Thermal management in electric vehicles (EVs) is fundamental to balancing battery safety, lifespan, and passenger comfort. As vehicle architectures become more complex, integrated systems capable of managing multiple thermal domains simultaneously are required. This thesis develops a supervisory predictive control architecture representing the upper layer of a two-stage control system. The supervisor, based on an adaptive model predictive control (AMPC), calculates the optimal compressor power and the refrigerant split factor, which are then implemented by a low-level control system to regulate compressor speed and the distribution of heat flow rates between the evaporator and the chiller. The core of the research involved the development of a control-oriented model of the refrigeration cycle, designed to estimate the future cooling capacity along the prediction horizon.