Fig 1.
Implementation steps of adaptive search mechanism.
Fig 2.
Flowchart of SMA optimized SVR model.
Fig 3.
Schematic diagram of the LightGBM principle.
Fig 4.
Intelligent diagnosis process for the health of river and lake ecosystems.
Table 1.
Experimental environment and model parameter settings.
Table 2.
Comparison of error results in the test set.
Fig 5.
Water quality prediction results at different time Windows.
Table 3.
Comparison of training set error results.
Fig 6.
Comparison of error results.
Fig 7.
Comparison of the accuracy curve and the F1 score curve.
Table 4.
Comparison of training set error results.
Table 5.
Comparison of training set error results.
Table 6.
Comparison of computing time and computational complexity.
Table 7.
Analysis of cross-validation results.
Fig 8.
Comparison of MU metrics.
Fig 9.
Comparison of model robustness indicators.
Fig 10.
Comparison of error values of six models.
Table 8.
Results of the ablation experiment.
Fig 11.
Comparison of the predictive effects of pH value and DO.
Fig 12.
Comparison of the prediction effects of permanganate index and total phosphorus index.
Fig 13.
The prediction results of ammonia nitrogen index and chemical oxygen demand.
Fig 14.
Results of correlation analysis.