guest ::
TY - THES AU - Reuscher, Tim Andreas TI - Multi-zone climatization of vehicle cabins using model predictive control PB - Rheinisch-Westfälische Technische Hochschule Aachen VL - Dissertation CY - Aachen M1 - RWTH-2024-05502 SP - 1 Online-Ressource : Illustrationen PY - 2024 N1 - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University N1 - Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2024 AB - For decades, climatization systems for vehicle cabins have been developed by both industry and academia. Especially the rise in electric vehicles has given the topic an increased relevance due to the requirement for energy savings. Multi-zonal control climatization approaches, in which multiple individually climatized zones are maintained within the vehicle, promise increased comfort and potentially improved efficiency. Due to the complexity of the climatization system, its multiple input and output values, and the slow time scales, model predictive control is a promising approach for cabin temperature regulation. However, the use of model predictive control for multi-zonal control has yet tobe researched in depth. This thesis presents an approach for the multi-zone climatization of vehicle cabins using model predictive control. The main contributions of this thesis are an integrated modeling approach including all subsystems, observer designs for both temperature and disturbance estimation, and a control architecture exploiting both model and observer to achieve application-oriented multi-zonal temperature control using mixing flap and blower inputs. For the system model, a recently published multi-zonal gray-box cabin model is combined with low dimensional gray-box representations of the mixing flaps and vent system. Experimental evidence is provided to show that the vapor compression circuit can be separated from the cabin temperature control loop without compromising the controlsystem’s performance. Experimental validation yields good agreement of the model with the system behavior and a clear advantage of the proposed gray-box model with alow-complexity white-box model. Observers are designed to estimate cabin temperatures with reduced sensor setups based on only a single sensor. Multiple setups are compared. Experiments show that a single sensor between the driver and front passenger allows to estimate three zonal temperatures to within ±0.5K of the actual value. A disturbance observer using a full sensor suite is then shown to improve the prediction accuracy ofthe derived models by 12 LB - PUB:(DE-HGF)11 DO - DOI:10.18154/RWTH-2024-05502 UR - https://publications.rwth-aachen.de/record/986972 ER -