Scholarlink

<p>Aiming at the problem that rolling bearing fault signals exhibit nonstationary characteristics and that feature extraction is challenging under background noise interference. This paper introduces entropy features combined with extreme learning machine (ELM) to construct a fault diagnosis method. Firstly, singular spectrum analysis and energy standard deviation are used to select CEEMDAN noise standard deviation, and particle swarm optimization (PSO) algorithm is used to select number of adaptive noises. Secondly, the correlation coefficient and variance contribution rate are used as comprehensive evaluation indexes to select the intrinsic mode functions (IMFs) components decomposed by CEEMDAN. The refined composite multi-scale dispersion entropy (RCMDE) is calculated as the feature vector, and the setting of RCMDE parameters is analyzed. Then, the joint mutual information maximization (JMIM) is used to perform dimensionality reduction on the high-dimensional feature vectors, and the feature vectors after dimensionality reduction are divided into the training set and the test set. Finally, the data set is input into the ELM model optimized by the salp swarm algorithm (SSA) to identify different fault bearing states. Two open-source datasets and our own collected data are used to verify the performance of the method. The results show that the proposed method can effectively identify different bearing fault states.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fault diagnosis method of rolling bearing based on adaptive CEEMDAN, RCMDE, and ELM

  • 期刊论文

  • 发表日期：

    2025年10月8日

查看全文
  • Shengjie Wang,
  • Zhenrui Peng

摘要

Aiming at the problem that rolling bearing fault signals exhibit nonstationary characteristics and that feature extraction is challenging under background noise interference. This paper introduces entropy features combined with extreme learning machine (ELM) to construct a fault diagnosis method. Firstly, singular spectrum analysis and energy standard deviation are used to select CEEMDAN noise standard deviation, and particle swarm optimization (PSO) algorithm is used to select number of adaptive noises. Secondly, the correlation coefficient and variance contribution rate are used as comprehensive evaluation indexes to select the intrinsic mode functions (IMFs) components decomposed by CEEMDAN. The refined composite multi-scale dispersion entropy (RCMDE) is calculated as the feature vector, and the setting of RCMDE parameters is analyzed. Then, the joint mutual information maximization (JMIM) is used to perform dimensionality reduction on the high-dimensional feature vectors, and the feature vectors after dimensionality reduction are divided into the training set and the test set. Finally, the data set is input into the ELM model optimized by the salp swarm algorithm (SSA) to identify different fault bearing states. Two open-source datasets and our own collected data are used to verify the performance of the method. The results show that the proposed method can effectively identify different bearing fault states.