Development of a group control strategy based on multi-step load forecasting and its application in hybrid ground source heat pump

Development of a novel group control strategy applied in hybrid ground source heat pump system (HGSHPs) is the purpose of the study, which helps to achieve a high-efficiency operation in cooling dominated areas. A HGSPHs model, based on the ground source heat pump system (GSHPs) in laboratory, was improved. A LSTM-based multi-step load forecasting model, as the core of group control strategy, was proposed and further opti-mized on network structure with the optimal MAE and MAPE of 982 W and 7.38%. The on-off and speed regulation of all units in improved HGSHPs are dynamically controlled in advance according to the results of forecasted load sequences. The energy saving advantages of the improved HGSHPs are reflected in low-load condition, accompanying with a 12.4% and 19.8% reduction in energy consumption for cooling compared with conventional HGSHPs and HGSHPs with on-off control strategy. A long-term operation for ten years of the conventional and improved systems were simulated and comparative evaluation of soil temperature rise was given with the values of 0.37 degrees C and 0.41 degrees C, respectively. Less energy performance degradation caused by soil heat accumulation and better energy-efficiency are attained of the HGSHPs with group control strategy.