena of signal distortion and energy leakage during LCD decomposition are discussed. In this paper, an improved LCD method based on cubic B-spline interpolation cubic B-Spline interpolation is proposed. The improved LCD method is applied to the simulation signal and the experimental vibration signal processing and analysis, and the effectiveness of the method is proved. In this paper, the basic theory and properties of WVD distribution are studied, and the causes of its crossover term are analyzed. It is shown that the WVD distribution without crossover term does not exist. In this paper, a WVD distribution based on improved LCD is proposed based on the idea of suppressing crossover term. In this method, the signal is decomposed into an approximate single component signal by improved LCD, then the WVD distribution is obtained for each component, and the WVD distribution of the original signal is obtained by summing up the original signal. The analysis results of the simulated signal and the experimental vibration signal show that, This method can successfully suppress the WVD crossover term. Finally, this method is compared with the traditional WVD method. The results show that the method proposed in this paper is more effective in suppressing the cross component. The fault analysis of gas turbine generator shows the effectiveness of the method. The analysis of the engineering example also proves that this method can be applied to the fault diagnosis of rotating equipment in engineering practice.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TH165.3

【参考文献】