Adaptive Control of Thermal System in Electrified Heavy Vehicles - Investigation in how adaptive parameters used in predictive control can minimize model prediction errors

Abstract

Abstract The development of battery electrical heavy vehicles is an important factor in reaching climate change goals and increasing the standard of living by reducing emissions and noise pollution. The cooling system serves as a crucial component for keeping parts in their ideal operating window which reduces wear, as well as increases performance. Applying predictive control to these vehicles is another step in improving performance with reduced power consumption and increased life expectancy of components. This work investigates the possibility of applying adaptive parameters to be used in predictive controllers to improve predictive estimates and thus also the control. We use estimation techniques, such as recursive least squares (RLS) and extended Kalman filtering (EKF) to estimate the adaptive parameters that can be fed back to a supervising controller, such as a model predictive controller (MPC). Results show that using adaptive parameters improves predictions with reduced prediction errors. The results indicate the possibility of accuracy improvements that combined with improved control structure could lead to better performance.