Enhanced LSTM-based AI model for accurate dissolved oxygen prediction in aquaculture systems

Accurate monitoring and prediction of dissolved oxygen (DO) levels in aquaculture systems are crucial for maintaining optimal water quality and ensuring fish health. This study presents an enhanced Long Short-Term Memory (LSTM)-based model for DO prediction, leveraging historical data on DO levels, water temperature, and other environmental parameters. Unlike traditional methods that rely on fixed assumptions, the proposed model dynamically adapts to changing environmental conditions, offering real-time, high-precision forecasts. Experimental results demonstrate that the enhanced LSTM model achieves a prediction accuracy of 92.28 %, outperforming existing models such as IFP (81.97 %), DOE (63.81 %), and DOP (86.79 %). The model’s superior accuracy and adaptability make it a reliable tool for aquaculture management, helping to optimize DO levels and reduce the risk of fish mortality. By integrating artificial intelligence into aquaculture monitoring, this approach contributes to improved system productivity and sustainability.