A NOVEL HYBRID MODEL TO PREDICT THE DISSOLVED OXYGEN OF THE WATER IN AQUACULTURE

Abstract

Dissolved oxygen (DO) is a critical parameter in aquaculture, directly affecting the health, growth, and survival of aquatic organisms. Accurate prediction of DO levels can help optimize water quality management, prevent fish mortality, and improve productivity. Traditional methods for measuring DO are often time-consuming, costly, or prone to errors. This study proposes a novel hybrid predictive model that integrates machine learning techniques with statistical modeling to forecast DO levels in aquaculture systems. The model combines Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) neural networks to capture both linear and nonlinear dynamics in water quality data, including temperature, pH, turbidity, and ammonia concentration. Experimental results demonstrate that the proposed hybrid approach outperforms conventional standalone models in terms of accuracy, robustness, and computational efficiency. The findings suggest that this predictive framework can serve as an effective tool for real-time monitoring and management of water quality in aquaculture environments, enabling proactive interventions and sustainable aquaculture practices.