Data-Driven Forecasting of Postflutter Responses of Geometrically Nonlinear Wings

Riso, Cristina; Ghadami, Amin; Cesnik, Carlos E. S.; Epureanu, Bogdan I.

doi:10.2514/1.j059024

Cited by 33 publications

(9 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At a time where data-driven technologies are increasing in popularity for numerous applications across all disciplines -of particular relevance to this paper in aeroelasticity [44][45][46][47] -and even certification agencies have begun to plan towards integrating them into aircraft certification processes [48], we consider our approach as hybrid between data-driven and physics-based modeling. This scheme is built on multiple training physical models as opposed to collections of input-output data more typically used in "classical" data-driven approaches.…”

Section: Parameter Space Sampling Methodsmentioning

confidence: 99%

Adaptive Sampling for Interpolation of Reduced-Order Aeroelastic Systems

2022

View full text Add to dashboard Cite

A new strategy for the interpolation of parametric reduced-order models of dynamic aeroelastic systems is introduced. Its aim is to accelerate the numerical exploration of geometrically-nonlinear aeroelastic systems over large design spaces or multiple flight conditions. The parametric reduced-order models are obtained from high-dimensional models by Krylov subspace projection. They are subsequently interpolated to acquire realizations inexpensively anywhere in the parameter space, where having the state-space as opposed to interpolating an output metric permits the use of linear analysis tools. The interpolation scheme is heavily conditioned by the available training points, thus a novel methodology based on adaptive sampling and a combinatorial use of the available true-systems knowledge is presented, whereby we reuse all known data of the true models as different combinations of training and testing data to build statistical surrogates of the interpolation error across the parameter space and refine the sampling in those regions that need it. This minimizes the number of costly-to-evaluate functions calls and ensures that parameter space regions are sampled according to the underlying system dynamics. The initial implementation of this adaptive sampling strategy is demonstrated on a very flexible wing with a complex stability envelope.

show abstract

Section: Parameter Space Sampling Methodsmentioning

confidence: 99%

Adaptive Sampling for Interpolation of Reduced-Order Aeroelastic Systems

2022

View full text Add to dashboard Cite

show abstract

“…Data-driven prediction methods have also offered a solution to the formidable challenge of predicting catastrophic events in a variety of complex systems. Recent studies have shown that features extracted from data can be used to predict critical transitions [165] and extreme events [166] in the dynamics of a variety of complex systems, including aeroelastic systems [39,40,44,167,168], ecological systems [51,[169][170][171][172][173][174][175], epidemiological systems [176][177][178][179], traffic flow systems [158,180] and fluid flows [166,[181][182][183].…”

Section: Introductionmentioning

confidence: 99%

Data-driven prediction in dynamical systems: recent developments

Ghadami

Epureanu

2022

Phil. Trans. R. Soc. A.

Self Cite

View full text Add to dashboard Cite

In recent years, we have witnessed a significant shift toward ever-more complex and ever-larger-scale systems in the majority of the grand societal challenges tackled in applied sciences. The need to comprehend and predict the dynamics of complex systems have spurred developments in large-scale simulations and a multitude of methods across several disciplines. The goals of understanding and prediction in complex dynamical systems, however, have been hindered by high dimensionality, complexity and chaotic behaviours. Recent advances in data-driven techniques and machine-learning approaches have revolutionized how we model and analyse complex systems. The integration of these techniques with dynamical systems theory opens up opportunities to tackle previously unattainable challenges in modelling and prediction of dynamical systems. While data-driven prediction methods have made great strides in recent years, it is still necessary to develop new techniques to improve their applicability to a wider range of complex systems in science and engineering. This focus issue shares recent developments in the field of complex dynamical systems with emphasis on data-driven, data-assisted and artificial intelligence-based discovery of dynamical systems. This article is part of the theme issue 'Data-driven prediction in dynamical systems'.

show abstract

“…Limit cycle oscillation (LCO) is frequently encountered in engineering applications, such as, aeroelastics (Thomas et al, 2002;Huang et al, 2018;He et al, 2019a;Li and Ekici, 2019;Jonsson et al, 2019;Riso et al, 2020), bipedal robotics (Grizzle et al, 2001;Shiriaev et al, 2008;Manchester et al, 2011), and combustion (Waugh et al, 2014;Xu et al, 2020). In the context of design optimization, the scalability of computational cost with respect to the number of design variables is essential for problems with a large number of design variables and with a PDE governing equation.…”

Section: Introductionmentioning

confidence: 99%

Adjoint-based limit cycle oscillation instability sensitivity and suppression

Jónsson

Martins

2022

Nonlinear Dyn

View full text Add to dashboard Cite

Limit cycle oscillations (LCO) can be supercritical or subcritical. The supercritical response is benign, while the subcritical response can be destructive. It is desirable to avoid subcritical responses by enforcing LCO stability through design optimization constraints. However, there is a need for a computational approach that can model this instability and can handle hundreds or thousands of design variables. We propose a simple metric to determine the LCO stability using a fitted bifurcation diagram slope to address this need. We develop an adjoint-based formula to compute the stability derivative efficiently with respect to many design variables. To evaluate the stability derivative, we only need to compute the time-spectral adjoint equation three times no matter the number of design variables in the optimization problem. The proposed adjoint method is verified with finite differences, achieving a five-digit agreement between the two approaches. We consider a stability-constrained LCO parameter optimization problem and demonstrate that the optimizer can suppress the instability. Finally, we consider a more realistic LCO speed and stability-constrained mass minimization problem for an airfoil. The proposed method can be extended to optimization problems with a PDE-based model, opening the door to applications where high-fidelity models are needed.

show abstract

Data-Driven Forecasting of Postflutter Responses of Geometrically Nonlinear Wings

Cited by 33 publications

References 50 publications

Adaptive Sampling for Interpolation of Reduced-Order Aeroelastic Systems

Adaptive Sampling for Interpolation of Reduced-Order Aeroelastic Systems

Data-driven prediction in dynamical systems: recent developments

Adjoint-based limit cycle oscillation instability sensitivity and suppression

Contact Info

Product

Resources

About