Predicting dissolved oxygen level using Young's double-slit experiment optimizer-based weighting model.

Area of Science:

• Environmental Science
• Water Resource Management
• Computational Intelligence

Background:

• Dissolved oxygen level (DOL) prediction is vital for environmental health and water resource management.
• Irregularity and volatility of DOL present forecasting challenges, limiting single model performance.
• Existing models often struggle with accuracy, limited applicability, and data acquisition difficulties.

Purpose of the Study:

• To develop a novel weighted model (PWM) for enhanced dissolved oxygen level prediction.
• To overcome the limitations of single predictive models in accuracy and scope.
• To improve the precision of water quality forecasting for environmental management.

Main Methods:

• Proposed a new weighted model (PWM) combining eight neural networks and one statistical method.
• Employed Young's double-slit experimental optimizer for intelligent weighting.
• Validated the PWM using real-world dissolved oxygen data from the Tualarin River Basin, Oregon, USA.

Main Results:

• The PWM demonstrated superior performance compared to individual machine learning and statistical models.
• Achieved the lowest mean absolute percentage error among all evaluated weighted models.
• Reported average mean absolute percentage errors of 1.0216%, 1.4630%, and 1.7087% for one-, two-, and three-step predictions, respectively.

Conclusions:

• The PWM effectively integrates diverse modeling approaches, enhancing predictive accuracy.
• This approach offers robust technical support for sustainable regional water environment development.
• The proposed weighted model provides a more reliable tool for dissolved oxygen forecasting.