Title: Deep LSTM model exploiting optical sensors for soil nutrient prediction

Authors: C.T. Lincy; A. Lenin Fred; J. Jalbin

Addresses: Computer Science and Engineering, Mar Ephraem College of Engineering and Technology, Elavuvilai, Marthandam, Tamil Nadu, India ' Computer Science and Engineering, Mar Ephraem College of Engineering and Technology, Elavuvilai, Marthandam, Tamil Nadu, India ' Computer Science and Engineering, Mar Ephraem College of Engineering and Technology, Elavuvilai, Marthandam, Tamil Nadu, India

Abstract: Artificial intelligence is a quickly expanding field incorporated into nearly every human life aspect. There is an enormous requirement for influential and speedy measurement systems to calculate precise macronutrients in soil, wherein fertiliser application can be spatially regulated concerning crop demand. However, there is a need to develop a potential avenue for research to design a highly developed model to forecast soil properties. For precise and speedy monitoring of soil macronutrients, a portable sensor device is an essential requirement of an agriculture system. Thus, soil nutrients recognised from gathered soil samples by optical sensors are estimated for their accuracy by the deep learning approach in this work. The deep long short-term memory (LSTM) model is utilised for prediction of soil nutrients following augmenting collected soil data. From investigational analysis, it is revealed that deep LSTM model gained least MSE value of 9.106 e^-06, least RMSE value of 0.00301, low MAE value of 0.0008 and low MAPE value of 0.0017.

Keywords: soil nutrient; soil samples; prediction; sensor; pre-processing; deep learning.

DOI: 10.1504/IJEP.2023.139855

International Journal of Environment and Pollution, 2023 Vol.73 No.1/2/3/4, pp.133 - 153

Received: 08 Sep 2023
Accepted: 18 Mar 2024
Published online: 08 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article