Intellectual Fertilizer Prediction System for Precision Agriculture Using Optimized Artificial Networks

Abstract

In agriculture, effective fertilizer management is essential to increasing crop yields and developing sustainable agricultural practices. Devices with artificial intelligence (AI) have produced agricultural tools that help farmers achieve controlled and accurate farming. To meet farmer expectations, increase yield production, and oversee agricultural operations, it is more crucial to anticipate fertilizer correctly. A unique Automated Fertilizer Management System (AFMS) using an Optimized Artificial Neural Network (OANN) model, augmented by back propagation and chain rule techniques, is presented in this study using deep learning (DL) models. The proposed method refines the back propagation process and increases learning efficiency by utilizing advanced optimization techniques such as regularization and hyper parameter modifying. During training, the model guarantees precise and effective weight updates by utilizing the chain rule for gradient calculation. A wide range of data, including soil nutrient levels and climate parameters unique to the Indian agricultural context, are used to train the ANN. These attributes are then sent into an "Optimized Artificial Neural Network (OANN)" that uses previously acquired data to forecast fertilizer results. Specifically, back propagation and chain rule techniques are used to modify the OANN weights in order to increase the prediction accuracy of the classifier.