The effect of model structure on the noise and disturbance sensitivity of Predictive Functional Control

An Independent Model (IM) structure has become a standard form used in Predictive Functional Control (PFC) for handling uncertainty. Nevertheless, despite its popularity and efficacy, there is a lack of systematic analysis or academic rigour in the literature to justify this preference. This paper seeks to fill this gap by analysing the effectiveness of different prediction models, specifically the IM structure and T-filter, for handling noise and disturbances. The observations are validated via both closed-loop simulation and real-time implementation and show that the sensitivity relationships are system dependent, which in turn emphasises the importance of performing this analysis to ensure a robust PFC implementation.