Power Prediction with Artificial Neural Network in Experimental Organic Rankine Cycle

In the simulation programs that used to estimate the power of the organic Rankine cycle; high error rates may have occurred due to accepting ideal or near-ideal behaviour differ from the actual behaviour of system components. Predictions made via artificial neural networks may be more close to actual results in the system which is of non-linear behaviour. In this study, network was trained by evaporator waste heat input- output temperatures and mass flow rate, cooling fluid input- output temperatures and mass flow rate taken from an experimental organic Rankine cycle. The power prediction was made with trained network and then the experimental and prediction results of the 10 kW organic Rankine cycle was compared. At the end of the study, the values obtained from artificial neural network were compared with experimental data and correlation coefficient which shows performance of network has calculated to be 0.99124. The prediction success of network was also checked via performing different test data input to the network.