摘 要

在如今越来越自动化的现代工业中，对机械设备的运行状态有着越来越高的要求。对机械设备来说，轴承是其重要的组成零部件。轴承在机械设备中起着重要的润滑、支撑等作用，一旦轴承运行状况不良或出现故障，则会导致设备故障、降低生产力甚至人员伤亡。统计表明：在旋转机械设备发生的故障中，大约30%都是因为滚动轴承发生故障而引起的。因此，通过轴承振动测试，采用一定的数据分析技术对其运行状态或出现故障的类型进行分析判断，有着十分重大的意义。

本文介绍了滚动轴承的基本机理以及关于滚动轴承固有振动频率与故障特征频率的公式计算等内容，主要研究了基于振动信号的故障诊断中应用较多的两种数据分析方法，即经验模态分解（EMD）以及共振解调技术。

通过对两种方法进行大量的调查研究可知，传统的共振解调技术虽然有着强大的特征信号提取的能力，但却不能自适应地筛选出共振频带，并且带通滤波器参数的选择通常取决于历史数据和具体使用者的经验，这使得分析结果具有很大的主观因素，而且对于复杂的振动信号，分解出的特征信号会很多，又由于故障信号一般来说引起的波动都十分微小，会被掩盖，所以很难提取需要的特征信号；经验模态分解并不像共振解调技术一样，有着强大的特征信号提取的能力，导致单纯的经验模态分解难以观察到明显的故障特征信号。本文提出了基于EMD和共振解调的数据分析技术，将两种方法结合起来，利用EMD分解的自适应性弥补了共振解调技术需要人为选择滤波参数的尴尬之处，并且通过分解的本征模态函数（IMF）分量，分选出重要的特征成分，从而顺利的提取数据信息。利用共振解调技术能提取调制高频固有振动中故障信息的能力，弥补了 EMD无法突显故障特征的缺陷。

对所提出的方法进行软硬件设计，硬件部分实现轴承的振动测试并进行Proteus仿真，软件部分编制了基于EMD和共振解调的仿真模型并进行LabVIEW仿真，对轴承振动信号进行EMD分解，利用Hilbert变换实现共振解调对滚动轴承故障边际谱的分析。

通过实际轴承振动测试实验研究表明，基于EMD和共振解调的数据分析技术能够准确可靠的判断滚动轴承的故障类型，同时，也能判断轴承状态的好坏，体现了该方法的实效性。

关键词：轴承；振动测试；故障诊断；EMD；共振解调技术

Abstract

In today's more and more automated modern industry, the running state of machinery and equipment has higher and higher requirements. For mechanical equipment, bearing is an important component. Bearing plays an important role in lubrication and support in mechanical equipment. Once the bearing is in poor operation or failure, it will lead to equipment failure, reduce productivity and even casualties. Statistics show that about 30% of the faults of rotating machinery and equipment are caused by the failure of rolling bearings. Therefore, through bearing vibration test, a certain data analysis technology is used to analyze and judge the running state or the type of fault. Breaking is of great significance.

This paper introduces the basic mechanism of rolling bearing and the formula calculation of natural vibration frequency and fault characteristic frequency of rolling bearing, and mainly studies two kinds of data analysis methods which are widely used in fault diagnosis based on vibration signal. Empirical mode decomposition (EMD) and resonance demodulation technology.

Through a large number of investigation and research on the two methods, it can be seen that although the traditional resonance demodulation technology has a strong ability to extract feature signals, it can not adaptively screen the resonance frequency band. And the selection of band-pass filter parameters usually depends on the historical data and the experience of specific users, which makes the analysis results have great subjective factors, and for complex vibration signals, there will be a lot of decomposed characteristic signals. And because the fluctuation of fault signal is very small, it will be covered up, so it is difficult to extract the required feature signal. Empirical mode decomposition does not have powerful characteristics like resonance demodulation technology. The ability of signal extraction makes it difficult to observe the obvious fault characteristic signal by simple empirical mode decomposition. In this paper, a data analysis technology based on EMD and resonance demodulation is proposed, which combines the two methods and makes use of the adaptability of EMD decomposition to make up for the embarrassment that resonance demodulation technology needs artificial selection of filtering parameters. Through the decomposition of the intrinsic modal function (IMF) component, the important feature components are selected, so as to extract the data information smoothly. Resonance demodulation technology can be used to extract fault information in modulation of high frequency natural vibration, which makes up for the defect that EMD can not highlight the fault characteristics.

The software and hardware of the proposed method are designed, the vibration test of the bearing is realized and the Proteus simulation is carried out in the hardware part. The simulation model based on EMD and resonance demodulation is compiled and the LabVIEW simulation is carried out in the software part, and the vibration signal of the bearing is decomposed by EMD. Hilbert transform is used to realize resonance demodulation to analyze the fault edge spectrum of rolling bearing.

Through the experimental study of bearing vibration test, it is shown that the data analysis technology based on EMD and resonance demodulation can accurately and reliably judge the fault type of rolling bearing, and at the same time, it can also judge the state of bearing, which reflects the effectiveness of this method.

Key words: bearing; vibration test; fault diagnosis; EMD; resonance demodulation technology

目 录

第1章 绪论 1

1.1 研究背景及意义 1

1.2 滚动轴承故障诊断的发展历程与研究现状 1

1.2.1 滚动轴承故障诊断的发展历程 1

1.2.2 滚动轴承故障诊断的研究现状 2

1.3 本文主要研究内容 3

第2章 滚动轴承机理分析 5

2.1 滚动轴承的基本结构 5

2.2 滚动轴承的振动机理 6

2.3 滚动轴承的失效形式 6

2.4 滚动轴承的固有频率与故障特征频率 7

2.4.1 滚动轴承零件固有频率 7

2.4.2 滚动轴承的故障特征频率 8

2.5 本章小结 11

第3章 EMD 和共振解调的原理与数据分析 12

3.1 共振解调技术 12

3.1.1 共振解调基本原理 12

3.1.2 共振解调技术的实现 13

3.1.3 共振解调法仿真分析 14

3.2 经验模态分解法 15

3.2.1 EMD的基本原理 15

3.2.2 瞬时频率 15

3.2.3 本征模态函数 16

3.2.4 EMD分解算法 16

3.2.5 希尔伯特谱分析 18

3.2.6 EMD算法仿真分析 19

3.3 基于 EMD 和共振解调的数据分析技术 21

3.3.1 基于 EMD 和共振解调数据分析技术实现步骤 21

3.3.2 基于 EMD 和共振解调法数据分析技术仿真分析 21

3.4 本章小结 22

第4章 滚动轴承数据分析系统软硬件设计 23

4.1 硬件数据采集系统设计 23

4.1.1 系统硬件设计框图 23

4.1.2 加速度传感器 23

4.1.3 数据采集系统 24

4.1.4 采样频率的设定 25

4.1.5 轴承振动数据采集系统与性能测试 26

4.2 系统软件设计 27

4.2.1 LabVIEW简介 27

4.2.2 EMD的LabVIEW实现 27