基于CEEMD-MPE和ELM的齿轮箱故障诊断研究  

Fault Diagnosis of Gearbox Based on CEEMD-MPE and ELM

作　　者：王斌 崔宝珍[1] WANG Bin;CUI Bao-zhen(School of Mechanical Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学机械工程学院,太原030051

出　　处：《组合机床与自动化加工技术》2019年第4期103-106,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金:基于粒子群优化和滤波技术的复杂传动装置早期故障诊断研究(50875247)

摘　　要：考虑到齿轮箱振动信号存在非平稳性和非线性等特点导致故障特征提取困难的问题,提出了一种基于互补集合经验模态分解(CEEMD)和多尺度排列熵(MPE)相结合的故障特征提取方法。首先对齿轮箱振动信号进行互补集合经验模态分解,并根据相关系数原则对各模态分量进行筛选,再利用多尺度排列熵对筛选出的模态分量进行特征提取;最后将提取出的故障特征输入到极限学习机(ELM)中进行分类识别,并与传统的径向基(RBF)神经网络进行对比,实验结果表明:采用CEEMD和MPE相结合的办法能够有效提取齿轮箱振动信号的故障特征,极限学习机能够准确、快速地进行齿轮箱故障识别。Considering the difficulty of fault feature extraction due to the non-stationary and non-linear characteristics of gearbox vibration signals, a fault feature extraction method based on complementary set empirical mode decomposition(CEEMD) and multi-scale permutation entropy(MPE) is proposed. Firstly, the vibration signal of gearbox is decomposed by CEEMD, according to the principle of correlation coefficient, the modal components are screened, then the multi-scale permutation entropy algorithm is used to extract the feature of the selected modal components, and finally the extracted fault features are inputted into the extreme learning machine(ELM) for fault classification and recognition. Compared with the traditional radial basis function(RBF) neural network, the experimental result shows that the fault characteristics of gearbox vibration signals can be effectively extracted by the combination of CEEMD and MPE, and the Extreme Learning Machine can accurately and rapidly detect the fault of gearbox.

关 键 词：齿轮箱 互补集合经验模态分解 多尺度排列熵 极限学习机 故障诊断 

分 类 号：TH17[机械工程—机械制造及自动化] TG506[金属学及工艺—金属切削加工及机床]