Building energy modelling (BEM) plays a critical role in designing sustainable buildings by minimizing energy consumption and associated costs. To ensure accurate and efficient simulations, this study proposes a graph-based modelling (GBM) strategy for multi-zone buildings. By leveraging the graph-based representation, the model captures interactions between various building components, including thermal and structural elements. To enhance model accuracy, a genetic algorithm (GA)-based optimization technique is employed for parameter calibration. The GA iteratively adjusts model parameters to minimize the discrepancy between simulated and measured data of the rooms’ indoor air temperatures. This approach enables the incorporation of additional building details, such as wall temperatures and specific humidity, leading to more realistic simulations. The effectiveness of the proposed methodology is demonstrated through a case study of a five-room building. The calibrated model predicts indoor air temperatures with appropriate accuracy under both heating and cooling conditions, validating its ability to capture the dynamic thermal behavior of the building.