Abstract

In this talk, I will argue that most experimental testing methods commonly used to characterise the behaviour of (nonlinear) systems are analogues to direct initial-value simulations and thus present numerous drawbacks when studying the dynamics of nonlinear systems. Alternative methods, considering tests in closed-loop conditions, have demonstrated their superiority over traditional approaches and are becoming increasingly popular. Control-based continuation (CBC) is one of these methods. It uses feedback control to apply the principles of numerical continuation directly to a physical system and obtain experimental bifurcation diagrams without the need for a mathematical model. I will introduce the principles of CBC and illustrate some of its capabilities on a mechanical structure exhibiting modal interactions. I will then discuss some of the challenges associated with the method and briefly present some recent developments improving the method’s robustness (using local surrogate models) and systematic control design (using adaptive nonlinear control). I will then conclude by discussing how the data gathered with CBC is helpful for nonlinear model development, parameter estimation, and validation. I will illustrate this discussion using limit cycle oscillation data collected on a 2-DOF aeroelastic rig tested in a wind tunnel, comparing mechanistic and physics-based machine-learning models. Bio: Dr Ludovic Renson is a Senior Lecturer in the Department of Mechanical Engineering at Imperial College London, where he leads a research group on nonlinear dynamics and control. Ludovic obtained his PhD at the University of Liege (Belgium) in 2014. Ludovic was a postdoctoral research fellow at the University of Bristol before his appointment at Imperial in 2019.