Prediction of Protein Content in Pea (Pisum sativum L.) Seeds Using Artificial Neural Networks

Pea (Pisum sativum L.) is a legume valued mainly for its high seed protein content. The content of pea characterized by lysine and low allergenicity. This has made consumers appreciate peas increasingly in recent years, not only their taste, but also nutritional value. An important element cultivation the ability to predict content, even before harvest. aim this research was develop linear non-linear model predicting percentage seeds perform comparative analysis effectiveness these models. focused on identifying variables with greatest impact included method machine learning (artificial neural networks) multiple regression (MLR). input parameters models were weather, agronomic phytophenological data from 2016–2020. predictive properties verified using six ex-post forecast measures. (N1) outperformed (RS) model. N1 had an RMS error magnitude 0.838, while RS obtained average value 2.696. MAPE 2.721 8.852, respectively. sensitivity performed best network showed that independent most influencing soil abundance magnesium, potassium phosphorus. results presented work can be useful study crop management. In addition, they help preserve country’s security.