基于随机共振的齿轮系统故障微弱信号提取技术研究

发布时间：2019-03-12 19:36

【摘要】：齿轮系统在机械工程及其他很多领域的应用范围都十分广泛。齿轮系统中所包含的零部件在发生故障时，在齿轮系统中会有周期性的脉冲冲击力产生，，表现在频谱上即为出现相应的故障信号频谱特征，例如齿轮发生磨损、点蚀，轴发生轻度弯曲或者裂纹等故障时都会在频谱图上产生相应的特征信号。如果能在齿轮系统的故障早期从微弱的特征信号中提取故障调制信息，分析其强度和频次从而判断出零部件损伤的程度和部位，就可以较好的达到机械系统早期故障诊断的目的。随机共振是一种将混合信号输入非线性系统后，通过非线性系统使噪声的部分能量转化给信号的方法，而以往的一些信号检测方法均为通过抑制噪声来达到检测信号的目的，很明显，在抑制噪声的同时也必然导致信号本身的能量受到抑制和影响，所以随机共振方法与传统的信号检测方法相比，其优点就在于，当噪声中信号能量比较微弱时，它在抑制噪声的同时反而能够增加微弱信号的能量，从而达到更好的检测微弱信号的目的。针对齿轮系统故障的振动机理和振动特点，本文中利用随机共振方法对强噪声中的齿轮系统故障振动信号的微弱故障特征量进行提取，以达到对齿轮系统的早期故障进行更好的识别和诊断的目的，论文主要进行了以下几方面的工作：本文首先基于非线性系统和随机共振理论模型，通过建模和仿真实验，研究了随机共振系统自身参数对于随机共振系统输出的影响规律。并对目前的几种随机共振大参数信号转换方法的性能和适用范围通过相应的仿真实验进行了分析与研究。然后结合齿轮系统的实际故障振动信号，研究了齿轮系统早期故障的振动机理及常见的故障信号特征，包括齿轮、轴系故障产生的机理和不同故障的频谱特征，得出了齿轮系统典型故障信号特征细化表。在对齿轮系统故障的微弱振动信号特征研究的基础上，设计了对于齿轮系统早期故障更有针对性的自适应随机共振算法，并通过对仿真及实验的结果分析对算法做了相应的改进。在MFS机械综合故障模拟实验台上进行了齿轮的磨损、点蚀，轴的裂纹和轻度弯曲等几种齿轮系统典型早期故障的实验验证，结果表明，本文设计的自适应随机共振算法可以对齿轮系统中的故障微弱信号进行较好的特征提取，即能够较好的对齿轮系统进行早期的故障识别。
[Abstract]:Gear system is widely used in mechanical engineering and many other fields. The components contained in the gear system will produce periodic impulse impact force in the gear system in the event of failure, which is shown in the spectrum of the corresponding fault signal spectrum characteristics, such as gear wear, pitting, and so on. When the shaft is slightly bent or cracked, the corresponding characteristic signals will be generated on the spectrum map. If the fault modulation information can be extracted from the weak characteristic signal in the early stage of the gear system fault, the intensity and frequency of the fault modulation information can be analyzed to determine the extent and position of the damage of the parts. It can achieve the purpose of early fault diagnosis of mechanical system. Stochastic resonance (SR) is a method by which the mixed signal is input into a nonlinear system and the partial energy of the noise is converted to the signal by the nonlinear system. However, some of the signal detection methods in the past are designed to suppress the noise to achieve the purpose of detecting the signal. Obviously, when the noise is suppressed, the energy of the signal itself must be suppressed and affected. So compared with the traditional signal detection method, the advantage of the stochastic resonance method is that when the signal energy in the noise is relatively weak, the random resonance method has the advantage over the traditional signal detection method. It can not only restrain the noise but also increase the energy of the weak signal, so that it can detect the weak signal better. In view of the vibration mechanism and vibration characteristics of gear system fault, the weak fault characteristic of gear system fault vibration signal in strong noise is extracted by stochastic resonance method in this paper. In order to better identify and diagnose the early faults of gear system, the main work of this paper is as follows: firstly, based on nonlinear system and stochastic resonance theory model, modeling and simulation experiments are carried out. The influence of the parameters of the stochastic resonance system on the output of the stochastic resonance system is studied. The performance and application range of several random resonance large parameter signal conversion methods are analyzed and studied through the corresponding simulation experiments. Then combined with the actual vibration signals of the gear system, the vibration mechanism and the common fault signal characteristics of the gear system early fault are studied, including the mechanism of the gear and shaft system fault generation and the spectrum characteristics of different faults. The characteristic refinement table of typical fault signal of gear system is obtained. Based on the study of the weak vibration signal characteristics of gear system fault, an adaptive stochastic resonance algorithm is designed, which is more specific to the early fault of gear system. The algorithm is improved by analyzing the simulation and experiment results. Several typical early failures of gear system, such as wear, pitting, shaft crack and slight bending, were tested on the MFS mechanical comprehensive fault simulator. The results show that: The adaptive stochastic resonance algorithm designed in this paper can extract the weak signal of the gear system better, that is, the early fault identification of the gear system can be better.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH132.41;TH165.3

【参考文献】