A novel fault diagnosis procedure based on improved symplectic geometry mode decomposition and optimized SVM. (March 2021)

• Record Type: Journal Article
• Authors: Zhang, Xiaoyuan; Li, Chaoshun; Wang, Xianbo; Wu, Huanmei
• Is Part Of: Measurement. Volume 173(2021)
• Publication Date: 2021-03
• Subjects: Symplectic geometry mode decomposition -- Cosine similarity -- Rotating machineries fault diagnosis -- Singular value decomposition -- Optimized support vector machines; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2020.108644
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: The proposed methodology is based on improved SGMD and optimized SVM. A new constraint condition based on cosine similarity is proposed to improve SGMD. A turning point is found in the sequence of singular values to eliminate noise. HHO combined with ICDF is used to optimize the parameters of SVM. The effectiveness of the proposed method is fully evaluated by experiments. Abstract: A novel fault diagnosis procedure based on improved symplectic geometry mode decomposition (SGMD) and optimized SVM is presented. In the proposed procedure, a vibration signal is firstly decomposed by SGMD into a set of components. Then a novel constraint condition based on cosine similarity is proposed to reconstruct the decomposed components into some independent superimposed symplectic geometry components (SGCs). Next, a series of singular values are got by singular value decomposition from the matrix whose rows are SGCs. And a turning point is found in the singular values sequence by two defined criteria, before which the singular values are selected to form the final feature vectors. Finally, SVM optimized by inter-cluster distance in the feature space (ICDF) and Harris hawks optimization algorithm (HHO) is used to diagnose faults. The proposed procedure is evaluated by experiments and comparative studies. The results demonstrate its effectiveness and robustness for rotating machineries fault diagnosis.