某型发动机转子振动故障的信号特征提取及诊断

发布时间：2018-03-01 22:18

本文关键词： 经验模式分解法信息熵故障诊断支持向量机证据理论　出处：《沈阳航空航天大学》2015年硕士论文　论文类型：学位论文

【摘要】：对于航空发动机故障诊断技术而言，故障特征的识别与提取将直接关系到故障诊断的准确性和有效性。目前利用振动信号对航空发动机进行故障诊断是比较常用的方法，根据航空发动机在不同状态下振动信号呈现的不同特征对其进行诊断。针对传统信号分析方法难以准确描述转子振动信号非平稳特性以及信号特征难以定量评价问题，本文采用经验模式分解法结合信息熵进行特征向量的提取。然后利用能量法选取包含主要故障信息的分量，并利用相关系数法对选取的主要分量的准确性进行验证。在选出主要分量的基础上计算其在时域、频域以及时-频域上的奇异谱熵、功率谱熵、小波能谱熵和小波包空间特征谱熵，以此四种信息熵作为转子故障诊断的特征向量。本文通过支持向量机结合证据理论的方法对发动机进行故障诊断，最后利用多分类支持向量机对发动机转子的三种典型故障进行多分类诊断。根据证据理论对信号在不同测点、不同工况下进行信息融合，，并且以概率形式输出结果。这种方法可以提高对发动机故障诊断的准确性，其结果能更直观的观测故障的严重程度。
[Abstract]:For aero-engine fault diagnosis technology, fault feature identification and extraction will be directly related to the accuracy and effectiveness of fault diagnosis. According to the different characteristics of aero-engine vibration signals in different states, it is diagnosed. The traditional signal analysis method is difficult to accurately describe the non-stationary characteristics of rotor vibration signal and the signal characteristics are difficult to evaluate quantitatively. In this paper, the method of empirical mode decomposition combined with information entropy is used to extract the eigenvector, and then the energy method is used to select the components containing the main fault information. The accuracy of the selected main components is verified by the correlation coefficient method. The singular spectral entropy and power spectral entropy in time domain and frequency domain are calculated on the basis of selecting the main components. Wavelet energy spectrum entropy and wavelet packet spatial characteristic spectrum entropy are used as the characteristic vectors of rotor fault diagnosis. In this paper, the support vector machine (SVM) combined with evidence theory is used to diagnose the fault of the engine. Finally, the multi-classification support vector machine is used to diagnose the three typical faults of the engine rotor. According to the evidence theory, the information fusion of the signal at different measuring points and different working conditions is carried out. This method can improve the accuracy of engine fault diagnosis, and the result can observe the severity of fault more intuitively.
【学位授予单位】：沈阳航空航天大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V263.6

【相似文献】