Comparative review of hybrid forecasting models for short-term prediction of building thermal load

A comparative review of 13 hybrid models for short-term building thermal load forecasting identifies the EMD-LSTM-Markov model as achieving the highest accuracy in predicting day-ahead heating and DHW demand patterns, though high power swells and spikes are underestimated. The assessment uses historical thermal demand data from Scottish households and weather forecasts to evaluate data-driven models enhanced with state-of-the-art techniques.