旋转机械故障诊断与预测方法及其应用研究
发布时间：2019-06-14 03:50
【摘要】：研究旋转机械的故障诊断与预测技术,对于保障机械设备运行的安全性和稳定性具有十分重要的意义。旋转机械的振动信号具有非稳定性和非线性,同时,在强背景噪声工作环境下,旋转机械的微弱故障特征很容易被噪声淹没,此外,当机械系统出现故障时,往往会产生位置不同的复合故障,故障之间相互耦合,从而给旋转机械故障精确诊断带来了挑战,因此,强噪下微弱、复合故障诊断是当今机械故障诊断领域的难点。论文将旋转机械作为研究对象,研究形态学滤波、局域均值分解、多元经验模态分解和噪声辅助多元经验模态分解等时频方法及其在旋转机械的微弱、复合故障诊断中的应用,为机械故障诊断、性能退化状态识别和趋势预测提供新的有效手段。主要内容如下：1、提出了一种基于LMD和形态滤波的轴承故障诊断方法。设计并搭建了铁路货车轮对滚动轴承测试系统,并对轴承典型故障振动信号进行分析,仿真实验与轴承故障试验结果验证了该方法的有效性。针对形态滤波器尺度选择缺乏自适应的问题,提出了基于遗传算法的自适应形态滤波方法,仿真和试验的分析结果表明,自适应形态学滤波器对于信号降噪处理和冲击特征提取两方面均有明显的效果。2、针对EMD和LMD等时频分析方法无法处理旋转机械多通道振动信号的缺点和旋转机械早期微弱故障、复合故障的特征提取问题,提出了基于改进的多元经验模态分解的旋转机械早期故障诊断方法。该方法利用多元经验模态分解将多通道振动信号分解得到一系列多元IMF分量,将峭度准则和互信息引入IMF的选取,进一步消除混入的噪声和伪分量的影响。仿真信号和旋转机械故障信号的分析结果表明,改进的MEMD方法在多通道信号分解的精确性和鲁棒性等方面具有明显的优越性和有效性,为旋转机械微弱故障、复合故障诊断和多通道振动信息融合分析提供了新的思路和手段。3、NAMEMD是一种新的非线性信号自适应时频分解方法,该方法克服了MEMD和EEMD的模态混叠等问题,但是经过研究发现,NAMEMD方法并不能完全抑制MEMD的模态混叠现象,得到的IMF仍存在模态混叠,需要后续处理。为了抑制NAMEMD方法分解中的模态混叠现象,提出了改进的NAMEMD方法。采用基于排列熵的随机性检测技术及时地检测异常信号和噪声信号,再对剩余信号进行NAMEMD分解,通过仿真信号验证了所提出方法的有效性,在此基础上,针对强噪下机械故障特征提取的问题,提出了基于改进的NAMEMD形态学与Teager能量算子解调的旋转机械故障诊断方法,并通过仿真信号和旋转机械故障信号将所提出的方法与EEMD和NAMEMD进行了对比,结果表明改进的NAMEMD方法消除了EEMD集成平均过程中因添加白噪声的时频特性差异带来的模态混叠,分解结果相对于EEMD具有较准确的IMF频谱分布和更好的降噪效果,分解结果更为精确。所提方法在抑制模态混叠、增强降噪效果和提高分解精确性上要优于EEMD和NAMEMD方法,结果验证了所提方法的有效性和优越性。4、在分析样本熵和排列熵原理的基础上,针对轴承振动信号的非线性特征,提出了基于NAMEMD和排列熵的轴承故障智能诊断方法。首先对振动信号进行NAMEMD分解,然后对前5个有意义的IMF分量进行排列熵计算,并将其作为特征向量输入训练好的SVM分类器,有效地实现轴承四种典型状态类型的识别,准确率高。5、将NAMEMD自适应分解与基于非线性动力学参数的信号复杂性的排列熵理论相结合,提出了基于改进NAMEMD和排列熵的旋转机械退化状态检测方法,该方法首先将多分量的振动信号自适应地分解得到一系列信噪比较高的IMF分量,利用对突变信号敏感的排列熵算法分别对各IMF进行排列熵分析,进行轴承运行状态及演化过程的准确识别。建立了滚动轴承振动信号和退化状态之间的联系。通过仿真试验和滚动轴承全寿命试验数据,证明了建立的状态指标能够准确、完整地反映滚动轴承的退化状态趋势,实现了滚动轴承全寿命周期状态的有效识别。所提方法具有较强的鲁棒性,为机械设备的性能退化状态检测提供了一种新的有效途径。6、针对滚动轴承退化状态趋势预测问题,提出了基于NAMEMD、PE和SVR的滚动轴承故障演化状态趋势预测模型,实现滚动轴承性能退化趋势的准确预测,评估在未来一段时间内的轴承状态的变化趋势,从而达到加强机械设备运行安全性与稳定性的目的。通过轴承全寿命试验,证明了所提方法的准确性和有效性,具有较高的预测精度和鲁棒性,对工程实践具有重要的指导意义。
[Abstract]:The research of the fault diagnosis and prediction of the rotating machinery is of great significance to the safety and stability of the operation of the mechanical equipment. the vibration signal of the rotating machine has the non-stability and the non-linearity, and at the same time, under the working environment of strong background noise, the weak fault characteristic of the rotating machine is easy to be flooded by noise, The mutual coupling between faults brings the challenge to the accurate diagnosis of the rotary mechanical failure. Therefore, the weak and complex fault diagnosis in the field of mechanical fault diagnosis is a difficult problem in the field of mechanical fault diagnosis. As a research object, the paper studies the time-frequency methods such as morphological filtering, local mean decomposition, multi-element empirical mode decomposition and noise-assisted multi-element empirical mode decomposition, and its application in the weak and complex fault diagnosis of rotating machinery, and it is a mechanical fault diagnosis. Performance degradation state identification and trend prediction provide new and effective means. The main content is as follows:1. A method of bearing fault diagnosis based on LMD and morphological filtering is presented. The rolling bearing test system of the railway wagon wheel is designed and constructed, and the typical fault vibration signal of the bearing is analyzed, and the simulation experiment and the bearing failure test result verify the effectiveness of the method. A self-adaptive morphological filtering method based on genetic algorithm is proposed for morphological filter scale selection. The results of simulation and experiment show that the adaptive morphological filter has obvious effect on signal noise reduction and impact feature extraction. In order to solve the shortcomings of the multi-channel vibration signal of the rotating machinery and the weak fault of the rotating machinery in the time-frequency analysis method such as the EMD and the LMD, the problem of the feature extraction of the composite fault is solved, and the early fault diagnosis method of the rotating machinery based on the improved multi-element empirical mode decomposition is proposed. According to the method, the multi-channel vibration signal is decomposed to obtain a series of multi-element IMF components by using the multi-element empirical mode decomposition, and the similarity criterion and the mutual information are introduced into the selection of the IMF, and the influence of the mixed noise and the pseudo component is further eliminated. The result of the analysis of the simulation signal and the rotating mechanical failure signal shows that the improved MEMD method has obvious advantages and effectiveness in the aspects of the accuracy and the robustness of the multi-channel signal decomposition, and is a weak fault of the rotating machinery, The composite fault diagnosis and the multi-channel vibration information fusion analysis provide a new idea and means.3. The NEMEMD is a new method of self-adaptive time-frequency decomposition of nonlinear signals, which overcomes the problems of the mode aliasing of the EMD and the EEMD, but has been found by the research, The NEMEMD method can not completely suppress the mode aliasing of the MEMD, and the obtained IMF still has the mode aliasing, and the subsequent processing is required. In order to suppress the mode aliasing in the decomposition of the NAMEMD method, an improved NEMEMD method is proposed. By adopting the random detection technology based on the arrangement entropy, the abnormal signal and the noise signal are detected in time, the residual signal is subjected to NAMOEMD decomposition, the validity of the proposed method is verified through the simulation signal, on the basis, the problem of the feature extraction of the mechanical failure under the strong noise is solved, A rotary mechanical fault diagnosis method based on improved NEMEMD morphology and Teager energy operator demodulation is proposed, and the proposed method is compared with the EEMD and the NAEMEMD by means of the simulation signal and the rotating mechanical failure signal. The results show that the improved NEMEMD method eliminates the mode aliasing caused by the difference of the time-frequency characteristics of the addition of white noise in the EEMD integration averaging process, and the decomposition result has a more accurate IMF spectral distribution and better noise reduction effect with respect to the EEMD, and the decomposition result is more accurate. The proposed method is superior to the EEMD and NEMEMD method in suppressing the mode aliasing, enhancing the noise reduction effect and improving the decomposition accuracy, and the validity and the superiority of the proposed method are verified. An intelligent diagnosis method for bearing failure based on NAEMEMD and permutation entropy is presented. The method comprises the following steps of: firstly, performing NAMOEMD decomposition on a vibration signal, and then arranging and entropy calculating the first five meaningful IMF components, and using the SVM classifier as a feature vector to input a trained SVM classifier, so that the identification of four typical state types of the bearing is effectively realized, and the accuracy is high. Based on the combination of the adaptive decomposition of the NAEMEMD and the arrangement entropy theory of the signal complexity based on the nonlinear dynamic parameters, a method for detecting the degradation state of the rotating machinery based on the improved NEMEMD and the arrangement entropy is proposed. The method comprises the following steps of: firstly, adaptively decomposing a multi-component vibration signal to obtain a series of IMF components with higher signal-to-noise ratio, and arranging and entropy analyzing the IMF according to an arrangement entropy algorithm which is sensitive to the abrupt signal, and carrying out accurate identification of the running state and the evolution process of the bearing. The relationship between the vibration signal and the degraded state of the rolling bearing is established. Through the simulation test and the whole life test data of the rolling bearing, it is proved that the established state index can accurately and completely reflect the degradation state tendency of the rolling bearing and realize the effective identification of the whole life cycle state of the rolling bearing. The proposed method has strong robustness, and provides a new effective method for the performance degradation state detection of mechanical equipment.6. Aiming at the problem of the trend prediction of the degradation state of the rolling bearing, the fault evolution state trend prediction model of the rolling bearing based on the NEMEMD, PE and SVR is put forward. The invention realizes the accurate prediction of the performance degradation trend of the rolling bearing, and evaluates the change tendency of the bearing state over a period of time, so as to achieve the purpose of strengthening the operation safety and the stability of the mechanical equipment. Through the full life test of the bearing, the accuracy and the effectiveness of the proposed method are proved, and the method has higher prediction accuracy and robustness, and has important guiding significance for engineering practice.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TH17
本文编号：2499074
[Abstract]:The research of the fault diagnosis and prediction of the rotating machinery is of great significance to the safety and stability of the operation of the mechanical equipment. the vibration signal of the rotating machine has the non-stability and the non-linearity, and at the same time, under the working environment of strong background noise, the weak fault characteristic of the rotating machine is easy to be flooded by noise, The mutual coupling between faults brings the challenge to the accurate diagnosis of the rotary mechanical failure. Therefore, the weak and complex fault diagnosis in the field of mechanical fault diagnosis is a difficult problem in the field of mechanical fault diagnosis. As a research object, the paper studies the time-frequency methods such as morphological filtering, local mean decomposition, multi-element empirical mode decomposition and noise-assisted multi-element empirical mode decomposition, and its application in the weak and complex fault diagnosis of rotating machinery, and it is a mechanical fault diagnosis. Performance degradation state identification and trend prediction provide new and effective means. The main content is as follows:1. A method of bearing fault diagnosis based on LMD and morphological filtering is presented. The rolling bearing test system of the railway wagon wheel is designed and constructed, and the typical fault vibration signal of the bearing is analyzed, and the simulation experiment and the bearing failure test result verify the effectiveness of the method. A self-adaptive morphological filtering method based on genetic algorithm is proposed for morphological filter scale selection. The results of simulation and experiment show that the adaptive morphological filter has obvious effect on signal noise reduction and impact feature extraction. In order to solve the shortcomings of the multi-channel vibration signal of the rotating machinery and the weak fault of the rotating machinery in the time-frequency analysis method such as the EMD and the LMD, the problem of the feature extraction of the composite fault is solved, and the early fault diagnosis method of the rotating machinery based on the improved multi-element empirical mode decomposition is proposed. According to the method, the multi-channel vibration signal is decomposed to obtain a series of multi-element IMF components by using the multi-element empirical mode decomposition, and the similarity criterion and the mutual information are introduced into the selection of the IMF, and the influence of the mixed noise and the pseudo component is further eliminated. The result of the analysis of the simulation signal and the rotating mechanical failure signal shows that the improved MEMD method has obvious advantages and effectiveness in the aspects of the accuracy and the robustness of the multi-channel signal decomposition, and is a weak fault of the rotating machinery, The composite fault diagnosis and the multi-channel vibration information fusion analysis provide a new idea and means.3. The NEMEMD is a new method of self-adaptive time-frequency decomposition of nonlinear signals, which overcomes the problems of the mode aliasing of the EMD and the EEMD, but has been found by the research, The NEMEMD method can not completely suppress the mode aliasing of the MEMD, and the obtained IMF still has the mode aliasing, and the subsequent processing is required. In order to suppress the mode aliasing in the decomposition of the NAMEMD method, an improved NEMEMD method is proposed. By adopting the random detection technology based on the arrangement entropy, the abnormal signal and the noise signal are detected in time, the residual signal is subjected to NAMOEMD decomposition, the validity of the proposed method is verified through the simulation signal, on the basis, the problem of the feature extraction of the mechanical failure under the strong noise is solved, A rotary mechanical fault diagnosis method based on improved NEMEMD morphology and Teager energy operator demodulation is proposed, and the proposed method is compared with the EEMD and the NAEMEMD by means of the simulation signal and the rotating mechanical failure signal. The results show that the improved NEMEMD method eliminates the mode aliasing caused by the difference of the time-frequency characteristics of the addition of white noise in the EEMD integration averaging process, and the decomposition result has a more accurate IMF spectral distribution and better noise reduction effect with respect to the EEMD, and the decomposition result is more accurate. The proposed method is superior to the EEMD and NEMEMD method in suppressing the mode aliasing, enhancing the noise reduction effect and improving the decomposition accuracy, and the validity and the superiority of the proposed method are verified. An intelligent diagnosis method for bearing failure based on NAEMEMD and permutation entropy is presented. The method comprises the following steps of: firstly, performing NAMOEMD decomposition on a vibration signal, and then arranging and entropy calculating the first five meaningful IMF components, and using the SVM classifier as a feature vector to input a trained SVM classifier, so that the identification of four typical state types of the bearing is effectively realized, and the accuracy is high. Based on the combination of the adaptive decomposition of the NAEMEMD and the arrangement entropy theory of the signal complexity based on the nonlinear dynamic parameters, a method for detecting the degradation state of the rotating machinery based on the improved NEMEMD and the arrangement entropy is proposed. The method comprises the following steps of: firstly, adaptively decomposing a multi-component vibration signal to obtain a series of IMF components with higher signal-to-noise ratio, and arranging and entropy analyzing the IMF according to an arrangement entropy algorithm which is sensitive to the abrupt signal, and carrying out accurate identification of the running state and the evolution process of the bearing. The relationship between the vibration signal and the degraded state of the rolling bearing is established. Through the simulation test and the whole life test data of the rolling bearing, it is proved that the established state index can accurately and completely reflect the degradation state tendency of the rolling bearing and realize the effective identification of the whole life cycle state of the rolling bearing. The proposed method has strong robustness, and provides a new effective method for the performance degradation state detection of mechanical equipment.6. Aiming at the problem of the trend prediction of the degradation state of the rolling bearing, the fault evolution state trend prediction model of the rolling bearing based on the NEMEMD, PE and SVR is put forward. The invention realizes the accurate prediction of the performance degradation trend of the rolling bearing, and evaluates the change tendency of the bearing state over a period of time, so as to achieve the purpose of strengthening the operation safety and the stability of the mechanical equipment. Through the full life test of the bearing, the accuracy and the effectiveness of the proposed method are proved, and the method has higher prediction accuracy and robustness, and has important guiding significance for engineering practice.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TH17
【引证文献】
相关博士学位论文 前2条
1 王宝祥;基于运动形态分解与多变量EMD的高速自动机动态监测与故障诊断研究[D];中北大学;2018年
2 彭延峰;自适应最稀疏时频方法及其在旋转机械故障诊断中的应用[D];湖南大学;2017年
相关硕士学位论文 前10条
1 马江飞;基于FPGA的齿轮箱故障诊断方法研究[D];西安理工大学;2018年
2 顾婉莹;基于激光自混合干涉的速度测量及其应用[D];东北石油大学;2018年
3 刘小勇;基于深度学习的机械设备退化状态建模及剩余寿命预测研究[D];哈尔滨工业大学;2018年
4 于小娟;基于EEMD的谷物测产信号去噪处理方法研究[D];东北农业大学;2018年
5 叶绪丹;基于变分模态分解的滚动轴承早期微弱故障诊断研究[D];安徽工业大学;2018年
6 陈博;高速列车车轮多边形的检测与识别方法研究[D];西南交通大学;2018年
7 卓仁雄;基于CEEMDAN和GWO-SVM的电机滚动轴承故障诊断[D];南华大学;2018年
8 萨仁朝格图;智能装备机械故障物联网监测诊断服务平台[D];大连理工大学;2018年
9 曾柯;基于SPSO优化TWSVM及Bayesian更新指数模型的轴承剩余寿命预测[D];重庆大学;2018年
10 徐国权;基于多特征的机车轴承振动故障诊断[D];北京信息科技大学;2018年
，本文编号：2499074
本文链接：https://www.wllwen.com/jixiegongchenglunwen/2499074.html
