Sparse Reconstruction Based on the ADMM and Lasso-LSQR for Bearings Vibration Signals

• Published in: IEEE access Vol. 5; pp. 20083 - 20088
• Main Authors: Song, Wanqing, Nazarova, Maria N., Zhang, Yujin, Zhang, Ting, Li, Ming
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2017; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: ADMM; Algorithm design and analysis; Algorithms; bearing vibration; Discrete cosine transform; Discrete cosine transforms; Image reconstruction; Lasso-LSQR; Mathematical model; Optimization; Reconstruction; Roller bearings; Rolling bearings; Sparse reconstruction; Termites; Vibration; Vibrations
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2017.2757026

In this paper, we introduce a novel method for reconstructing roller bearings vibration signals. As well as the sparse reconstruction algorithm, our approach is based on the Lasso via the alternate direction multiplier method (ADMM) and optimized by least square QR-factorization (LSQR), which takes the priority over the Basis Pursuit and Lasso in iterations and errors. First, we use the discrete cosine transformation to achieve sparse signals, then we compress signals by using the Gaussian random matrix, and, finally, we reconstruct the original signals with the Lasso-LSQR by using the ADMM. According to the results, vibration signals can keep sufficient reconstruction accuracy with high compressive ratio, which validates the effectiveness of the method for vibration signals.