RL-based EMS control

Control and optimization
Automobile

Challenge

To increase an electric vehicle's power consumption efficiency and thus to achieve a longer range, the air conditioning system (HVAC) needs to be controlled by adjusting various parameters. The current HVAC system engineering incorporates a fluid dynamics simulator. This, however, is both slow and does not describe the actual system sufficiently. As a result, the control performance obtained from the simulation cannot be realized in the actual system, along with slowed model development speeds.

Approach

To create a dynamic model that simulates the internal condition of the HVAC system, measurement data from an actual electric car's air conditioning system is used. In place of an HVAC simulator, the trained dynamics model is used to predict changes in the HVAC's internal state, and it serves as the basis for development of an efficient air conditioning control model.

Results

As a result, the target temperature can be reached in less time and with less energy than the conventional control method, resulting in improved battery performance for electric vehicles.

🔔 Let's Rock Up! Join in our meetup for Jupyter users

RL-based EMS control

Challenge

Approach

Results

Related Use Cases

Enterprise MLOps for seamless ML/DL model operations in complex processes

Chip Placement with RL

Multi-robot task ＆ Motion planning