改进AMD广义形态分形维数和KFCMC的液压泵故障诊断方法  

Hydraulic pump fault diagnosis method based on the improved AMD, generalized morphological fractal dimensions and kernel fuzzy C-means clustering

作　　者：郑直 姜万录[2,3] 王宝中 王莹 ZHENG Zhi;JIANG Wanlu;WANG Baozhong;WANG Ying(College of Mechanical Engineering,North China University of Science and Technology,Tangshan 063210,China;Hebei Provincial Key Laboratory of Heavy Machinery Fluid Power Transmission and Control,Yanshan University,Qinhuangdao 066004,China;Key Laboratory of Advanced Forging&Stamping Technology and Science,Ministry of Education of China,Yanshan University,Qinhuangdao 066004,China)

机构地区：[1]华北理工大学机械工程学院,河北唐山063210 [2]燕山大学河北省重型机械流体动力传输与控制重点实验室,河北秦皇岛066004 [3]燕山大学先进锻压成形技术与科学教育部重点实验室,河北秦皇岛066004

出　　处：《振动与冲击》2019年第18期46-52,102,共8页Journal of Vibration and Shock

基　　金：华北理工大学博士科研启动基金(0088/28412499);国家自然科学基金(51875498);河北省自然科学基金重点项目(E2017209059;E2018203339)

摘　　要：针对液压泵故障诊断问题,提出了一种基于改进解析模态分解(AMD)、广义形态分形维数(GMFD)和核模糊C均值聚类(KFCMC)相结合的新方法。根据故障特征频率先验知识,在有效二分频范围内对实测液压泵多模态故障振动信号进行AMD分解,并基于欧氏距离法选定实现最优分解的二分频;将基于最优二分频所提取含有丰富运行特征信息的故障分量信号作为数据源,并提取GMFD作为特征向量;利用KFCMC实现对液压泵不同故障的诊断。此外,还利用原始AMD、经验模态分解(EMD)、集总经验模态分解(EEMD)、局部均值分解(LMD)、变分模态分解(VMD)和模糊C均值聚类(FCMC)方法对上述信号进行分析,结果表明所提方法效果要优于上述传统分解和诊断方法。通过对仿真和实测液压泵故障振动信号的实验验证,表明该方法可以有效地诊断液压泵不同故障。Aiming at the fault diagnosis of hydraulic pumps,a new fusion method was proposed based on the analysis mode decomposition(AMD),general morphological fractal dimensions(GMFD)and kernel fuzzy C-means clustering(KFCMC).Based on the priori knowledge about fault feature frequencies,the AMD was applied to decompose multi-mode vibration fault signals of a hydraulic pump in the effective range of bisecting frequency,and the best bisecting frequency for realizing optimal decomposition was chosen according to the Euclidean distance.Then,the mode extracted by virtue of the optimal bisecting frequency,which was rich in fault feature informations,was used as data sources to extract the GMFD and adopt it as feature vectors.Finally,the KFCMC was used to diagnose hydraulic pump faults.In addition,the methods of original AMD,experience mode decomposition(EMD),ensemble experience mode decomposition(EEMD),local mode decomposition(LMD),variational mode decomposition(VMD)and fuzzy C-means clustering(FCMC)were also used to decompose the signals,and it is shown that the proposed method is better than the others.Through the simulation and experiment verification on the fault signals of the tested hydraulic pump,it is shown that the proposed method is available to diagnose different hydraulic pump faults with an enough accuracy.

关 键 词：液压泵 解析模态分解(AMD) 广义形态分形维数(GMFD) 核模糊C均值聚类(KFCMC) 

分 类 号：TH137[机械工程—机械制造及自动化] TP277[自动化与计算机技术—检测技术与自动化装置]