【作者】 王晓伟；

【导师】 刘占生；

【作者基本信息】 哈尔滨工业大学 ， 动力机械及工程， 2009， 博士

【摘要】 动静部件碰摩是旋转机械中常见典型故障,其对旋转机械的安全稳定运行有着很大的危害。因此,对碰摩故障进行及时有效地识别和诊断对于保证旋转机械安全运行有着重要意义。为了对旋转机械碰摩接触产生的声发射信号有更好的理解,以实现对碰摩故障的有效识别,有必要对碰摩产生声发射信号的机理进行分析。本文分析了旋转机械动静碰摩的过程,指出动静碰摩接触的实质为多微凸体接触;基于GW接触模型,建立了声发射信号均方根值与碰摩接触力之间关系的理论模型,从该模型能够看出测量声发射信号均方根值的概率分布与接触表面微凸体的概率分布相同;通过进行全周碰摩和局部碰摩实验研究,证明声发射信号能量在转子结构中传播是线性衰减的。对于早期的动静碰摩故障,其产生的声发射信号与背景噪声相比往往很弱,因此需要对声发射信号的降噪技术进行研究。根据声发射信号的特点,提出运用自适应阈值方法对连续小波变换的系数进行的消噪,采用逆连续小波变换对消噪后的小波变换系数进行重构,研究发现该方法能够有效的去除声发射信号中的背景噪声,保留信号中的弱冲击成分。对转子结构分析的基础上,将碰摩故障的定位归为声发射源的线性定位问题,通过研究指出碰摩故障定位的关键技术是确定声发射波到达传感器的时间和声发射波传播速度的确定。针对突发型声发射信号的特点,提出采用AR模型确定声发射波的到达时间。针对连续型的声发射信号,通过带通滤波提取信号的特征频率,采用互相关技术确定其时间差,并对其传播速度进行识别。为了对声发射源信号进行有效的识别,更好的对其进行分类,将盲目反卷积技术用于声发射信号的处理,从检测到的声发射信号中消除信号在传播路径中受到的干扰,以提取真实的声发射源信号。在600MW超超临界汽轮发电机组轴系模拟实验台上进行了密封瓦与密封盘碰摩、径向可倾瓦滑动轴承碰摩实验。实验结果表明本文中提出的对碰摩声发射信号的处理方法能够对含强背景噪声的检测信号进行处理,有效的降低噪声。可倾瓦径向滑动轴承故障检测实验表明:声发射信号能够诊断出可倾瓦轴承的瓦块与轴的碰摩,通过合理布置声发射传感器在轴承座上的位置,根据声发射信号到达多个传感器的时间、声发射信号的均方根值以及轴承的结构可以推断出与轴发生碰摩的瓦块位置,为碰摩故障的早期诊断,提供了一种有效手段。更多还原

【Abstract】 The rubbing and impact between the static and dynamic components is the common typical malfunction of rotating machinery, which is great harmful to the safe and steady running. So identifying and diagnosing the rubbing and impact malfunction have heavy significance on the warranting the safe running of the rotating machinery.It is necessary to analyze the mechanism of acoustic emission signals generated from the rubbing and impact, which will make better understand of acoustic emission signal and carry out the effective identifying of rubbing impact malfunction. By analyzing the process of static and dynamic rubbing impact, the principle of static and dynamic rubbing impact is the contact of multi-micro asperity, which is proposed in this paper; based on the GW contact model, the theoretical model of acoustic emission signals which have the relationship between the rubbing impact contact forces is established. By analyzing this model, it indicates that the distribution of measured acoustic emission signals is the same as the distribution of contact micro asperity. From the full circular and local rubbing impact experimental research, it is verified that the propagation of acoustic emission energy in rotor structure is the linear attenuation.It is necessary to study the noise reduction techniques for acoustic emission signals because that the acoustic emission signal of early static dynamic rubbing impact malfunction is weak compared with background noise. So based on the characteristic of acoustic emission signals, the self adapting threshold method is employed to reduce the noise of continuous wavelet transformation coefficients and then the coefficients will be rebuilt by using inverse continuous wavelet transformation. The results indicate that this method can eliminate the noise of the signals effectively and keep the weak impulse components.By analyzing the structure of the rotor, the rubbing impact locating problem is the case of linear location for acoustic emission source. It indicates that the key techniques of allocating the malfunction position are how to obtain the arriving time and the velocity of the wave. Then the AR model is employed to get the arriving time of acoustic emission wave based on the characteristic of burst acoustic emission signals. For successive acoustic emission signals, the frequency characteristic is obtained by applying the bandpass filtration and the time delay is determined by employing the cross correlation technique to extract signals, also the spreading velocity will be identified.The blind deconvolution technique is applied to the treat the acoustic emission signals, which can identify the signals effectively and classify the signals better. This method of analyzing the effect of signals attenuating on the transmission path from detected acoustic emission signals can extract the true acoustic emission source signals possibly.The rubbing impact between the sealing pads and sealing disks experiments, the rubbing impact of journal tilting pad bearing experiments are carried out on the 600MW super-critical steam turbine-generator simulated test rig. The results indicate that the concerned method from the research above can treat the detected signals which have strong background noise and reduce the noise. The results of tilting pad bearing malfunction experiments indicate that by arranging the acoustic emission sensors on the housing reasonably, and employing the arriving time of different acoustic emissions sensors and the mean square root of acoustic emission signals, the rubbing impact position of the pads can be identified, which provides the effective way for the early diagnose of rubbing impact malfunction.更多还原