基于增强稀疏分解的发动机叶片监测振动辨识  

Vibration Identification of Engine Blade Monitoring Based on Enhanced Sparse Decomposition

作　　者：柳雅龙 李峰 LIU Yalong;LI Feng(Guangzhou Hengyuan Power Machinery Co.,Ltd.,Guangzhou 510660,China;School of Computer Science and Technology,Henan Polytechnic University,Zhengzhou 454003,China)

机构地区：[1]广州恒源动力机械有限公司,广东广州510660 [2]河南理工大学计算机科学与技术学院,河南郑州454003

出　　处：《机械制造与自动化》2024年第6期278-282,共5页Machine Building & Automation

摘　　要：以往将应变片贴在叶片外表的监测方法仅可对少许叶片的信号进行监测,且信号传输难度较大。为此设计一种基于增强稀疏分解(ESD)的发动机叶片监测振动辨识方法,开展非欠采样和欠采样状态下的振动信号辨识研究。研究结果表明:非欠采样振动下,相比较矩阵特征空间分解(MUSIC)法、非线性最小二乘拟合(NLS)方法,增强稀疏分解实现了充分滤除其他频率成分干扰。欠采样振动下,MUSIC方法和NLS方法不理想,而ESD的辨别精准性依然很高。欠采样同步振动下,MUSIC方法和NLS方法对于频率成分难以辨别;ESD能够实现对各成分的辨别。通过仿真实验验证了ESD方法可以精准识别叶片的振动信号,对后续的性能优化具有一定的理论支撑意义。With regard to the limit of blade signal monitoring by the monitoring method of sticking strain gauge on the blade surface and difficulty in signal transmission,a vibration identification method based on enhanced sparse decomposition(ESD)for engine blade monitoring was designed,and vibration signal identification under non-undersampled and undersampled conditions was studied.The results show that under non-under-sampled vibration,compared with the comparison matrix eigenspace decomposition(MUSIC)method and the nonlinear least square fitting(NLS)method,the enhanced sparse decomposition can fully filter out the influence of other frequency components.Under under-sampled vibration,MUSIC method and NLS method are far from satisfaction,but the discrimination accuracy of ESD is still very high.Under under-sampled synchronous vibration,both MUSIC method and NLS method are difficult to identify the frequency components,while ESD can still realize the discrimination of each component.The simulation experiment verifies that the ESD method can accurately identify the vibration signals of blades,which plays a certain theoretical supporting significance for the subsequent performance optimization.

关 键 词：发动机 振动检测 叶片监测 增强稀疏分解 欠采样 参数辨识 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]