【作者】 赵玲；

【导师】 秦树人；

【作者基本信息】 重庆大学 ， 机械电子工程， 2010， 博士

【摘要】 机械故障诊断领域的研究中,最重要、最关键、也是最困难的问题之一就是故障特征信号的特征提取。机械系统自身的复杂结构,设备故障的产生、发展、工况的变化导致了振动信号的非平稳特性。这就使得振动信号的非平稳特征提取与表示成为被广泛关注的重要研究方向,其中信号的降噪预处理、瞬时频率特征提取、诊断依据的确定是研究的主要内容。分形理论的出现使人们能够以新的观念和手段来处理复杂世界里的难题,透过扑朔迷离、无序的混沌现象和不规则的形态,揭示隐藏在复杂现象背后的规律,为解决机械工程领域中的许多问题提供了行之有效的新途径,也为机械设备的故障诊断提供了新的思路和方法。本文综合利用非平稳、非高斯信号处理中最受关注的小波变换(WT)和Hilbert-Huang变换(HHT)等时频处理方法的优点,结合新兴的分形理论,提出了基于分形、小波和神经网络的故障诊断方法,并对机械振动信号的降噪、故障特征提取技术进行了深入的研究,基于文中所提出的信号分析方法,提出了基于总线系统的测试系统架构。主要研究工作如下:首先介绍了课题的来源、背景、研究现状及其研究意义,对研究对象进行分析,研究了旋转机械的典型故障及其振动特征,对旋转机械故障诊断的基本过程做了详细阐述;介绍了论文的主要工作和创新点。对于旋转机械设备故障,所分析的信号常是非平稳信号,可能包含着尖峰、突变等表现,还可能包含非平稳的白噪声,而这些往往是信号的高频部分,同时由于信号中包含的故障信息一般较弱,常常淹没在噪声信号中,要分析这类信号,必须对信号进行降噪处理。因此,在第二章中介绍了在信号降噪方面做的研究,提出并实现了基于SVD和高斯小波的滤波降噪方法。该方法可以很好的降低噪声信号,有效提取信号中有用成分,具有良好的瞬态信息提取能力,为正确识别故障特征提供了保证。HHT方法是最近比较受到关注的信号分析方法,但是自身存在的端点效应、模态混叠等问题阻碍了它在故障特征提取领域的应用。在第三章中对完善HHT分析方法方面进行了研究。在介绍HHT基本理论的基础上,结合实际应用中遇到的问题,提出并实现了基于改进掩膜信号法的提取振动信号非平稳特征方法。该方法通过添加掩膜信号有效抑制了模态混叠现象,通过对原始数据依据掩膜信号进行延拓,解决端点效应问题。分形学是一门横断学科的新理论,它是对传统几何分析手段不能描述的极不光滑、无任何规则集合以及对无序的、不稳定的、非平衡的和随机的自然界绝大多数事物的一般结构进行研究的一门新诞生的学科。分形理论的引入,为解决机械工程领域中的许多问题提供了行之有效的新途径。第四章中通过对分形理论以及分形维数计算方法的深入研究,提出并实现了基于HHT分析结果的分形维数改进算法。通过EMD的结果来确定权重因子q取值范围,既能全面反映出多重分形的特征,又可以把各参数之间的相关性尽量减少到最小。降低了计算量,提高了算法的运行效率。在第五章中提出了综合利用分形理论、小波、神经网络进行机械系统故障特征辨识的方法。通过将时频分析方法与分形理论相融合,实现了信号时频域内非平稳特征的参数化描述。在系统开发方面,提出了基于消息总线的测试系统架构。通过采用多线程同步数据采集技术、内存数据库技术、总线技术以及虚拟仪器技术来实现多层结构的测试系统,应用模块化开发方法,对旋转机械振动信号分析系统进行了总体设计。文章最后对本文工作进行了总结并对故障特征提取技术的研究进行了展望。更多还原

【Abstract】 Vibration signals generated by rotary machinery contain lots of fault information. By analysis into them, state changing of the parts in the mechanical equipment can be recognized, and the fault type can be found out. The most important and crucial problem in the mechanical fault diagnosis is the feature extraction method of the fault characteristic signal, while it is the very problem most difficult to solve. The vibration signals of mechanical equipment often represent nonstationarities due to occurrence and variance of fault, influence of nonnormal working condition and inherent nonlinearity of equipment. Because of the complexity of the dynamic signal and the multidisciplinary cross and fusion characteristic of the extracted signal, the feature extraction method has been the most important research direction concerned by the researchers, in which signal de-noising, instantaneous frequency feature extraction,diagnostic bases determination are the main elements.Making use of the non-stationary signal processing methods merits such as Wavelet Transform (WT) and Hilbert-Huang Transform (HHT), combining with fractal theory, the fault diagnosis method based on wavelet packet transform, fractal theory and neural network are put forward, and this dissertation investigates noise reducing method of the machinery vibration signal and the feature extraction technique of the mechanical equipment thoroughly. Based on these methods, a new kind of message bus based testing system architecture was proposed. Its contributions list as follows:In terms of signal de-noising, a new method of filtering and de-noising based on optimal Gaussian wavelet and Singular Value Decomposition (SVD) is put forward. Feature extraction and signal de-noising of the fault signal have been the most important investigation in the signal processing. By reducing the noise of machinery vibration signal, the mechanical fault information can be obtained effectively. The new de-noising method, which possesses better transient information extraction ability, could reduce the noise and extract the period of the signal effectively and assure the validity of the fault feature recognition.In terms of HHT improving, a new extraction method based on improving masking signal is put forward. Aim at solve the mode mixing problem, the masking signal is added before Empirical Mode Decomposition (EMD). As a result of simulation and experiment, it is shown that mode mixing can be effectively avoided. Through extending the data, it’s the effective technique to restrain the ending effect of HHT. The example shows that the method improves the rationality and veracity of the signal feature extraction with HHT.After the investigating of fractal theory and fractal dimension calculation, the impvoved algorithm of fractal dimension calculation based on the result of HHT is proposed. The algorithm is determined the range of the weighting factor q by the EMD result, that not only reflecting the characteristics of the multifractal, but also minimizing the correlation among the parameters as much as possible. The method reduced the computation and improved the operating efficiency of the algorithm.By the integration of time-frequency analysis methods and fractal theory, the fault characteristics identification method based on utilization of fractal theory, wavelet, neural network is proposed. It has realized parametric description on signal characteristics of non-stationary in both time and frequency domain.In terms of system exploitation, a new kind of message bus based testing system architecture is proposed. The specific implementation is that gathering signal by computer standard bus based data acquisition card; processing, analyzing the data and illustrating the result by virtual instrument; and operating virtual instrument by message bus. Investigating the system structure of the module software, designed the uniform framework of the system module.There are the summarization of the article and expectation of the feature extraction technology development in the end of article.更多还原