Highlights The database is established with an experimental study of the state of the Soil. The programming and the training showed a performance of the models of soil resistance according to variables. Resistance of penetration increases by increasing the number of passages. The highest penetration resistance values are observed with soil moisture contents of around 15%. A concordance of the intelligent prediction model compared to the experimental state.
Abstract The investigation and evaluation of the phenomenon of soil compaction after the passage of repetitive equipment (Tractor, cultivator..) is considered as a preventive solution to preserve agricultural soils against degradation and maintain sustainable agriculture. In reality, there is currently no reliable method for predicting the primary causes of soil compaction, particularly in moist soils. This study applicates Artificial Neural Network (ANN) modeling to make a resistance penetration prediction. Resistance penetration (Rp) test data acquired from measured experimental values are used to train the models. The learning score coefficient (0.96), the RMSE (0.51) and MAE (0.39) show that the forecasts from the ANN models coincide with the measured field data. It is concluded that the developed ANN models would be used effectively to make predictions that are more accurate on the soil state compaction in different moisture conditions. This work will help the farmers to optimize the machine use and hence to enhance production and improve yields with minimal costs.
Machine learning approach for predicting soil penetration resistance under different moisture conditions
Journal of Terramechanics ; 110 ; 39-45
2023-08-09
7 pages
Article (Journal)
Electronic Resource
English
Utilizing Machine Learning for Cone Penetration Test-Based Soil Classification
Transportation Research Record | 2024
|Moisture penetration in microcircuit packages
NTRS | 1980
|