Research on construction and optimal scheduling strategy of machine learning-driven energy consumption prediction model for HVAC systems in office buildings

Promoting building energy conservation is a key path to achieving the dual-carbon goal. The heating, ventilation, and air conditioning (HVAC) system in office buildings is a core energy-consuming terminal, and its accurate prediction and optimal scheduling are of great significance for improving energy utilization efficiency and reducing operating costs. Aiming at the limitation that traditional methods are difficult to adapt to dynamic loads and complex nonlinear coupling characteristics, this paper constructs a machine learning-driven integrated framework for HVAC system energy consumption prediction and optimal scheduling. Based on multi-dimensional data from authoritative measured databases and public building energy consumption monitoring platforms, typical machine learning prediction models are systematically constructed and compared to achieve high-precision prediction of short-term cooling and heating loads as well as system energy consumption, providing reliable feedforward support for optimal scheduling. Taking the prediction results as the foundation, an optimal scheduling model based on model predictive control (MPC) integrated with dynamic electricity price mechanism is established, targeting the coordination of operating cost optimization and power grid peak load shifting, with indoor thermal comfort standards as hard constraints. Case verification through professional simulation platforms shows that the proposed integrated method effectively improves the electrical load characteristics and enhances the operational economy and low-carbon performance of the system while ensuring indoor environmental quality, providing a technical path with theoretical value and practical significance for the intelligent and low-carbon upgrading of HVAC systems in office buildings.