Authors: Hadi Taghavifar Ehsan Shabahangnia
Publish Date: 2014/04/26
Volume: 50, Issue: 11, Pages: 1515-1524
Abstract
Free convection from cubical air channel equipped with copper plate was taken into consideration in the presence of electrostatic field as the channel position was varied The paper examines the artificial neural network capability in modeling and prediction of five output parameters for plate temperatures as affected with four input parameters of heat flux applied voltage the inclination angle of channel and inlet ambient temperature The proposed network’s performance was measured with increasing number of neurons in hidden layer The best network structure was found 4205 with Levenberg–Marquardt training algorithm and mean squared error of 006366 The mean relative error for all output cases were 29 The best coefficient of determination was resulted at 30 cm from channel entrance section to the amount of 09807 The discrepancies of the results are chiefly attributed to 90° channel inclination angle The network was able to predict accurately temperature trend of airflow and plate with voltage heat flux and channel positions
Keywords: