Data-driven estimation of time-dependent solar gain coefficient in a two-zone building with synthetic occupants: Two B-splines integrated grey-box modeling approaches

Solar gains are a significant share of the heat input to the buildings' interior energy balance. Therefore, it is important to estimate the dynamic details of solar gains precisely for predictive works, such as in model predictive control (MPC). Effective solutions to gauge the dynamic details of solar gains are still lacking. To fill this gap, a data-driven B-splines integrated grey-box modelling technique has been presented recently, and its potential to estimate dynamic solar gains based on the limited size of low-frequency on-site data has been demonstrated in a simplified one-zone building case, with a single level on/off heating input signal. Its robustness is examined in a two-zone building with heating inputs of synthetic occupants, in this study. Two thermal zones are established to mimic day and night zones in a test building, and corresponding synthetic occupancy profiles and thermostat systems have been implemented in each zone. This study shows that the Bspline integrated grey-box model is useful in this more realistic case considering the impacts of occupants, but demands more advanced modellings. Specifically, two updated solutions on B-splines integrated grey-box modelling are offered, although challenges in the model selection are noted, mainly attributed to the low statistical significance of solar gain, which had fortunately almost no impact on the effectiveness of the proposed techniques to characterize solar gain dynamics. The reported challenge could be further investigated in the future based on more case studies.

Automated summary

This study demonstrates the potential application of a data-driven B-splines integrated grey-box modelling technique for estimating dynamic solar gains in real buildings. Improved solutions are proposed to consider the impact of occupants.