Building energy consumption has become an important component of global energy consumption, especially the cooling and heating loads of buildings have a significant impact on energy demand. Traditional building load simulation methods mostly target a single building type, while for load simulation of multiple types of buildings, it is usually necessary to establish separate models for each building type. Therefore, this paper proposes a multi-task learning-based simulation method for the heating and cooling loads of multi-type buildings. Firstly, by calculating the correlation coefficient between meteorological parameters and cold and hot loads, the input features most relevant to the load are selected. Then, based on the Multi-Task Mixed Expert Model (MMoE) framework, a unified model was constructed that can simultaneously simulate the heating and cooling loads of multiple building types. Finally, based on simulation data, a case study was conducted to verify the effectiveness of the proposed method. The experimental results show that the proposed method has good simulation performance, with an overall simulation error RMSE of 2.180 W/m².
Liu et al. (2026) studied this question.