Quasi-random-sampling high dimensional model representations for the construction of reduced discrete time state space dynamic models

In the context of real time model-based applications, complex high fidelity models may be computationally too expensive. Model order reduction and system identification techniques have been employed to transform complex models into equivalent reduced order models. However, most of the literature on model order reduction concerns linear time invariant dynamic systems, and the research into non linear model reduction is still on early stage. In this paper, we present a novel approach using quasi random sampling – high dimensional model representation (QRS-HDMR) to generate reduced discrete time state space dynamic models. The approach has the advantages of being able to handle the high dimensional case and produce affine discrete state space models, readily usable in control engineering applications