Prediction of Power Requirements and Soil Compaction in Random Traffic Farming in Northern Iraq by Using Neural Networks Method

Abstract Random traffic farming (RTF) is an approach in Iraq‘s cropping practices where uncontrolled machinery traffic frequently causes soil compaction. Predicting machinery draft force and resulting compaction under random traffic farming is therefore essential for improving machine efficiency and enhancing long-term soil sustainability. This study applies artificial neural networks (ANNs) for predicting draft force, soil penetration resistance, and bulk density under various operational parameters. These parameters included tractor mass (3000 kg vs 6000 kg), traffic intensity (0–3 passes), and tillage depth (150 mm vs 250 mm). Experimental fieldwork was conducted on silty clay soil at Ninawa governorate. Draft force (DF) was directly measured, while soil compaction was evaluated using soil penetration resistance (SPR) and bulk density (BD). Field experiment results indicated that machinery traffic was the most influential factor, a first pass increased SPR and DF by up to 76% and 113%, respectively. The ANN model demonstrated high predictive accuracy for DF (R = 0.985) and SPR (R = 0.858), though BD predictions were less accurate (R = 0.631). These findings highlight that ANN modelling is an effective tool for optimizing machinery use and traffic management in RTF systems, thereby supporting sustainable soil management in arid regions.