Hybrid surrogate model for the prediction of uncertain friction-induced instabilities

Abstract

This paper presents a hybrid surrogate model for the prediction of friction-induced instabilities in uncertain mechanical friction systems. The proposed hybrid meta-model is developed in order to predict the occurrence of mode coupling instabilities submitted to random and interval parameter uncertainties. This predictor is built from the combination of the generalized polynomial chaos formalism, known to be useful to deal with random uncertainties, together with the inclusion function based on Chebyshev polynomials used to deal with interval uncertainties. The feasibility of the proposed approach and its efficiency are assessed by investigating the stability analysis of a four degree-of-freedom model with two sets of uncertain parameters described by probabilistic and interval models. Numerical results are compared with those obtained by applying a classical parametric approach to demonstrate the efficiency of the proposed methodology. The suggested hybrid meta-model is then shown to have an interesting potential to deal with stability analysis of mechanical systems subjected to friction-induced vibration.