瞬态成分建模与参数辨识方法及其旋转机械故障诊断应用研究

发布时间：2018-05-06 17:53

本文选题：故障诊断 + 瞬态成分建模　；参考：《苏州大学》2011年硕士论文

【摘要】：机械系统中轴承、齿轮等零部件出现剥落、裂纹等局部故障时,在运行时振动信号中出现瞬态冲击响应成分,而且随着损伤程度的发展,其特征波形也会发生变化。振动信号中反映零部件局部故障的瞬态冲击响应成分提取是机械设备故障诊断的关键问题之一。本论文依托于“变工况下旋转设备轻微局部故障的特征增强检测与诊断方法研究”国家自然科学基金青年基金项目(批准号:50905121)和“基于瞬态振动特征辨识的轴承局部故障定量诊断研究”江苏省自然科学基金项目(批准号:BK2010225),以旋转机械设备故障诊断为目标,以旋转机械中滚动轴承和齿轮两种典型零部件为对象,提出了瞬态成分建模与参数辨识的冲击响应成分提取方法,并分别就有关问题进行了深入的理论研究和应用研究。首先分析了轴承和齿轮在局部故障时振动信号的特点,并进行了轴承和齿轮在局部故障下的振动试验和振动信号采集,保证理论研究工作都建立在试验验证的基础上。试验对象包括滚动轴承和齿轮这两种典型旋转机械零部件,其中滚动轴承包括圆柱滚子轴承和深沟球轴承。根据旋转机械零部件故障导致的瞬态冲击响应成分特点,在Laplace小波相关滤波法的基础上,将原有的Laplace小波建模扩展为多种小波参数化建模,提出了基于最大相关系数准则的瞬态成分建模与参数辨识方法。模型包括单瞬态成分模型和周期性多瞬态成分模型,因此方法除提取瞬态冲击响应成分的自身参数之外,还可提取反映故障类型或者故障位置的周期参数。对仿真信号以及轴承和齿轮实测振动信号的应用验证了该方法对轴承和齿轮故障特征提取的有效性和适用性。针对变工况下机械设备故障诊断问题,提出基于最小二乘法的瞬态成分建模与参数辨识方法。方法中,双边不对称参数化建模,能用于分析多种机械零部件故障诊断分析;迭代提取瞬态成分,并运用Wigner-Ville分布表示在时频平面上,既保证了较高的时频聚集性,又避免了瞬态成分之间的交叉项。本文基于瞬态冲击响应成分建模与参数辨识方法研究,确定了两种基于参数辨识的旋转机械设备关键部件在局部故障下的状态特征提取的有效方法,对旋转机械故障诊断具有一定的理论意义和实践价值。
[Abstract]:When some parts such as bearing, gear and other parts in mechanical system appear exfoliation, crack and other local faults, the transient shock response component appears in the vibration signal during operation, and its characteristic waveform will change with the development of damage degree. It is one of the key problems in fault diagnosis of mechanical equipment to extract the transient impulse response components from vibration signals which reflect the local faults of components. This paper is based on "Research on feature Enhancement Detection and diagnosis method of minor Local Fault of rotating equipment under variable working conditions" (Grant No.: 50905121) and "bearing based on transient Vibration feature Identification" of National Natural Science Foundation of China (Grant No. 50905121) Research on quantitative diagnosis of partial faults "Jiangsu Natural Science Foundation Project (Grant No.: BK2010225m), aimed at fault diagnosis of rotating machinery and equipment," Taking two typical parts of rolling bearing and gear in rotating machinery as an example, a method of extracting impulse response components for transient component modeling and parameter identification is proposed, and the related problems are studied in theory and application respectively. Firstly, the characteristics of vibration signal of bearing and gear in local fault are analyzed, and the vibration test and vibration signal collection of bearing and gear under local fault are carried out to ensure that the theoretical research work is based on experimental verification. The experimental objects include two kinds of typical parts of rotary machinery, such as roller bearing and gear, including cylindrical roller bearing and deep groove ball bearing. According to the characteristics of the transient impulse response caused by the fault of rotating machinery parts, the original Laplace wavelet modeling is extended to a variety of wavelet parametric modeling based on the Laplace wavelet correlation filtering method. A method of transient component modeling and parameter identification based on maximum correlation coefficient criterion is proposed. The model consists of a single transient component model and a periodic multiple transient component model. Therefore, in addition to extracting the parameters of the transient impulse response component, the periodic parameters reflecting the fault type or fault location can also be extracted. The validity and applicability of this method to the fault feature extraction of bearing and gear are verified by the application of simulation signal and vibration signal of bearing and gear. A method of transient component modeling and parameter identification based on least square method is proposed to solve the problem of mechanical equipment fault diagnosis under variable working conditions. In the method, the bilateral asymmetric parametric modeling can be used to analyze the fault diagnosis and analysis of various mechanical parts, to extract the transient components iteratively, and to use the Wigner-Ville distribution to represent the transient components in the time-frequency plane, which ensures the high time-frequency aggregation. In addition, the crossover between transient components is avoided. Based on the research of transient shock response component modeling and parameter identification, two effective methods for extracting the state characteristics of the key components of rotating machinery under local faults are established in this paper. It has certain theoretical significance and practical value for rotating machinery fault diagnosis.
【学位授予单位】：苏州大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【相似文献】