基于EEMD-LVQ的机电作动器故障诊断方法  

Fault Diagnosis based on EEMD-LVQ for Electro-mechanical Actuator

作　　者：王晓明 付继伟 韩松 白云鹤 李少石 WANG Xiaoming;FU Jiwei;HAN Song;BAI Yunhe;LI Shaoshi(Beijing Institute of Aerospace Systems Engineering,Beijing,100076;School of Automation Science and Electrical Engineering,Beihang University,Beijing,100191)

机构地区：[1]北京宇航系统工程研究所,北京100076 [2]北京航空航天大学,自动化科学与电气工程学院,北京100191

出　　处：《导弹与航天运载技术（中英文）》2023年第5期1-7,共7页Missiles and Space Vehicles

摘　　要：针对传统基于集合经验模态分解算法在故障特征区分性和LVQ算法在训练效率和稳定性方面存在的问题,提出一种基于集合经验模态分解-学习矢量量化网络(Ensemble Empirical Mode Decomposition,Learning Vector Quantization,EEMD-LVQ)的机电作动器(Electro-mechanicalActuator,EMA)的故障诊断方法。首先,通过EEMD算法对信号进行分解并计算能量分布向量,并利用相关系数筛选特征实现降维,增强故障特征向量的区分性;然后,利用经过余弦衰减算法优化的LVQ神经网络对故障特征向量集进行训练和检测,从而获得诊断结果。实际EMA数据的试验验证和对比分析证明了提出的故障诊断方法可提高LVQ算法的训练效率,并且可以兼顾后期的稳定性。The fault diagnosis method based on ensemble empirical mode decomposition(EEMD)and learning vector quantization(LVQ)is proposed for electro-mechanical actuator(EMA),which is aiming to address the problems of the traditional EEMD algorithm in fault feature discrimination and the LVQ algorithm in training efficiency and stability.Firstly,the signal is decomposed and the energy distribution vectors are calculated using the EEMD algorithm.The features are then selected using the correlation coefficient and dimensionality reduction is performed to enhance the discriminability of the fault feature vectors.Next,the LVQ neural network,optimized with cosine decay algorithm,is used to train and detect the fault feature vector set,obtaining the diagnostic results.Experimental validation and comparative analysis of actual EMA data demonstrate that the proposed fault diagnosis method improves the training efficiency of the LVQ algorithm while also considering its stability in the later stages.

关 键 词：机电作动器 故障诊断 集合经验模态分解 学习矢量量化网络 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] V42[自动化与计算机技术—控制科学与工程]