Enterprise AI Analysis
Prediction of Tractor Drawbar Pull Under Different Tillage Tools Using Machine Learning and Low-Cost Sensors
Machine-learning models were developed to predict the drawbar pull of a 78-kW-class tractor for moldboard, chisel, and subsoiler plows. Four models were tested: random forest (RF), extreme gradient boosting (XGB), artificial neural network (ANN), and support vector machine (SVM). The training variables included engine speed (ES), engine torque (ET), travel speed (TS), tillage depth (TD), and slip ratio (SR). Unlike prior studies that focused mainly on engine parameters, this study incorporated nonlinear variables to improve both accuracy and practical applicability. Data were collected from three Korean paddy fields with different soil conditions, and the dataset was divided into 70% for training and 30% for testing. Five input variable combinations were used: Model A (ES, ET), Model B (ES, ET, TD), Model C (ES, ET, TS, SR), Model D (TD, TS), and Model E (ES, TD, TS). The results showed that, for the moldboard plow, RF in Model E achieved the highest performance (R2 = 0.977). For the chisel plow, ANN in Models B and C provided strong predictive accuracy (R2 = 0.953). The subsoiler also performed well with ANN in Models B and E (R2 = 0.953). Overall, the proposed models-particularly RF and ANN—proved effective in predicting drawbar pull and outperformed XGB and SVM. This study is distinguished by its comparison of various input variable combinations for different plows (moldboard, chisel, and subsoiler) and by its proposal of a cost-effective approach using low-cost sensors.
Quantifiable Enterprise Impact
This research provides a framework for predictive maintenance and operational optimization in agricultural machinery, leading to significant improvements in efficiency and cost-effectiveness.
0
R² (Highest Accuracy)
0
Peak R² Accuracy (Moldboard Plow)
0
Peak R² Accuracy (Chisel/Subsoiler)
0
Cost Reduction vs. Conventional Sensors
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Enhanced Predictive Accuracy Across Tillage Tools
The study developed machine learning models (RF, XGB, ANN, SVM) to predict tractor drawbar pull for moldboard, chisel, and subsoiler plows. For the moldboard plow, RF in Model E achieved the highest performance (R² = 0.977). For the chisel plow, ANN in Models B and C provided strong predictive accuracy (R² = 0.953). The subsoiler also performed well with ANN in Models B and E (R² = 0.953). Overall, the proposed models—particularly RF and ANN—proved effective in predicting drawbar pull and outperformed XGB and SVM. The analysis highlighted that increasing the number of input variables (engine speed, engine torque, tillage depth, travel speed, slip ratio) improved prediction accuracy, especially when incorporating non-engine parameters.
Enterprise Process Flow
Data Acquisition
→
Model Development (ML, Optuna)
→
Validation (Test Set, ASABE, Robustness)
Significant Cost Reduction in Data Acquisition
This study distinguished itself by proposing a cost-effective approach using low-cost sensors. While a conventional load cell costs approximately $5,500, the total cost for the potentiometer, proximity sensor, and GPS receiver used in this study was only about $220. This represents a significant cost reduction, making the system accessible at roughly 4% of the conventional cost, even accounting for potential fluctuations.
0
Total Cost of Low-Cost Sensors (USD)
Machine Learning Outperforms Traditional Models
A comparison was made between the ASABE drawbar pull equation and the machine learning models. The ASABE equation showed very low coefficients of determination (R² = 0.029-0.158) across all plow types, indicating its limited ability to explain variations in drawbar pull, performing even worse than its reported error ranges (±40-50%). This discrepancy is attributed to regional soil differences and the ASABE model's reliance on implement-related factors without accounting for key powertrain variables like engine torque, speed, and slip ratio. In contrast, ML models achieved substantially higher R² values, with ANN performing best for chisel (0.950) and moldboard (0.901) plows, and RF/ANN for subsoiler (0.749/0.754).
| Plow Type | ASABE Standard R² | ML Model D (RF R²) | ML Model D (XGB R²) | ML Model D (ANN R²) | ML Model D (SVM R²) |
|---|---|---|---|---|---|
| Moldboard | 0.158 | 0.846 | 0.718 | 0.901 | 0.741 |
| Chisel | 0.155 | 0.861 | 0.805 | 0.950 | 0.805 |
| Subsoiler | 0.029 | 0.749 | 0.590 | 0.754 | 0.592 |
Key Insights for Agricultural AI Implementation
The study found that tractor traction performance is highly dependent on the type of plow and the combination of input variables. Increasing input variables (engine speed, engine torque, tillage depth, travel speed, slip ratio) consistently improved prediction accuracy. The RF model consistently achieved superior performance across all plow types, though with a slight risk of overfitting, while ANN showed strong performance for chisel and subsoiler. The low-cost sensor approach was validated as a viable and highly cost-effective alternative to conventional load cells.
Calculate Your Potential ROI
Estimate the efficiency gains and cost savings for your enterprise by implementing AI-driven agricultural predictive analytics.
Projected Annual Savings
Annual Cost Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
A structured approach to integrating AI-driven predictive analytics into your agricultural operations.
Phase 1: Data Infrastructure & Sensor Integration
Establish robust data pipelines for real-time sensor data (engine, speed, depth, slip). Integrate low-cost sensor solutions as validated in this research, ensuring data quality and connectivity across your tractor fleet.
Phase 2: Model Training & Customization
Leverage existing research models (RF, ANN) as a baseline. Train and fine-tune machine learning models with your specific operational data, customizing for different tillage tools and regional soil conditions to optimize predictive accuracy.
Phase 3: Real-time Predictive Analytics & Alerts
Deploy the validated models for real-time prediction of drawbar pull. Implement an alert system for operators to optimize settings, prevent excessive slip, and reduce fuel consumption, translating predictions into actionable insights.
Phase 4: Continuous Optimization & Scalability
Establish a feedback loop for continuous model improvement. Expand the system across diverse agricultural operations and integrate with broader farm management platforms to scale efficiency gains.
Ready to Transform Your Agricultural Operations with AI?
Schedule a personalized consultation to explore how our AI solutions can enhance efficiency, reduce costs, and optimize performance for your enterprise.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat