eprllib.Agents.Rewards#
Reward functions in eprllib#
The reward functions are called in the step() method in the environment iteratively for
each agent present in the respective timestep.
This module contains the reward functions used in reinforcement learning and implemented in eprllib.
You can implement your own reward function by creating a new class that inherits from
BaseReward and overriding the calculate_reward method.
The module includes the following classes:
BaseReward: The base class for creating reward functions.
ComfortRewards: Contains classes to calculate rewards based on comfort.
EnergyRewards: Contains classes to calculate rewards based on energy.
CombinedRewards: Contains classes that combine comfort and energy methods to calculate rewards.
These classes are used to provide feedback to the agents, guiding them to optimize their policies based on the defined reward criteria.
Modules
Reward Function |
|
Comfort rewards |
|
Energy and ASHRAE 55 Simple Model reward function |
|
Energy and CEN 15251 reward function |
|
Energy rewards |