Data assimilation with sensor-informed resampling for building occupancy simulation

Abstract

A building occupancy simulation and estimation simulates the dynamics of occupants and estimates the real time spatial distribution of occupants in a building. It needs a simulation model and a data assimilation algorithm that assimilates real-time sensor data into the simulation model. In our previous works, we presented a graph-based agent-oriented simulation model, and a data assimilation framework based on Sequential Monte Carlo (SMC) methods for efficient real time occupancy estimation involving a large number of occupants. As the occupancy and the building environment size increases, there is a major problem with data assimilation caused by the high dimensional system states. To address this issue, this paper presents a new sensor-informed resampling method which utilizes sensor data to improve resampling of particles. Experiment results show the new method was able to provide better estimation of the system state with a limited number of particles compared to the standard bootstrap filter.