Effects of Time Horizon on Model Predictive Control for Hybrid Electric Vehicles

Abstract

One of the challenges in Model Predictive Control (MPC) for Hybrid Electric Vehicles (HEVs), is real time implementation, Bo-Ah et al. (2012). Computation time can be reduced by limiting the time horizon of the prediction. Limiting the time horizon results in sub optimal control, but may yield nearly optimal control if the time horizon is chosen appropriately. This paper investigates the sensitivity of MPC to predicted horizon length with regard to Fuel Economy (FE). The results show that predicting Driver's Desired Power (DDP) for the next 10 seconds on the highway and 20 seconds in the city, is sufficient for MPC to perform close to the Globally Optimized Controller (GOC). In other words: Regarding fuel economy optimization on the highway, knowing DDP for the next 10 seconds is almost equivalent to knowing the DDP for the whole trip.