完备及欠定条件下盲分离在故障诊断中的应用研究

发布时间：2018-04-18 00:08

本文选题：独立分量分析 + 稀疏分量分析　；参考：《昆明理工大学》2011年硕士论文

【摘要】：在旋转机械振动状态监测与故障诊断过程中,常常面临着各种干扰或者各种故障相互耦合的情况,研究如何从复杂的监测信号当中分离提取故障特征具有重要意义。本文以旋转机械振动信号为研究对象,以盲源分离(BSS)为研究方法,系统地研究了三种盲源分离技术及其在机械故障诊断领域中的应用,重点针对滚动轴承复合故障分离问题,结合形态滤波技术研究了独立分量分析(ICA)及稀疏分量分析(SCA)在实际滚动轴承复合故障当中的分离方法。本文主要研究内容如下： (1)以机械设备故障诊断行业为背景,对盲源分离方法中的三种典型分析方法-独立分量分析、稀疏分量分析以及非负矩阵分解(NMF)的相关研究现状进行了简要综述,并总结了盲分离方法在机械故障诊断领域的研究应用情况。 (2)在研究独立分量分析相关理论基础之上,针对常用梯度算法易使优化问题陷入局部最优,且步长选择对算法速度影响较大的问题,将基于微粒群优化的独立分量分析应用到旋转机械转子复合故障诊断当中。 (3)针对滚动轴承复合故障各个故障相互耦合,振动信号成分复杂,常用盲源分离方法不能有效分离各故障特征的问题,以实验分析的方法探讨了滚动轴承复合故障的耦合机制,并结合滚动轴承故障振动信号所呈现的非平稳、高频调制等特性,提出一种基于形态滤波的独立分量分析方法,并将其应用于滚动轴承复合故障信号分离中。 (4)针对多数盲信号处理方法在分离信号的过程中,要求观测信号数目大于等于源信号数目的问题,将欠定盲分离处理方法中应用较为成功的稀疏分量分析(SCA)引用到滚动轴承复合故障诊断当中。针对轴承故障信号难以满足稀疏分量分析所要求的信号需要呈稀疏分布的问题,提出一种基于形态滤波的稀疏分量分析方法,并分别在完备与欠定的条件下对滚动轴承复合故障进行了实验研究。 (5)研究了非负矩阵分解及其在盲分离中的应用,并尝试将其应用于滚动轴承故障诊断当中。 (6)在理论与实验研究的基础上,利用MATLAB设计实现了三种盲分离技术在旋转机械复合故障诊断中的应用系统。并分别采用仿真与实验的方法对系统进行了验证。
[Abstract]:In the process of vibration state monitoring and fault diagnosis of rotating machinery, it is often faced with various disturbances or mutual coupling of various faults. It is of great significance to study how to separate and extract fault features from complex monitoring signals.In this paper, three blind source separation techniques and their applications in the field of mechanical fault diagnosis are systematically studied, with the vibration signal of rotating machinery as the research object and BSS as the research method. The emphasis is on the composite fault separation of rolling bearings.The separation methods of independent component analysis (ICA) and sparse component analysis (SCA) in the composite fault of rolling bearing are studied by means of morphological filtering technique.The main contents of this paper are as follows:1) based on the background of mechanical equipment fault diagnosis industry, this paper briefly reviews the research status of three typical analysis methods in blind source separation: independent component analysis (ICA), sparse component analysis (SAA) and nonnegative matrix factorization (NMFs).The research and application of blind separation in the field of mechanical fault diagnosis are summarized.2) based on the research of the theory of independent component analysis (ICA), aiming at the problem that the common gradient algorithm is easy to make the optimization problem fall into the local optimum, and the selection of step size has a great influence on the speed of the algorithm,The independent component analysis (ICA) based on particle swarm optimization (PSO) is applied to rotor complex fault diagnosis of rotating machinery.In view of the problems that each fault of rolling bearing composite fault is coupled with each other, the vibration signal component is complex, and the common blind source separation method can not effectively separate the fault characteristics, the coupling mechanism of the rolling bearing composite fault is discussed by the method of experimental analysis.Based on the characteristics of non-stationary and high-frequency modulation of rolling bearing fault vibration signal, an independent component analysis method based on morphological filtering is proposed and applied to the separation of rolling bearing composite fault signal.Aiming at the problem that most blind signal processing methods require the number of observed signals to be greater than or equal to the number of source signals in the process of separating signals,The sparse component analysis (SCA), which has been successfully applied in the under-determined blind separation method, is applied to the composite fault diagnosis of rolling bearings.Aiming at the problem that the bearing fault signal is difficult to meet the need of sparse component analysis (SAA), a sparse component analysis method based on morphological filtering is proposed.The complex fault of rolling bearing is studied experimentally under the condition of complete and underdetermined.In this paper, the nonnegative matrix decomposition and its application in blind separation are studied and applied to the fault diagnosis of rolling bearings.On the basis of theoretical and experimental research, three blind separation techniques are designed and implemented by using MATLAB in complex fault diagnosis system of rotating machinery.Simulation and experiment are used to verify the system.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3;TN911.7

【参考文献】