Data-driven structural identification of nonlinear assemblies: Asymmetric stiffness and damping nonlinearities

Nonlinear model identification for mechanical structures is a challenging task, particularly when the structure exhibits asymmetric nonlinear behaviour related to both stiffness and damping. In this paper, a new method for parametric nonlinear model identification of structures from measured data is proposed by cascading optimisation problems defined for model selection and parameter estimation steps. Model selection is firstly carried out using an algebraic-based cost function to extend the equation of motion of the underlying linear system to include nonlinearities. Afterwards, parameter tuning is conducted by utilising a simulation-based cost function that relies on the instantaneous characteristics of the measured response. A numerical study is conducted to investigate the behaviour of the optimisation problems used for the identification. Further, the validity of the proposed method is demonstrated based on the responses of the systems with symmetric and asymmetric nonlinearities. Then, the method is successfully applied to experimental data from a pylon subassembly mockup structure in order to identify its asymmetric nonlinear damping and stiffness effects.