A Reinforcement Learning Based Approach to Recommend Appropriate Occupant Behaviours

To the best of our knowledge, many machine learning techniques are not yet investigated for building Occupant Behaviors (OB) modeling tasks like advising the best decision to save energy. For instance, Reinforcement Learning (RL) with its key concept of reward could be a good fit for occupant education towards a saving energy. The proposal is an application of RL and its variants to help building occupants adjusting their behaviors to reduce electricity consumption while preserving indoor comfort. We propose a RL framework, where an agent interacts with building environments through tailored states, rewards and actions in a way to derive an optimal policy for OB.