A Bayesian model updating with incomplete complex modal data. (February 2020)

• Record Type: Journal Article
• Authors: Das, Ayan; Debnath, Nirmalendu
• Is Part Of: Mechanical systems and signal processing. Volume 136(2020)
• Publication Date: 2020-02
• Subjects: Finite element method -- Bayesian approach -- Complex modal data -- Damping -- Maximum a posterior (MAP); Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2019.106524
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: Bayesian FE model updating targeting complex modal data is formulated. Damping is included in updating in classical and non-classical viscous form. Mass, stiffness and damping parameters are updated in efficient iterative manner. Scopes remain for using incomplete modal data and for avoiding mode matching. Formulation for probabilistic damage detection is developed. Abstract: Finite element (FE) model updating using measured dynamic data has gained much attention in view of its contribution towards important areas like model-correction, structural health monitoring. Usually model updating is observed to be performed using measured modal data like modal frequencies and mode shapes – indeed very few works are observed to provide scope for using damping measurements in FE model updating. The present work proposes a Bayesian probabilistic FE model updating using multiple sets of complex modal data (viz. complex modal frequencies and complex mode shapes) and thereby enabling the scope for incorporating damping measurements in updating. Moreover there remains the opportunity for using incomplete complex modal data and avoiding the requirement of mode matching. In this paper, a computationally efficient model updating approach based on maximum a posteriori (MAP) is formulated where all the mass, stiffness and damping parameters are updated in a sequential iterative manner. Detailed formulations are obtained considering damping to be contributed by classical damping (Rayleigh damping) and non-classical viscous damping. Besides, formulations for uncertainty estimation and probabilistic damage detection are also developed. The proposed approach is validated using two numerical examples while utilizing incomplete complex modal data. Performance in updating is evaluated for both undamaged and damaged cases. The proposed approach is observed to be successful for updating of FE model as well as for detection of changes/damages in probabilistic manner, while being computationally efficient as well.