Artificial neural network based modelling of performance of a beta-type Stirling engine

In this article, artificial neural network has been used in order to predict the power (P) and torque (T) values obtained from a beta-type Stirling engine that uses air as working fluid. Experimental data have been obtained for different charge pressures and hot source temperatures using ZrO2-coated and uncoated displacers. The closest artificial neural network results to experimental torque and power values were obtained with double hidden layer 5-13-9-1 and 5-13-7-1 network architectures, respectively. The best prediction values were obtained by Levenberg-Marquardt learning algorithm. Correlation coefficient (R-2) for the torque values were 0.998331 and 0.997231 for the training and test sets, respectively, while R-2 value for power values were 0.998331 and 0.997231 for the training and test sets, respectively. R-2 values show that the developed artificial neural network is an acceptable and powerful modelling technique in predicting the torque and power values of the beta-type Stirling engine.