Applying an artificial neural network approach to the analysis of tractive properties in changing soil conditions

Abstract

For better performance of a micro-tractor during agricultural operations, it is necessary to select the optimal tractor configuration for the operation. The purpose of this study was to analyse the effects of soil texture, soil moisture and compaction as well as horizontal deformation and vertical load on traction force and traction efficiency. Analysis and mathematical modelling were performed with the use of artificial neural networks (ANN). Accurate mathematical models were obtained with high values of coefficient of determination R2 for the validation data set (R2=0.945 for traction force and R2=0.963 for traction efficiency). Based on neural models, analysis of the contribution of independent input variables was performed. Soil texture and soil moisture had the highest influence on traction force and traction efficiency; vertical load significantly affected traction force. Horizontal deformation and soil compaction had minor influences on both dependent variables. Evolutionary algorithm was used for the determination of soil conditions and vertical load which produce high traction force and traction efficiency. The vertical load is considered as an easily managed parameter during agricultural operations. Since the increase in vertical load results in increasing traction force and, at the same time, in decreasing traction efficiency, the appropriate vertical load value was calculated for optimization of traction force and traction efficiency in changing soil conditions, which is crucial for tillage management.