Fault diagnosis of rolling bearing combining improved AWSGMD-CP and ACO-ELM model

The signal of rotating machinery is usually non-stationary, non-linear, and with noise interference. The early fault signal is too weak to extract fault features and the accuracy cannot be effectively guaranteed. A novel fault diagnosis method based on adaptive weighted symplectic geometry mode decomposition (AWSGMD-CP) with Cosine difference factor (CDF) and Pearson correlation coefficient (PCC) and extreme learning machine (ELM) optimized by ant colony optimization (ACO) is proposed. Firstly, the vibration signal is decomposed by SGMD, and several symplectic geometry components (SGCs) are obtained, which can effectively capture the signal characteristics. Secondly, the constrains based on CDF, PCC and variable entropy weighted matrix (VEWM) are adopted to reconstruct SGCs into weighted symplectic combined components (WSCCs). Then calculate the power spectrum entropy (PSE) weighted singular values as the fault feature vectors. Finally, the ELM optimized by ACO is introduced to perform fault classification. Simulation and experimental results show that the proposed method can effectively extract rich fault features from vibration signals, and has a higher diagnosis accuracy, up to 99.18%.

Automated summary

A novel fault diagnosis method based on adaptive weighted symplectic geometry mode decomposition (AWSGMD-CP) with Cosine difference factor (CDF) and Pearson correlation coefficient (PCC) and extreme learning machine (ELM) optimized by ant colony optimization (ACO) is proposed. The method can effectively extract fault features from vibration signals and has a high diagnosis accuracy of up to 99.18%.