基于蜣螂算法优化深度极限学习机的中介轴承故障诊断方法  

Inter-shaft fault diagnosis method based on deep extreme learning machine optimized with dung beetle optimizer

作　　者：栾孝驰 汤捷中 沙云东 LUAN Xiaochi;TANG Jiezhong;SHA Yundong(College of Aircraft Engine,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学航空发动机学院,沈阳110136

出　　处：《振动与冲击》2024年第21期96-106,127,共12页Journal of Vibration and Shock

基　　金：辽宁省教育厅系列项目;中国航发产学研合作项目的支持

摘　　要：针对中介轴承故障信号传递路径复杂、受背景噪声干扰大、故障特征提取难,且传统诊断模型准确率受限于测点位置的问题,提出了一种基于自适应噪声完全经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)与蜣螂算法(dung beetle optimizer,DBO)优化深度极限学习机(deep extreme learning machine,DELM)结合的中介轴承故障诊断方法。首先,使用CEEMDAN和由能量比-相关系数-峭度值组成的固有模态分量筛选准则对原始信号进行分解、筛选、重构,在重构信号的时域与频域中提取特征组成特征矩阵;其次,将诊断准确率作为DBO的适应度值,对DELM模型的初始权重进行优化构建出全新的DELM;最后,将特征矩阵输入DELM完成故障诊断。以中介轴承故障数据为例,经DBO优化后的DELM诊断准确率取得了较大提升,在诊断较为困难的45°方向上诊断准确率仍达到了98.75%。结果表明,该诊断方法有效识别了中介轴承故障类型,展现了较强的鲁棒性与泛化能力。Here,aiming at problems of complex fault signal transmission path,high interference from background noise,difficulty in extracting fault features and limited accuracy of traditional diagnosis model due to measurement point positions,a fault diagnosis method for inter-shaft was proposed based on combination of complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and deep extreme learning machine(DELM)optimized with dung beetle optimizer(DBO).Firstly,original signals were decomposed,screened and reconstructed using CEEMDAN and the intrinsic mode component screening criterion composed of energy ratio-correlation coefficient-kurtosis value,features were extracted in time and frequency domains of the reconstructed signals to form feature matrix.Secondly,the diagnosis correct rate was taken as the fitness value of DBO,and initial weights of DELM model were optimized with DBO to reconstruct a brand new DELM.Finally,the feature matrix was input into DELM to complete fault diagnosis.Taking fault data of inter-shaft as an example,the diagnosis correct rate of DELM after optimized with DBO was more largely improved,and it was shown that the diagnosis correct rate in 45°direction which is difficult to diagnose still reaches 98.75%.The results showed that the proposed method can effectively identify types of inter-shaft faults,and demonstrate stronger robustness and generalization ability.

关 键 词：中介轴承 故障诊断 模态分解 蜣螂算法(DBO) 深度极限学习机(DELM) 

分 类 号：V263.6[航空宇航科学与技术—航空宇航制造工程]