Modeling building occupancy using a novel inhomogeneous Markov chain approach

Accurate modeling of building occupancy is an important issue in building energy optimization, but it is a difficult problem due to its stochastic property. This paper proposed an inhomogeneous Markov chain approach to model building occupancy under two scenarios of multi-occupant single-zone (MOSZ) and multi-occupant multi-zone (MOMZ). In this study, we define the state of the Markov chain as an increment of the number of the occupants, which is quite different from the traditional method using the number of the occupants as the state. Simulations have been done to verify the performance of the proposed method, and mean occupancy profile and four random variables such as time of first arrival are evaluated to assess the advantages of the proposed approach. In the MOSZ scenario, the proposed model performs very well and significantly outperforms the existing agent-based model using the actual measurement data from a real building. The proposed model can also adequately capture the trend of the occupancy dynamics in the MOMZ scenario.