Data-based hybrid modelling of the component placement process in pick-and-place machines

In this paper we present an experimental study in the identification of the electronic component placement process in pick-and-place machines. Unilateral contact and saturation phenomena characterize the hybrid dynamics of the system. Furthermore, the mode switch cannot be measured and identification algorithms for hybrid systems, that are capable to reconstruct both the modes and the switching law, must be used. Piece-Wise AutoRegressive eXogenous (PWARX) models, which consist of a number of ARX modes together with the partition of the regressor space into regions where each model is valid are identified. We show that the reconstructed models are able to capture the relevant dynamics of the experimental setup. Practical insights on hybrid system identification and comments on possible improvements of the identification algorithm complement the quantitative results.