[Comparative Study of Water Quality Prediction Methods Based on Different Artificial Neural Network].

Area of Science:

• Environmental Science
• Data Science
• Water Resource Management

Context:

• Accurate water quality prediction is crucial for effective watershed management and regional planning.
• Traditional methods like Back Propagation Neural Networks (BPNN) can suffer from overfitting.
• Advanced modeling techniques are needed to improve prediction accuracy and reliability.

Purpose:

• To compare the predictive performance of BPNN, particle swarm optimization-modified BPNN (PSO-BPNN), and Convolutional Neural Networks (CNN) for water quality index prediction.
• To evaluate the effectiveness of different neural network architectures in handling complex environmental data.
• To identify the most suitable model for accurate permanganate index prediction in the Xiangjiang River Basin.

Summary:

• The study evaluated BPNN, PSO-BPNN, and CNN for predicting water quality in the Xiangjiang River Basin.
• PSO-BPNN demonstrated improved stability over traditional BPNN by mitigating overfitting.
• CNN achieved superior prediction accuracy, with lower RMSE and MAE, and higher R² compared to both BPNN and PSO-BPNN, indicating a more robust fitting method.

Impact:

• The findings highlight CNN as a highly effective tool for water quality forecasting, enhancing regional planning and watershed management capabilities.
• Improved prediction accuracy can lead to better environmental protection strategies and resource allocation.
• This research provides a benchmark for applying advanced machine learning models in environmental monitoring and management.