基于LCD和MSE的往复压缩机故障诊断方法研究

发布时间：2018-05-06 07:55

本文选题：往复压缩机 + 故障诊断　；参考：《东北石油大学》2015年硕士论文

【摘要】：往复压缩机是用于压缩和输送气体的机械设备,已成为石油化工领域必不可少的关键设备,随着社会对安全生产重视程度的日益提高,对其实施故障诊断已势在必行。然而,往复压缩机结构复杂,激励源众多,其故障诊断过程仍存在诸多问题亟待深入研究。本文以往复压缩机振动信号为信息源,针对信号所具有的非线性、强非平稳特性,进行基于局部特征尺度分解(LCD)和多尺度熵(MSE)的特征提取方法研究。首先,在对往复压缩机常见故障形式进行调研以及故障机理进行深入分析的基础上,确定以往复压缩机气缸与传动机构部位的常见故障为研究对象,以振动信号分析为手段,开展其故障诊断方法研究。以2D12型往复压缩机为具体实验对象,进行故障模拟实验并测试状态数据,为故障诊断过程实施奠定了基础。其次,针对往复压缩机振动信号的非平稳特性,使用LCD时频分析方法对振动信号进行自适应分解,突出了状态信息。并针对LCD方法在分解强非平稳信号时,基线出现的波形毛刺失真现象,提出了以三次Hermite插值代替线性变换的基线构造方法,对LCD进行了改进。通过对仿真信号和实测信号的对比分析,验证了该方法的有效性。再者,针对往复压缩机振动信号的非线性特性,提出了基于LCD和MSE的故障特征提取方法。利用相关系数法筛选了LCD分解结果中包含主要状态信息的ISC分量,采用MSE对ISC分量进行了复杂性定量描述,并以奇异值分解方法提取了状态特征矩阵的特征值,作为最终的状态特征向量。最后,以往复压缩机的气缸与传动机构部位的常见故障形式为对象,利用基于LCD和MSE的故障特征提取方法实现了状态特征的有效提取,并以支持向量机(SVM)为模式识别器,实现了故障的准确诊断,通过与多种特征提取方法对别,验证了本文方法的优越性。
[Abstract]:Reciprocating compressor is a kind of mechanical equipment used to compress and transport gas. It has become an indispensable key equipment in petrochemical field. With the increasing attention to safety in production, it is imperative to implement fault diagnosis of reciprocating compressor. However, the reciprocating compressor has a complex structure and a large number of excitation sources, so there are still many problems in the fault diagnosis of reciprocating compressor. In this paper, based on the local characteristic scale decomposition (LCD) and multi-scale entropy (MSE), the method of feature extraction based on local characteristic scale decomposition (LCD) and multi-scale entropy (MSE) is studied in view of the nonlinear and strong non-stationary characteristics of reciprocating compressor vibration signal as information source. First of all, on the basis of the investigation of the common fault forms of reciprocating compressor and the analysis of the fault mechanism, it is determined that the common faults in the cylinder and transmission mechanism of reciprocating compressor are taken as the research object, and the vibration signal analysis is taken as the means. The method of fault diagnosis is studied. Taking the 2D12 reciprocating compressor as the concrete experiment object, the fault simulation experiment and the test state data were carried out, which laid the foundation for the implementation of the fault diagnosis process. Secondly, according to the non-stationary characteristic of reciprocating compressor vibration signal, the LCD time-frequency analysis method is used to decompose the vibration signal adaptively, which highlights the state information. Aiming at the waveform burr distortion of LCD method when decomposing strong non-stationary signals, a method of constructing baselines with cubic Hermite interpolation instead of linear transformation is proposed, and the LCD is improved. The effectiveness of the method is verified by comparing the simulated signal with the measured signal. Furthermore, a fault feature extraction method based on LCD and MSE is proposed for the nonlinear characteristics of reciprocating compressor vibration signals. The correlation coefficient method is used to screen the ISC components containing the main state information in the LCD decomposition results, the complexity of the ISC components is quantitatively described by MSE, and the eigenvalues of the state characteristic matrices are extracted by the singular value decomposition method. As the final state eigenvector. Finally, taking the common fault form of cylinder and transmission mechanism of reciprocating compressor as the object, the effective state feature extraction method based on LCD and MSE is used, and the support vector machine (SVM) is used as the pattern recognizer. The fault diagnosis is realized, and the superiority of this method is verified by distinguishing with various feature extraction methods.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TE964

【相似文献】