A dual-layer dimension-wise fuzzy finite element method (DwFFEM) for the structural-acoustic analysis with epistemic uncertainties. (1st August 2019)

• Record Type: Journal Article
• Authors: Xu, Menghui; Du, Jianke; Wang, Chong; Li, Yunlong; Chen, Jianbin
• Is Part Of: Mechanical systems and signal processing. Volume 128(2019)
• Page Start: 617
• Page End: 635
• Publication Date: 2019-08-01
• Subjects: Structural-acoustic analysis -- Epistemic uncertainty propagation -- Interval perturbation method -- Subinterval technique -- Orthogonal polynomial approximation; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2019.04.006
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A dimension-wise analysis FFEM (DwFFEM) is formulated in a dual-layer framework. The current MIPM was corrected. The interval shift effect for IPM-based methods is eliminated by the DwFFEM. The efficiency of DwFFEM is higher than that of IPM-based methods for fuzzy analysis. The DwFFEM applies to any analysis procedure for the epistemic uncertainty propagation. Abstract: Epistemic uncertainty quantification in the structural-acoustic analysis was performed by a dimension-wise analysis fuzzy finite element method in this paper. This novel procedure was formulated in a dual-layer framework, i.e. a fundamental layer at zero-cut and filtrating layer at any nonzero-cut. The first layer was to identify the minimal and maximal (min/max) points of each slice of the response surface at zero-cut while the seconder layer was to filtrate slice's min/max points at any nonzero-cut. Subsequently, the min/max input vectors at any alpha-cut were dimension-wisely assembled at which the interval response of the coupled system was obtained by two crisp finite element analysis. The membership function of the fuzzy response was finally recomposed by intervals at discrete alpha-cuts. Besides, important corrections of current high-order interval perturbation methods were simultaneously made in this paper. The proposed method applies to any analysis procedure to quantify epistemic uncertainties of outputs due to its nonintrusive property. Accuracy and efficiency of the proposed method were validated by numerical examples.