Hybrid deep‐Q Elman neural network for crop prediction and recommendation based on environmental changes

Abstract

AbstractCrop recommendation is a potential research topic that relies on environmental conditions such as temperature, humidity, rainfall, and soil pH to identify suitable crops for cultivation. There are diverse models available in the literature for crop recommendation. Still, those models are not accurate enough to predict the appropriate crop when there is a sudden change in the environmental factors. The models cannot map the raw data exactly with the prediction values, and the output relies on the quality of the input features used. To resolve these issues, the concept of Q‐learning is hybridized with deep learning to enable exact mapping of the raw data with the prediction values. In this article, Q‐learning is combined with the Elman neural network, trained with the input parameters selected by the improved Archimedes optimization algorithm from the dataset. The model evaluations are carried out with the user dataset constructed using the sensor information collected from the regions of Maharashtra. The overall accuracy provided by the proposed crop recommendation model is 99.44%, and the average inference time of the model is 0.0117 s.