Performance Prediction Of Vertical Type Ground-Sourca Heat Pump Via Artificial Neural Network For Summer Season In Ankara

A vertical type ground-source heat pump having depth of 40 m has been set up to be cooled a space in the present study. Energy analysis of the set up system was determined for cooling season as a function of depth. Coeffient of performances of heat pump (COPHP) and the whole system (COPsystem) were calculated as 3,12 and 2,81 for cooling season, respectively. Coefficent of performances obtained from the experiments have been modelled via artifical neural network by using Levenberg-Marquardt (LM) back propagation learning algorithm and Fermi transfer function. R2, RMSE and MAPE values were predicted for cooling season data. It can be concluded that coeffient of performaces of the system will be accurately predicted by this modelling for different cooling conditions