Reinforcement learning for heat pump regulation (RLWP)

Development of a control system for heat pumps based on reinforcement learning (RL). Unlike the current control system, the optimal controller settings and setpoints are used for each operating point, thereby significantly improving operational reliability and efficiency.

Key data

Description

This project aims to develop a control system for heat pumps based on reinforcement learning (RL). Unlike the current control system, the optimal controller settings and setpoints are used for each operating point, thereby significantly improving operational reliability and efficiency.

Furthermore, the dynamic simulations (digital twin) of the heat pump required to train the RL agent represent highly valuable expertise that can be leveraged after the project to make progress in other areas, such as predictive maintenance.

This not only creates a future USP due to the significantly improved operational reliability but also paves the way for strategically important further developments.