PREDICTION OF FUEL CONSUMPTION CRITERIA OF TRACTOR USING NEURAL NETWORKS AND MATHEMATICAL MODELS

The aim of this study is to predict fuel consumption parameters using artificial neural networks and mathematical models. This research included a study of the effect of three different operational factors on three types of fuel consumption (fuel consumption per unit time (FCT), fuel consumption per unit of area (FCA), fuel consumption per unit draft power (FCP)). The operational factors are tire pressure (50, 100 and 150 kPa), plowing depths (15, 20 and 25cm), forward speeds (0.51, 0.85 and 1.45 m/sec) and three repetitions for each experimental unit. The results indicated that increasing FCP increased plowing depth and forward speed by 48% and 36%, respectively. The results also indicated that a tire pressure of 100 kPa gave the lowest FCT. FCA increased by 54% with an increase in plowing depth. While it decreased by 85% with the increase in forward speed. The lowest FCP was recorded at a depth of 20 cm by 0.49 [kg (kW h)-1]. Increasing the forward speed from 0.51 to 1.45 m/sec led to a decrease in FCP by 135%. Tire pressure of 100 kPa also gave the lowest FCP. The results showed that all models (mathematical and ANN models) had excellent performance for predicting fuel consumption parameters under different field conditions. The ANN technique produced superiority over mathematical models for fuel consumption prediction (FCT, FCA, and FCP) based on the MSE and R2 statistical criteria.