Complex machine dynamics: systematic recurrence quantification analysis of disk brake vibration data

Abstract

Complex machine dynamics, as caused by friction-induced vibrations and related to brake squeal, have gained significant attention in research and industry for decades. Today, remedies heavily rely on experimental testing due to the low prediction quality of numerical models. However, there is considerable lack of in-depth studies in characterizing self-excited oscillations encoded in scalar measurements. We complement previous works on phase-space reconstruction and recurrence plots analysis to a larger data base by applying a novel systematic approach using a large data base. This framework considers appropriate delay embedding, time series partitioning into squealing and non-squealing parts and comparison to operational parameters of the brake system. By means of recurrence plot analysis, we illustrate that friction-excited vibrations are multi-scale in nature. Results confirm the existence of low-dimensional attractors in squealing regimes with increasing values of determinism and periodicity with rising vibration levels. It is shown that the squeal propensity can be directly linked to recurrence quantification measures. Using determinism and the clustering coefficient as metrics, we show for the first time that is possible to predict instabilities in regions of non-squealing conditions.