基于振动分析的滚动轴承早期故障诊断研究

发布时间：2019-04-12 10:02

【摘要】：滚动轴承是传动机械的核心部件，其运行状态直接影响到整台设备的精度、可靠性及寿命等性能。由于其本身的结构特点及工作环境等因素，滚动轴承极易出现故障。轴承故障的特征向量和识别模式之间呈复杂的非线性关系，在轴承早期微弱故障和复合故障的定量诊断与预示中，如何从非平稳、非线性振动信号中提取有效的故障信息就成了关键，对这一问题进行研究在机械故障诊断中具有重要的理论及现实意义。论文主要研究内容如下：首先，在对滚动轴承故障机理和故障形式及成因进行全面分析的基础上，模拟滚动轴承主要故障，通过滚动轴承振动检测与诊断试验系统实现对正常和故障状态下的振动信号的采集，并对所得到信号进行时域参数特征统计和时频域处理，以分析滚动轴承不同状态下的振动特性。其次，研究了基于随机共振的滚动轴承早期故障识别方法，分析了单稳随机共振模型下的变尺度级联效应，通过正常状态以及外圈早期故障的仿真和实测数据，验证了随机共振在抑制轴承背景噪声、早期故障特征提取方面的可行性和实用性。再次，提出了随机共振（SR）消噪下的总体平均经验模式分解（EEMD）的滚动轴承特征提取方法，探讨了EEMD方法在自适应分解、抗模式混叠方面的优势，并结合包络解调技术，将其成功应用于滚动轴承早期单点故障及耦合故障的特征提取。最后，在SR-EEMD方法所构建故障特征向量的基础上，利用BP和RBF两种神经网络模型分别对滚动轴承状态样本集进行训练和预测，再通过遗传算法对RBF网络进行参数优化，提高了网络性能。
[Abstract]:Rolling bearing is the core component of transmission machinery, and its running state directly affects the precision, reliability and life of the whole equipment. Because of its structural characteristics and working environment, rolling bearings are prone to fault. There is a complex nonlinear relationship between the characteristic vector and the recognition pattern of bearing fault. In the quantitative diagnosis and prediction of weak and compound faults in the early stage of bearing, how to solve the problem from non-stationary to non-stationary? It is very important to extract effective fault information from nonlinear vibration signals. The research on this problem is of great theoretical and practical significance in mechanical fault diagnosis. The main contents of this paper are as follows: firstly, the main faults of rolling bearings are simulated on the basis of a comprehensive analysis of the fault mechanism, fault form and cause of failure. Through the rolling bearing vibration detection and diagnosis test system, the vibration signals under normal and fault conditions are collected, and the time-domain parameter characteristic statistics and time-frequency domain processing of the obtained signals are carried out. In order to analyze the vibration characteristics of rolling bearings under different conditions. Secondly, the early fault identification method of rolling bearing based on stochastic resonance is studied, and the variable scale cascade effect under the monostable stochastic resonance model is analyzed. The simulation and measured data of the normal state and the early fault of the outer ring are carried out. The feasibility and practicability of stochastic resonance in suppressing bearing background noise and extracting early fault features are verified. Thirdly, the general average empirical mode decomposition (EEMD) method of rolling bearing feature extraction based on stochastic resonance (SR) de-noising is proposed, and the advantages of EEMD method in adaptive decomposition and anti-mode mixing are discussed, and the envelope demodulation technique is combined with the method of self-adaptive decomposition and anti-mode aliasing. It is successfully applied to feature extraction of early single point fault and coupling fault of rolling bearing. Finally, based on the fault eigenvector constructed by SR-EEMD method, two neural network models, BP and RBF, are used to train and predict the sample set of rolling bearing state, and then the parameters of RBF network are optimized by genetic algorithm. Improved network performance.
【学位授予单位】：中国计量学院
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TH133.33;TH165.3

【参考文献】