A Study of a Model for Predicting Pneumatic Subsoiling Resistance Based on Machine Learning Techniques

In order to explore the drag reduction mechanism of pneumatic subsoiling and study influence on soil, this used machine learning models predict working resistance a subsoiler adopted random forest (RF), error back-propagation (BP), eXtreme gradient boosting (XGBoost) support vector regression (SVR) analyze compare predictions these four models. Field experiments were carried out in two fields with different bulk densities moisture content. The effects parameters studied by changing air pressure, depth forward speed. RF, SVR, XGBoost BP models, five (working depth, speed, density content) inputted as independent variables, operating was predicted value. After training results showed that R2 value RF model highest smallest, which made it better than values MAPE, RMSE for model’s test set 0.01, 0.99, 3.61 N, respectively, indicating could well. When input parameters, experimental contribution pressure reducing reached 29%, can reduce resistance, consumption.