液压设备故障诊断的分形方法研究

【作者】 邢志鹏；

【导师】 吴冉泉； 吴百海；

【作者基本信息】 广东工业大学 ， 机械电子工程， 2003， 硕士

【摘要】 分形是一门新的学科，它特别适用于分析非线性系统，而将分形用于机械设备故障诊断则是近年来发展的新动向。可以采用分形方法分析机械系统的非线性信号，从中提取出特征信息，并与其它的分析方法相结合，共同实现机械设备的故障模式识别与故障诊断。 本文首先探讨了分形用于机械系统故障诊断的理论基础。在介绍了分形和分形理论中的自相似性、无标度性、分形维数等概念后，讨论了分形与复杂机械系统的关系。分形中的关联维数反映了复杂机械系统运动过程中的吸引子的动态特性，因而可以采用分形的方法来分析复杂机械系统信号。在此基础上，提出了基于分形的机械系统故障诊断的方法与流程。 本文以轧机液压系统中的液压泵为研究对象，制定了运用分形方法对其进行故障诊断的试验方案，包括数据的获取、数据的采集方案、特征参数的提取和故障诊断方法等。在特征参数的提取中采用关联维数作为系统的特征参数，讨论了关联维数的计算方法和计算参数的选择。 在液压泵故障分形分析与诊断实例部分，对不同工作(故障)状态下的液压泵外壁振动速度信号的关联维数进行了分析，并与频谱分析进行了对比，结果表明，关联维数对设备故障比较敏感，能够反映复杂系统的动态特性，同时它也解释了频谱分析方法无法确定的非线性特征产生的原因，给出了定性和定量的结果。 为了课题研究的方便，单独开发了基于分形的数据分析系统，并对其作了简单的介绍。最后对课题的研究作了总结和展望。更多还原

【Abstract】 Fractal is a new subject and it is especially fit for researching nonlinear system. Recently signal processing methods based on fractal have been applied to mechanical system fault diagnosis. The objective of the proposed research is to develop a method based on fractal theory for analyzing the nonlinear signals of mechanical system. This method will provide further useful information about the underlying dynamics. It is possible to implement mechanical system fault diagnosis by using this method combined with other signals analysis methods.The basic theory of mechanical system fault diagnosis based on fractal is firstly introduced. The correlation dimension reflects the attractor characteristic of complex mechanical system, so the signal processing methods based on fractal may be used to analyze complex mechanical system. In this research, the flow of mechanical system fault diagnosis is provided and the judgment principle of fractal characteristic of signals is also given.Hydraulic pumps are the important part of rolling mill and they are also the researched objects of this paper, hi this research, The experimentation project of fault diagnosis of hydraulic pump is established, including data sampling, obtaining character parameters, fault diagnosis method and so on. The correlation dimension is adopted as the character parameter of hydraulic pump dynamical system. The calculating method of correlation dimension and the principle of selecting parameters such as time delay and embedding dimension are discussed in this paper.In the application of hydraulic pump fault diagnosis by fractal method, this paper analyzes the correlation dimensions of vibration speed signals of hydraulic pump under different working states. The result shows that the correlation dimension is sensitive to hydraulic pump faults and it reflects the dynamic characteristic of dynamics system. The qualitative and quantitative conclusion can be gained by correlation dimension.The data analysis system based on fractal theory is developed during the research and it is introduced. At last, the research is summered and prospected in this paper.更多还原