Parallelized sigma-point Kalman filtering for structural dynamics

• Published in: Computers & structures Vol. 92; pp. 193 - 205
• Main Authors: Eftekhar Azam, Saeed, Ghisi, Aldo, Mariani, Stefano
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.02.2012; Elsevier
• Subjects: Applied sciences; Composite delamination; Composite structures; Computation; Computer science; control theory; systems; Control theory. Systems; Degrees of freedom; Dynamical systems; Dynamics; Evolution; Exact sciences and technology; Fracture mechanics (crack, fatigue, damage...); Fundamental areas of phenomenology (including applications); Interface models; Kalman filter; Kalman filtering; Mathematical analysis; Modelling and identification; Nonlinear dynamics; Physics; Sigma-point (unscented) transformation; Solid mechanics; Structural and continuum mechanics; Vibration, mechanical wave, dynamic stability (aeroelasticity, vibration control...); Sigma-point (unscented) transformation; Nonlinear dynamics; Interface models; Composite delamination; Kalman filter; Set point; Unscented transformation; Parallel algorithm; Vibration; Scalability; Punctual transformation; Dynamical system; Modeling; System with n degrees of freedom; Mechanical system; System identification; Non linear effect; Delamination; Shared memory; Structural analysis; Mechanical shock; Damaging
• ISSN: 0045-7949; 1879-2243
• DOI: 10.1016/j.compstruc.2011.11.004

► Sigma-point Kalman filter is adopted to track the state of composite structures undergoing impact-induced delamination. ► Starting from the independent evolution of sigma-points, we propose a parallel implementation of the filter. ► A shared-memory (OpenMP) architecture is adopted. In this paper, the sigma-point Kalman filter (S-PKF) is adopted to track the state of composite structures undergoing impact-induced delamination. Estimates provided by the S-PKF are obtained through a set of sigma-points, which independently evolve in time according to the system dynamics. Since the number of sigma-points grows proportionally to the number of degrees of freedom of the space-discretized structural system, the S-PKF can become computationally demanding. Starting from the aforementioned independent evolution of the sigma-points, we propose a parallel implementation of the S-PKF within a shared-memory (OpenMP) architecture. Scalability and accuracy issues are eventually discussed.