Performance analysis of hybrid ground source heat pump systems based on ANN predictive control

Ground source heat pump system attracts increasing attention worldwide in the past decades. It uses soil as the heat source/sink to supply heating and cooling to buildings. In the cooling dominated area, ground source heat pump system couples with the supplemental heat rejecter to avoid the unbalance underground, which is called the hybrid ground source heat pump system. The control strategy of the ground heat exchanger and cooling tower is very important to optimize the performance of the hybrid ground source heat pump systems. A new control strategy is proposed, which is to compare the cooling water temperature exiting the ground heat exchanger and cooling tower directly. Models based on artificial neural networks are built to realize the proposed control method. Four years’ performance of the hybrid ground source heat pump system controlled based on the new method is calculated and compared with another two frequently used methods. Results show that the new control method is more energy efficient and can make full use of the heat exchange advantage of outdoor air and the soil.