Intelligent diagnosis method for river and lake ecosystem health based on improved slime mold algorithm-optimized SVR

Area of Science:

• Environmental Science
• Ecology
• Data Science

Background:

• Complex nonlinear relationships and dynamic changes in water quality parameters challenge traditional statistical methods for ecosystem diagnosis.
• Limitations in accuracy, timeliness, and applicability of existing methods hinder precise assessment and early warning systems.

Purpose of the Study:

• To develop an intelligent diagnostic model for river and lake ecosystems that overcomes the limitations of traditional statistical analysis.
• To enhance the accuracy, timeliness, and applicability of water quality prediction and ecosystem health diagnosis.

Main Methods:

• An improved slime mold algorithm-optimized support vector regression model was developed, featuring dynamic weight strategy and adaptive search mechanisms.
• LightGBM feature selection was integrated to construct a combined model, addressing high-dimensional data modeling and dynamic adaptability.
• The model's performance was evaluated using metrics like Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R-squared (R2).

Main Results:

• The optimized model demonstrated significant improvements over comparative models, with RMSE of 0.031, MAE of 0.021, and R2 of 0.942.
• The final combined model achieved superior prediction accuracy with RMSE of 0.022, MAE of 0.016, and R2 of 0.976.
• The model showed good prediction effects for key water quality parameters including pH, dissolved oxygen, and chemical oxygen demand, with low sensitivity to outliers.

Conclusions:

• The proposed intelligent diagnostic model offers efficient and accurate methods for water quality prediction and river and lake ecosystem health diagnosis.
• The study's findings support the digital transformation of water environment management by enhancing ecological risk prevention and control capabilities.
• The developed model exhibits superior performance and practical prediction effectiveness for diagnosing river and lake ecosystems.