基于参数优化的ICEEMDAN滚动轴承故障诊断  

ICEEMDAN Fault Diagnosis of Rolling Bearings Based on Parameter Optimization

作　　者：李雨晴 马洁 LI Yuqing;MA Jie(College of Mechanical Electrical Engineering,Beijing Information Science&Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学机电工程学院,北京100192

出　　处：《机床与液压》2025年第6期21-27,共7页Machine Tool & Hydraulics

基　　金：国家自然科学基金面上项目(61973041)。

摘　　要：滚动轴承长期处于噪声污染的工作环境中,其故障诊断常受到噪声干扰,难以对故障特征信息进行有效提取。针对此问题,提出基于冠豪猪优化算法(CPO)的改进自适应噪声完备经验模式分解(ICEEMDAN)联合卷积神经网络(CNN)的故障诊断方法。通过CPO对ICEEMDAN的白噪声幅值权重及噪声添加次数进行参数寻优,将最优参数返回并进行信号分解,以最小包络熵作为相关度函数,筛选出相关程度高的特征模态分量(IMF);将重构的有效特征分量IMF转化为特征向量并输入到CNN模型中,从而实现轴承的故障诊断。与已有模型进行对比,结果表明:该方法具有较强的泛化能力,诊断精度明显优于现有方法,并且具有更高的诊断效率。Rolling bearings is in a noise-polluted working environment for a long time,and its fault diagnosis is often disturbed by noise,so it is difficult to extract fault feature information effectively.In view of this problem,an improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)combined with convolution neural network(CNN)based on crested porcupine optimizer(CPO)was proposed.The white noise amplitude weight and noise addition times of ICEEMDAN were optimized by CPO,and the optimal parameters were returned for signal decomposition.The intrinsic mode function with high correlation range was selected by using the minimum envelope entropy as the correlation function,and the reconstructed effective feature component IMF was transformed into eigenvector as input data to the CNN model to realize bearing fault diagnosis.Compared with the existing models,the results show that the proposed method has strong generalization ability,it is obviously better than the existing methods in terms of diagnosis accuracy,and has higher diagnosis efficiency.

关 键 词：故障诊断 冠豪猪优化算法(CPO) 改进自适应噪声完备经验模式分解(ICEEMDAN) 卷积神经网络(CNN) 

分 类 号：TH133.33[机械工程—机械制造及自动化]