Succinct and fast empirical mode decomposition. (15th February 2017)

• Record Type: Journal Article
• Authors: Li, Hongguang; Hu, Yue; Li, Fucai; Meng, Guang
• Is Part Of: Mechanical systems and signal processing. Volume 85(2017)
• Page Start: 879
• Page End: 895
• Publication Date: 2017-02-15
• Subjects: Empirical mode decomposition -- Order-statistics filter -- Vibration signal; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2016.09.031
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: Empirical mode decomposition (EMD) has been extensively studied and widely utilized in various areas. In this paper, order-statistics filters are used to replace the traditional interpolation methods and estimate envelopes of the EMD method. Window size selection criteria are proposed to optimize the processing results. Both simulated and experimental signal are applied to investigate characteristics and effectiveness of the proposed method. The results demonstrate that the envelope estimation procedure provides a tremendous enhancement of the EMD method. In the proposed method, application of the order-statistics filter can simplify the process of estimating envelopes and minimize the end effect of the traditional EMD method. Moreover, sifting stop criterion and window width selection are optimized to improve the decomposition speed and processed results. Hence, the processed method is fast, time efficient and effective, and is named as succinct and fast EMD (SF-EMD) in this paper. An application of multiple fault diagnosis for a rotor test rig verifies its potential in practical engineering. Highlights: Order-statistics filters are used to replace the traditional interpolation methods and estimate envelopes of the EMD method. Window size is selected from natural oscillatory feature of the original signal. The method is suitable for extracting faint and/or transient signal component. Computation time is greatly decreased and there is no end effect in the decomposed modes.