Development and Validation of Prediction Model for Exhaust Emissions During Tractor Plow Tillage

In this study, to compensate for the constraints of high unit cost of portable emission measurement system (PEMS) and measurement environment, we developed a tractor operation-based emission prediction model. We also evaluated the developed prediction model using validation metrics. In addition to engine load data, correlation analysis was conducted on engine temperature and fuel consumption variables. The results showed a high correlation of more than 0.5 between emissions and engine temperature, and a high correlation of more than 0.5 between emissions and fuel consumption for emissions except CO and THC. The R2 values of the CO, THC, NOx, and PM emission prediction models were 0.81, 0.82, 0.85, and 0.97, respectively, showing good overall predictive performance. The prediction models for CO, THC, NOx, and PM emissions developed using the third-order regression analysis all showed excellent performance with an average absolute percentage error of around 2%. Therefore, the developed emission regression model can be used to predict tractor emissions using various variables. Through the exhaust emissions prediction model developed in this study, eco-friendly technology according to the optimal engine design is expected to increase. In addition, it is expected that agricultural machinery prices will be stabilized and export competitiveness will be secured.