Prediction of output power and efficiency for a free piston Stirling oscillator using artificial neural network

The proposed paper describes a new method to predict output power and efficiency of a free piston Stirling oscillator (FPSO) using artificial neural network (ANN). First, the mathematical description of the proposed oscillator is derived. Then, a unified method for developing an ANN is considered to achieve the aim of this work. The input parameters include the strokes of displacer and power pistons, hot and cold sinks, heat transfer coefficient, and phase angle while output power and efficiency are considered as the output parameters of the ANN. Next, the regression and mean square error (MSE) analyses are employed to obtain the appropriate ANN model. Besides, three experimental case studies from NASA and Shiraz University of Technology are taken into account to verify the performance of the proposed ANN. As a consequence, the maximum prediction error for output power and efficiency are found to be less than 3% and 6% respectively. Finally, the proposed scheme can be used as a powerful tool for predicting important parameters of an FPSE.

Keywords:

Stirling oscillators , Artificial neural network (ANN) , Performance prediction , Output power , Efficiency.