Artificial neural network modeling (ANNs) forcombine harvester header losses

ABSTRACT: Combine function is measured by several factors including power threshing, yield loss and fuel consumption. Certainly, the most important issue is the amount of loss which is influenced by several factors including reel rotational speed, ground speed, and reel height and cutting bar knives quality. Three main treatments for this study were considered. Each factor with three levels was evaluated in a factorial experiment with four replications. In this study, MLP network Feed Forward modeling was used. Input consisted of three layers including reel rotational speed, reel rotational speed and reel height. Output of the artificial neural network determined by head loss. Considering both MSE and coefficient of determination, the descending gradient with momentum was chosen. After training and validation of the network, MSE and coefficient of determination were 8.58729E-05 and 0.837 respectively. The results from neural network were compared with regression equation. Coefficient of determination in regression equation was obtained 0.63. In both methods, efficiency of reel rotational speed was greater than the other factors.