An IoT-based water contamination analysis for aquaculture using lightweight multi-headed GRU model

Area of Science:

• Aquatic Health
• Environmental Monitoring
• Machine Learning Applications

Background:

• Water quality is critical for aquaculture productivity and fish survival.
• Water contamination poses significant risks to aquatic ecosystems.
• Early detection of water contamination is essential for effective management.

Purpose of the Study:

• To develop a smart monitoring system for detecting water contamination in aquaculture.
• To address limitations in existing deep learning models for time-series water quality data.
• To propose an enhanced Multi-Headed Gated Recurrent Unit (MHGRU) model for improved classification accuracy.

Main Methods:

• Utilized Internet of Things (IoT) devices for real-time water quality data collection.
• Developed a Water Toxic Index (WTI) to categorize water contamination levels.
• Implemented an enhanced light-weight Multi-Headed Gated Recurrent Unit (MHGRU) model to capture spatial and temporal dependencies.

Main Results:

• The proposed MHGRU model achieved 99.7% accuracy on real-time data.
• The model demonstrated high performance on a public dataset, reaching 99.12% accuracy.
• Outperformed existing Artificial Neural Network (ANN) models, which achieved accuracies of 99.52% and 98.71%.

Conclusions:

• The smart monitoring system with the enhanced MHGRU model is highly effective for detecting water contamination.
• The model's ability to capture temporal and spatial dependencies leads to superior prediction accuracy.
• This approach offers a robust solution for maintaining water quality in aquaculture.