A novel method of composite multiscale weighted permutation entropy and machine learning for fault complex system fault diagnosis. (1st July 2020)

• Record Type: Journal Article
• Authors: He, Cheng; Wu, Tao; Liu, Changchun; Chen, Tong
• Is Part Of: Measurement. Volume 158(2020)
• Publication Date: 2020-07-01
• Subjects: Extreme-point symmetric mode decomposition -- Composite multiscale weighted permutation entropy -- Gravitational search algorithm based on multiple adaptive constraint strategy -- Least squares support vector machine -- Fault diagnosis; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2020.107748
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A modified algorithm named ESMD is developed, which can improve signal decomposition performance. The new integration framework (ESMD-CMWPE) is proposed for fault classification. The MCAGSA optimized LSSVM is presented to improve the algorithm's fault diagnosis accuracy and diagnosis recognition rate. The validity of the proposed scheme is verified through the analysis of the experimental data. Abstract: A novel fault diagnosis method is proposed for rolling bearing by combining extreme-point symmetric mode decomposition (ESMD) composite multiscale weighted permutation entropy (CMWPE) and gravitational search algorithm based on multiple adaptive constraint strategy (MACGSA) optimized least squares support vector machine (LSSVM). In order to solve the problem of intrinsic mode function (IMF) modal aliasing and small differences in fault features, ESMD and CMWPE are used to obtain a more sensitive high-dimensional feature vector set. Aiming at the low accuracy of LSSVM fault diagnosis, MACGSA was used to optimize LSSVM to improve the accuracy of fault diagnosis. ESMD is used to process the rolling bearing data to obtain a series of IMFs; Then, extracting the CMWPE values of IMFs to form a high-dimensional feature vector set; Finally, the MACGSA-LSSVM model is adopted to achieve fault classification. Compared with other diagnostic methods, this method has higher diagnostic accuracy.