Water, Год журнала: 2024, Номер 16(21), С. 3099 - 3099

Опубликована: Окт. 29, 2024

Traditional single prediction models struggle to address the complexity and nonlinear changes in water quality forecasting. To this challenge, study proposed a coupled model (RF-TVSV-SCL). The includes Random Forest (RF) feature selection, dual signal decomposition (Time-Varying Filtered Empirical Mode Decomposition, TVF-EMD, Sparrow Search Algorithm-Optimized Variational SSA-VMD), deep learning predictive (Sparrow Algorithm-Convolutional Neural Network-Long Short-Term Memory, SSA-CNN-LSTM). Firstly, RF method was used for selection extract important features relevant prediction. Then, TVF-EMD employed preliminary of data, followed by secondary complex Intrinsic Function (IMF) components using SSA-VMD. Finally, SSA-CNN-LSTM utilized predict processed data. This evaluated predicting total phosphorus (TP), nitrogen (TN), ammonia (NH3-N), dissolved oxygen (DO), permanganate index (CODMn), conductivity (EC), turbidity (TB), across 1, 3, 5, 7-d forecast periods. performed exceptionally well short-term predictions, particularly within 1–3 d range. For 1-, 3-, 5-, forecasts, R2 ranged from 0.93–0.96, 0.79–0.87, 0.63–0.72, 0.56–0.64, respectively, significantly outperforming other comparison models. RF-TVSV-SCL demonstrates excellent capability generalization ability, providing robust technical support forecasting pollution prevention.

Язык: Английский