An Improved Data Fusion Model for Secondary Return Water Temperature of Heating System Employing EKF

Predicting the secondary return water temperature in district heating systems is vital for boosting energy efficiency. The intricate nature, nonlinear dynamics, and pronounced time delays in these systems, however, complicate precise forecasting. To overcome these obstacles, we introduce an enhanced data fusion approach rooted in heat transfer fundamentals, coupled with the extended Kalman filter (EKF) for handling nonlinear state estimation. This method adeptly captures time delays and estimates return water temperature using physical principles, while minimizing reliance on extra sensors in the secondary network. Experimental findings reveal that our approach surpasses alternative techniques, delivering reduced Root Mean Squared Error (RMSE) and superior prediction precision, all with straightforward computation. By merging physical insights with EKF, this solution effectively resolves challenges posed by nonlinearity and delays, providing a practical tool for optimizing heating system performance.