Water Quality Prediction System Based on Adam Optimised LSTM Neural Network for Aquaculture: A Case Study in Kerala, India

Accurate water quality prediction (WQP) and assessment have a significant role in making aquaculture production profitable and sustainable. The water quality (WQ) parameters in aquaculture undergo dynamic changes, generally nonlinear and complex. The conventional prediction mechanisms show insufficient and poor accuracy with high computation time. This research work proposes Adam optimized long short-term memory (LSTM) deep learning neural network-based WQP system for aquaculture. The WQ data collected from the aqua-ponds located in Kerala, India, from January 2016 to January 2019 are utilized for training and testing the proposed LSTM-based prediction model. The proposed LSTM model results show that predicted and actual values accurately match and outperform the autoregressive integrated moving average model in terms of prediction accuracy. The results show the viability and effectiveness of utilizing LSTM to accurately predict the aquaculture WQ parameters.