【作者】 黎少辉；

【导师】 李意民；

【作者基本信息】 中国矿业大学 ， 流体力学， 2009， 博士

【摘要】 失速颤振是风机中常见的故障之一,同时叶片裂纹又是风机中普遍存在的一种严重安全隐患,所以探明失速颤振机理及尽早检测出裂纹现象的存在对于风机的安全运行具有重要意义。论文对叶片结构与气流脉动的关系、流固耦合互体现特性、以及根据以上两点进行叶片故障诊断等内容进行系统化研究。详细工作如下:为探讨风机失速颤振产生的机理,论文提出从气流脉动和叶片结构之间的关系来寻找失速颤振发生频率的方法。通过对风机叶片的模态分析和叶轮出口气动信号频谱分析发现,一定来流速度范围内气流脉动引起的颤振和某一阶固有频率有关,即气流的脉动频率与叶片固有频率趋于相同,远离失速攻角时气流脉动频率逐步摆脱固有频率的影响,在大攻角下叶片结构仍影响气流脉动频率,但已没有一个明显的频率锁定区域。这为探明叶片失速颤振的机理提供了实验依据,证明叶片结构影响气流脉动频率。为研究风机叶轮系统气固耦合振动互体现的特性,论文利用刻画非线性系统的多特征分析方法,从相空间重构、关联维、近似熵、L-Z复杂度等不同角度对多次重复实验采集的不同工况下叶轮轴向振动信号及叶轮出口气动信号进行分析。实验分析结果证明这两种信号在某些量化特征上具有耦合互体现特征。这为利用气动信号进行叶片故障诊断奠定基础。基于信号的能量主要集中在低频附近,高频信号的能量随频率的增加而迅速衰减的特点以及耦合振子距离越近,越具互体现特性。论文提出了对叶片不同状态下的气动信号进行多带小波分解,与基于二带小波分解的方法互补,检测叶片裂纹故障的方法。首先对多次实验采集的气动信号进行小波分解;然后计算每个频段的能量,形成多维特征向量集;最后由特征向量集训练分类器,利用设计好的分类器对特征向量进行分类检测,识别故障。多次实验证明,该方法能够快速、有效的检测出早期裂纹现象的存在。最后为进一步了解叶片裂纹扩展时气动信号的变化,论文对不同工作状态下的气动信号的小波子带能量进行统计分析。分析结果表明,随着叶片裂纹的扩展,高频带相对于正常叶片的气动信号的同子带能量明显降低,而低频带对应的气动信号同子带能量大幅增强。由此可以说明,叶片的裂纹扩展过程中,低频进一步降低,高频进一步升高,气流脉动频率范围逐步扩大。这为转子叶片裂纹故障的监测与诊断提供了依据。同时裂纹的扩展带来叶片结构刚度逐步降低和气动信号低频部分的逐步降低,进一步说明叶片结构影响气流的脉动频率。更多还原

【Abstract】 The stall flutter is one of the common faults which exists in fans, at the same time, the blade crack is another hidden danger, fan blade crack is also prevalent in a serious security risk while fans are working. So, it is important for fans’safe operation to prove the theory of stall flutter and detect as early as possible the existence of the crack. The paper systematically researches the relationship between blade natural frequency and flow fluctuation frequency, then studies the fluid-structure coupling feature of embodiment, as well as carries on the fault diagnosis of blade according to above two points. The detailed work is as follows:Firstly, for exploring the generation mechanism of fan stall flutter, the method of seeking the stall flutter frequency is put forward according to the relationship between flow fluctuation and blade structure. The method analyses the fan blade’s modal and the spectrum of flow fluctuation signal. The result shows that the flutter caused by flow fluctuation relates with natural frequencies of blade within the scope flow velocity. That is, the frequency of flow fluctuation and the blade natural frequency tend to be the same. The frequency of flow fluctuation will escape the influence of natural frequency when flow field is far way from stall attack angle. The natural frequency of the blade still affects the frequency of separation flow in the large attack angle, but there is not an obvious frequency lock region. The result provides the experimental basis for exploring the theory of fan stall flutter, and proves that the blade structure influent the flow fluctuation frequency.Secondly, a multi-feature analysis method is proposed to study embodiment feature of the fluid-solid coupling. This method analyses the vibration signals of axial impeller and flow fluctuation collected in different working condition using the phase-space reconstruction, correlation dimension, approximate entropy and L-Z complexity. Experimental results prove that two kinds of signals have coupled characteristics of embodiment; it is the fundamental of diagnosing the blades’crack fault using by the flow fluctuation signals.Thirdly, the research is performed to detect the existence of the blade crack. The signal energy mainly concentrates in low-frequency and decreases with the increasing of frequency. Coupling vibrations are more nearer more embodied each other. For detecting the blade crack fault, the method which combines M-band wavelet and 2-band wavelet analysis is proposed.Aerodynamic signals collected on different working conditions are decomposed and reconstructed. Then the energies of every frequency band are calculated as the character vectors. Finally, the classifier is trained to class the character vectors and recognize the fault. The experimental results indicate that this method can recognize the fault of early blade crack effectively.Finally, for finding out the changes of the flow fluctuation signal with the blade crack spreading, the statistical analysis on the wavelet sub-band energy of the flow fluctuation signal collected on the different working condition is carried out. Analysis results show that high frequency band energy reduces and the low frequency band energy increases with the spread of blade cracks. So, the flow fluctuation frequency range gradually expands because of the spread of blade cracks. The results provide the basis for blade crack fault monitoring and diagnosis. At the same time, the reducing of blade stiffness and the low-frequency of flow fluctuation signal with the spread of crack indicate that the separated flow pulse frequency will be consistent with blade natural frequency.更多还原