【作者】 张君；

【导师】 韩璞；

【作者基本信息】 华北电力大学（河北） ， 热能工程， 2005， 博士

【摘要】 汽轮发电机组属于典型的旋转机械,其振动信号在本质上可分为平稳振动信号和非平稳振动信号两大类,这两类信号都包含有能反映旋转机械工作状态的重要特征信息,有效地利用这些信息对设备的监测和诊断有重要价值。本论文对信号处理的前沿技术之一——小波分析技术进行了深入研究和探索,初步实现了基于小波分析技术的汽轮发电机组故障诊断。1.在查阅大量文献的基础上,系统地回顾了旋转机械故障诊断的有关方法和原理。指出了当前信号分析技术对旋转机械非平稳振动信号分析的不足和现有方法存在的问题。2.研究了小波分析技术的最新发展,深入地探讨了小波分析技术应用于汽轮机故障诊断中必须解决的关键技术难题。3.本文就最优小波包基和消噪阈值的选取问题进行了深入研究。提出了一种基于最优小波包基的消噪方法;对于消噪阈值的选取,提出一种以小波包能量为基础,以原始信号与降噪后信号之间的均方误差(MSE)极小化为目标的基于小波包的降噪算法。检测结果表明,在故障检测前先采用最优小波包基方法对故障信号进行消噪,有利于提高汽轮机振动检测的准确性。4.本文提出了一种新的汽轮机振动监测和故障诊断的特征提取方法——小波包特征提取。小波包作为一种时-频分析手段引入到振动信号分析中,小波包系数可以非常灵活地提供信号在时域和频域的信息。在转子实验台上的实验证明,该方法能够完全满足振动信号分析的要求,是非常有效和切实可行的。5.将上述的理论研究系统整合后,本文设计了一套基于信号处理方法的故障诊断系统样机。针对复杂的小波包分解算法,为了满足振动信号实时分析的要求,设计了一种有效的基于DSP的小波包分解算法。大量实验表明,该方法能够满足振动信号实时分析的需要。该样机不仅能对平稳信号表征的汽轮机组故障进行诊断,而且可以检测机组运行过程中的突变故障,可以有效降低故障的漏报率和误报率。更多还原

【Abstract】 Turbo-generator is a kind of typical rotation machine. The vibration signature of rotating machinery can be classified to stationary and non-stationary signature. Both of the signatures contain important information that can reflect the working station of rotating machinery. It is valuable to make good use the information on supervising machine working station. In this dissertation, wavelet transform technique (an advanced technique in signature processing) was deeply researched and explored. Turbo-generator fault diagnosis based on wavelet transform technique was implemented. 1. With the study of a great deal of literatures, the methods and theory of rotating machinery fault diagnosis were reviewed orderly. The disadvantage and problem existing in the signature processing methods are pointed out. 2. The latest development of wavelet transform technique is researched. The key technique problems that wavelet transform technique applied to turbine generator fault diagnosis are discussed in depth. 3. In this dissertation, we study the problem of wavelet packet basis (BWPB) and denosing threshold in depth. A new method based on BWPB is presented to denoise vibration signature; What about the selection of denoising threshold, we present a new denoising algorithm which is aimed to reduce the mean square error (MSE) between the reconstructed signal and the original signal. The detection results show that signal denoising applying BWPB method is in favor of enhancing the detection accuracy of turbine faults. 4. In this dissertation, we put forward a new method based on wavelet packet transform (WPT) for vibration monitoring and fault diagnosis of turbo-generator. The WPT is introduced as an alternative means of extracting time-frequency information from vibration signature. The results of WPT coefficients provide one with arbitrary time-frequency resolution of vibration signal. Extensive experiments on rotor laboratorial platform show that the implementation meets the requirement of vibration signals analysis. It is feasible and effective. 5. After the above-mentioned theoretic research organized together, a set of fault diagnosis system is implemented. For the complicated WPT algorithm, we design an effective method based on DSP to meet the real-time requirement of vibration signals analysis. Extensive experiments show that the DSP based application and it is a feasible and effective solution. The system not only can diagnose the turbo-generator fault presented by stationary signature, but also can diagnose the break fault. It can efficiently reduce the fault failed to report and the fault missed to report.更多还原