Artificial Neural Network and Stepwise Approach for Predicting Tractive Efficiency of the Tractor (Case Jx75t)

Abstract

The aim of this study is to develop and predict models of tractive efficiency using the artificial neural network and stepwise approach. The tractive efficiency of tractor (CASE JX75T) was measured experimentally. Experiments were conducted in the site of Basrah University. Which had silty clay soil texture. The field conditions included effect of two level of cone index (550 and 980 kPa), two level of moisture content (8 and 21%), three forward speeds (0.54, 0.83 and 1.53 m/s) and four level of tillage depths (10, 15, 20 and 25 cm). The results illustrated that both developed models (stepwise approach and ANN technique) had acceptable performance for predicting tractive efficiency of tractor under various field conditions. However, ANN model outperformed stepwise model, where 4-7-1 topology showed the best power for predicting tractive efficiency with R-squared of 0.97 and MSE of 0.0074 with Levenberg-Marquardt training algorithm. The analysis of variance demonstrated that the studied parameters had single significant effect on tractive efficiency. The most parameter influential on tractive efficiency was tillage depth followed forward speed, cone index and moisture content.

Smart Computing Techniques for Predicting Soil Compaction Criteria under Realistic Field Conditions

“…The appropriate model for the cone index is represented in Eq.7, in which the coefficients are in the coded unit form. Almaliki et al (2019) during their study to predict the tractive efficiency.…”

“…Intelligent computing technology has been used in various disciplines and fields of research in computational sciences with various software technologies such as statistics, machine learning, artificial neural networks (ANN), analysis of fuzzy data, and artificial intelligence -to solve many problems and manage technical processes in all kinds of medical and engineering sciences (Almalki et al, 2016;Kamilaris & Prenafeta-Boldú, 2018;Shafaei et al, 2018;Almaliki et al, 2019;Almaliki et al, 2021;Monjezi, 2021;Monjezi & Hosseinzadeh, 2021).…”

“…Artificial Neural networks were used by Almaliki et al (2019) to predict the tractive efficiency of the tractor during the tillage process and correlate this with a set of influencing factors such as soil penetration resistance, forward speed, and different tillage depths. These techniques gave high compatibility with the presented experimental data.…”