A non-negative Bayesian learning method for impact force reconstruction. (29th September 2019)

• Record Type: Journal Article
• Authors: Yan, Gang; Sun, Hao
• Is Part Of: Journal of sound and vibration. Volume 457(2019)
• Page Start: 354
• Page End: 367
• Publication Date: 2019-09-29
• Subjects: Impact force reconstruction -- Ill-posed deconvolution -- Bayesian learning -- Non-negative regularization constraints; Sound -- Periodicals; Vibration -- Periodicals; Son -- Périodiques; Vibration -- Périodiques; Sound; Vibration; Periodicals; Electronic journals; 620.205
• DOI: 10.1016/j.jsv.2019.06.013
• Languages: English
• ISSNs: 0022-460X

Abstract

Abstract: Detecting and identifying impact events, which may cause severe damages, is important in assessment of the integrity of many engineering structures. This paper presents a new approach for reconstructing the impact forces applied on engineering structures (e.g., composites) using sensor recordings of the structural responses. The problem is firstly formulated as a discretized deconvolution problem in the time domain with the impact force vector as the unknown. Then with consideration of the physical property of the impact force, an inverse analysis approach of Bayesian learning with non-negative regularization constraints is employed to solve the ill-posed deconvolution problem. The newly proposed impact force reconstruction approach is illustrated by experimental examples performed on a sandwich composite structure. Results have demonstrated the effectiveness and applicability of the proposed approach to reconstruct impact forces. Highlights: A new inversion approach is presented for impact force identification. Non-negative Bayesian learning is employed to solve the ill-posed inverse problem. This approach has been validated by the experiment of a sandwich composite panel.