Abstract: It is difficult to extract the fault features of rolling bearings under noise interference.Aiming at this problem,a new feature extraction method based on improved Variational Mode Decomposition (VMD) and Composite Zoom Permutation Entropy (CZPE) was proposed,and the optimized Stacked Denoising Auto-Encoders (SDAE) was used for fault classification.An improved VMD method was proposed to adaptively optimize the decomposition parameters by the new comprehensive evaluation index of ′cosine similarity-kurtosis-envelope entropy′,and the decomposed Intrinsic Mode Function (IMF) were screened by this index.To extract more comprehensive fault features,a new composite scaling permutation entropy was introduced to quantify the fault features of each effective IMF.A hybrid algorithm based on Rat Swarm Optimization (RSO) and Sparrow Search Algorithm (SSA) was proposed to optimize the hyperparameters of SDAE network,and the fault features were input into the optimized SDAE network to obtain the classification results.The American CWRU bearing data set was used for verification.The experimental results showed that the method could extract the fault features under the background of noise comprehensively and stably,and had better anti-noise performance and higher fault diagnosis accuracy than other methods.

Key words: variational mode decomposition, comprehensive evaluation index, composite zoom permutation entropy, hybrid algorithm, stacked denoising autoencoders

摘要： 针对滚动轴承在噪声干扰下故障特征难以提取的问题,提出一种改进变分模态分解(VMD)和复合缩放排列熵(CZPE)的特征提取新方法,并利用优化堆叠降噪自编码器(SDAE)进行故障分类。首先,提出由“余弦相似度—峭度—包络熵”新综合评价指标自适应优化分解参数的改进VMD方法,并通过该指标筛选分解后的本征模态函数 (IMF) 分量;然后,为提取更全面的故障特征,引入新的复合缩放排列熵对各有效IMF的故障特征进行量化;最后,提出一种基于鼠群优化算法(RSO)与麻雀搜索算法(SSA)的混合算法优化SDAE网络超参数,将故障特征输入优化后SDAE网络中得到分类结果。采用美国CWRU轴承数据集进行验证,实验结果表明该方法能全面稳定地提取背景噪声下的故障特征,且与其他方法相比具有更好的抗噪性能和更高的故障诊断准确率。

关键词: 变分模态分解, 综合评价指标, 复合缩放排列熵, 混合算法, 堆叠降噪自编码器