Stochastic occupancy-integrated MPC for multi-objective optimal built environment control

Efficient built environment control is essential for balancing energy consumption, thermal comfort, and indoor air quality (IAQ), especially in spaces with highly dynamic and intermittent occupancy patterns. Traditional control strategies, such as fixed schedules or simple occupancy-based rules, often fail to address the stochastic nature of occupancy behaviors, leading to suboptimal performance. This study proposes a stochastic occupancy-integrated model predictive control (MPC) strategy that advances built environment optimization through several innovative contributions. First, the proposed MPC integrates stochastic occupancy number predictions into its control scheme, enabling multi-objective optimization considering thermal comfort and IAQ for spaces with sudden occupancy changes and irregular usage. Second, the stochastic differential equations (SDE)-based building dynamic models are developed considering the stochasticity and time-inhomogeneity of occupancy heat gains and CO₂ generations in the prediction of indoor temperature, CO₂ concentration and energy consumption. Third, a TRNSYS-Python co-simulation platform is established to evaluate the MPC strategy’s performance, addressing the discrepancies between the SDE models used for MPC and the actual process of the target system. Finally, the study comprehensively evaluates the MPC’s multi-dimensional performance under different optimization weight combinations and benchmarks it against two baseline strategies: a fixed-schedule (FIX) strategy and occupancy-based control (OBC) strategies with varying per-person fresh airflow rates. Simulation results demonstrate that the proposed MPC achieves 32% energy savings and 17% IAQ improvement compared to the FIX strategy, and 30% thermal comfort improvement and 20% IAQ improvement with the same energy consumption compared to OBC. These findings highlight the robustness and enhanced performance of the proposed MPC in addressing the complexities of stochastic and time-varying occupancy, offering a state-of-the-art solution for energy-efficient and occupant-centric built environment control.