On the neurocomputing based intelligent simulation of tractor fuel efficiency parameters

Abstract

Tractor fuel efficiency parameters (TFEPs) (fuel consumption per working hour (FCWH), fuel consumption per tilled area (FCTA) and specific volumetric fuel consumption (SVFC)) were intelligently simulated. A neurocomputing based simulation strategy (adaptive neuro-fuzzy inference system (ANFIS)) was used to simulate the TFEPs. A comparison was also made between results of the best ANFIS environment and those of another neurocomputing based simulation strategy, artificial neural network (ANN). Field experiments were conducted at plowing depths of 10, 20 and 30 (cm) and forward speeds of 2, 4 and 6 (km/h) using a disk plow implement. Statistical descriptor parameters applied to evaluate simulation environments indicated that the best simulation environment of both ANFIS and ANN were able to perfectly predict the TFEPs. However, the best comprehensive ANN simulation environment with a simple architecture of 2-6-3 was easier to use than three individual ANFIS simulation environments. The ANN results revealed that simultaneous increase of forward speed from 2 to 6 (km/h) and plowing depth from 10 to 30 (cm) led to nonlinear increment of the FCWH from 5.29 to 14.89 (L/h) and nonlinear decrement of the SVFC from 2.95 to 0.67 (L/h kW). Meanwhile, forward speed increment along with plowing depth decrement resulted in nonlinear decrement of the FCTA from 28.13 to 12.24 (L/ha). Interaction of forward speed and plowing depth on the FCWH and SVFC was congruent, while it was incongruent for the FCTA. It is suggested to employ the ANN environment in developing future fuel planning schemes of tractor during tillage operations. Keywords: Adaptive neuro-fuzzy inference system, Artificial neural network, Fuel consumption per working hour, Fuel consumption per tilled area, Specific volumetric fuel consumption