Two ANN models with MLPNN and RBFNN architectures have been developed in this study to predict the evaporating temperature, cooling capacity and the compressor power of an AAC system. 108 sets of experimental data in steady state condition have been obtained by conducting several tests on the AAC test rig. These experimental data were used to train and test the developed ANN models and the performance of the models was evaluated based on three performance indicators: MSE, RMSE and correlation coefficient R.