摘要
针对液压系统常见的泄漏、气穴故障问题,从时域分析和频域分析两个方面建立液压系统故障诊断体系,提出了一种基于EMD-SVD变换的液压机泄漏、气穴故障特征提取方法。通过经验模态函数将各类故障信号分解为8类不同时间尺度的本征模态函数IMFs,对其中能量集中的前5类IMFs组成的初始向量矩阵进行SVD(奇异值分解)得到特征向量,组成故障特征矩阵。为比较各类故障诊断方法的最终识别效果,实验同时利用小波分析和Hilbert-Huang变换2类方法获得了2类不同特征信号进行对比,最后通过模糊聚类分析对各类特征信号进行样本隶属度计算来判断故障信号所属类别。结果表明,基于EMD-SVD变换的故障特征提取方法取得了最佳识别效果,其识别准确率可达99.3%。
Aiming at the common hydraulic system leakage,cavitation failure problems,the fault diagnosis system of hydraulic system was established from two aspects:time domain analysis and frequency domain analysis and frequency domain analysis.A feature extraction method for hydraulic press leakage and cavitation fault based on EMD-SVD transform was presented,all kinds of fault signals were decomposed into 8 kinds of IMFs with different time scales by empirical modal functions,the eigenvectors were obtained by SVD(singular value decomposition)of the inital vector matrix composed of the first five IMFs in the energy concentration,the fault feature matrix was constituted.In order to compare the final recognition effect of various fault diagnosis methods,three different kinds of feature signals were obtained by using correlation domain analysis,wavelet analysis and Hilbert-Huang transform methods for comparison.The classification of fault signals was determined by calculating the sample membership degree of various feature signals through fuzzy clustering analysis.The results show that the fault feature extraction method based on EMD-SVD transform achieves the best recognition effect,and its recognition accuracy can reach 99.3%.
作者
钟岳
王钊
方虎生
殷勤
刘帅
Zhong Yue;Wang Zhao;Fang Husheng;Yin Qin;Liu Shuai(Army Engineering University of PLA,Nanjing 210042,China)
出处
《机电工程技术》
2020年第11期104-108,共5页
Mechanical & Electrical Engineering Technology
基金
国家重点研发计划项目(编号:2016YFC0802904)
国家自然科学基金项目(编号:61671470)。