Machine Learning Methods for the Prediction of Wastewater Treatment Efficiency and Anomaly Classification with Lack of Historical Data

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(22), P. 10689 - 10689

Published: Nov. 19, 2024

This study examines an algorithm for collecting and analyzing data from wastewater treatment facilities, aimed at addressing regression tasks predicting the quality of treated classification preventing emergency situations, specifically filamentous bulking activated sludge. The feasibility using obtained under laboratory conditions simulating technological process as a training dataset is explored. A small collected actual plants considered test dataset. For both tasks, best results were achieved gradient-boosting models CatBoost family, yielding metrics SMAPE = 9.1 ROC-AUC 1.0. set most important predictors modeling was selected each target features.

Language: Английский

---

Fusion of In-Situ and Modelled Marine Data for Enhanced Coastal Dynamics Prediction Along the Western Black Sea Coast

Journal of Marine Science and Engineering, Journal Year: 2025, Volume and Issue: 13(2), P. 199 - 199

Published: Jan. 22, 2025

This study explores the use of Temporal Fusion Transformers (TFTs), an AI/ML technique, to enhance prediction coastal dynamics along Western Black Sea coast. We integrate in-situ observations from five meteo-oceanographic stations with modelled geospatial marine data Copernicus Marine Service. TFTs are employed refine predictions shallow water by considering atmospheric influences, a particular focus on wave-wind correlations in regions. Atmospheric pressure and temperature treated as latitude-dependent constants, specific investigations into extreme events like freezing solar radiation-induced turbulence. Explainable AI (XAI) is exploited ensure transparent model interpretations identify key influential input variables. Data attribution strategies address missing concerns, while ensemble modelling enhances overall robustness. The models demonstrate significant improvement accuracy compared traditional methods. research provides deeper understanding atmosphere-marine interactions demonstrates efficacy Artificial intelligence (AI)/Machine Learning (ML) bridging observational gaps for informed zone management decisions, essential maritime safety

Language: Английский

---

A deep learning model for real-time forecasting of 2-D river flood inundation maps

Published: Aug. 5, 2024

Abstract. Floods are among the most hazardous natural disasters worldwide. Accurate and rapid flood predictions critical for effective early warning systems management strategies. The high computational cost of hydrodynamic models often limits their application in real-time simulations. Conversely, data-driven gaining attention due to efficiency. In this study, we aim at assessing effectiveness transformer-based forecasting spatiotemporal evolution fluvial floods real-time. To end, model FloodSformer (FS) has been adapted predict river inundations with negligible time. FS leverages an autoencoder framework analyze reduce dimensionality spatial information input water depth maps, while a transformer architecture captures correlations between inundation maps inflow discharges using cross-attention mechanism. trained can long-lasting events autoregressive procedure. model's performance was evaluated two case studies: urban flash scenario laboratory scale along segment Po River (Italy). Datasets were numerically generated two-dimensional model. Special given analyzing how accuracy is influenced by type severity used create training dataset. results show that prediction errors generally align uncertainty observed physically based models, larger more diverse datasets help improving accuracy. Additionally, time procedure compared physical simulated event. also benchmarked against state-of-the-art convolutional neural network showed better These findings highlight potential enhancing responsiveness, contributing improve resilience.

Language: Английский

---

A Systematic Modular Approach for the Coupling of Deep-Learning-Based Models to Forecast Urban Flooding Maps in Early Warning Systems

Hydrology, Journal Year: 2024, Volume and Issue: 11(12), P. 215 - 215

Published: Dec. 12, 2024

Deep learning (DL) approaches to forecast precipitation and inundation areas in the short-term horizon have up until now been treated as independent research problems from model development perspective. However, for urban hydrology area, coupling of these models is necessary order upcoming area maps is, therefore, utmost importance successful flood risk management. In this paper, three deep-learning-based are coupled a systematic modular approach with aim analyze performance chain an operative setup pluvial flooding nowcast: nowcasting adapted version NowcastNet model, manhole overflow hydrographs Seq2Seq generation spatiotemporal sequence catchment hour encoder–decoder model. It can be concluded that quality still largely depends on accuracy With increasing DL both nowcasting, presented enables substitution individual blocks better newer without jeopardizing operation system.

Language: Английский

---

Analysis and Forecasting of Temperature Based on Temporal Fusion Transformer Model: A Case Study of Urumqi

Published: Nov. 8, 2024

Language: Английский

---