Physics-Enhanced Sparse Identification of Nonlinear Oscillator with Coulomb Friction

The identification of the nonlinear governing equations of a single degree of freedom oscillator under harmonic excitation, including Coulomb friction damping, is investigated. The so-called RK4-SINDy approach is enhanced here by incorporating part of the known physics to handle the non-smooth friction law since state-of-the-art approaches fail to capture such nonlinearities. This simple, yet representative, case study is examined using both artificially generated noisy data and data obtained from an experimental setup, in both cases for continuous motion. The obtained results highlight the potential of this framework in nonlinear system identification, given sparsely collected corrupted data, and the benefits of incorporating already known system information.