Improved Quantum Artificial Bee Colony Algorithm-Optimized Artificial Intelligence Models for Suspended Sediment Load Predicting

Improved Quantum Artificial Bee Colony Algorithm-Optimized Artificial Intelligence Models for Suspended Sediment Load Predicting

River sediment plays a critical role in maintaining the health of river ecosystems, and accurately predicting suspended sediment load (SSL) is vital for effective river management. This study proposes a quantum bee colony algorithm based on Boltzmann’s selection and dimensional dynamic up...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Peng Wei, Wang Yu
Format: Artikel
Sprache:Englisch
Veröffentlicht: IEEE 2025-01-01
Schriftenreihe:IEEE Access
Schlagworte:
Suspended sediment load
artificial intelligence
quantum computing
quantum bee colony algorithm
hybrid algorithm
Online-Zugang:https://ieeexplore.ieee.org/document/10318821/
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:River sediment plays a critical role in maintaining the health of river ecosystems, and accurately predicting suspended sediment load (SSL) is vital for effective river management. This study proposes a quantum bee colony algorithm based on Boltzmann’s selection and dimensional dynamic update strategy (BDQABC). The BDQABC is utilized to optimize artificial intelligence (AI) models, specifically BDQABC-SVR and BDQABC-ANN. Both models are applied to predict SSL at the Caguas Municipio station in Loíza River, Puerto Rico. To evaluate the predictive capability, the models are compared with quantum bee colony algorithm-optimized AI models (QABC-SVR and QABC-ANN), genetic algorithm-optimized AI models (GA-SVR and GA-ANN) and traditional AI models (SVR and ANN). The models using various combinations of lagged river flow and SSL values as inputs. The performance of the models is assessed using metrics such as root mean square error (RMSE), mean absolute error, nash-Sutcliffe efficiency, coefficient of determination, and graphical comparison methods. The results indicate that the BDQABC-SVR and BDQABC-ANN models exhibit stronger predictive capabilities for SSL than other models. Specifically, BDQABC-SVR improves the RMSE accuracy of the QABC-SVR model by 5.6% and 22.0% respectively, while BDQABC-ANN improves the RMSE accuracy of the QABC-ANN model by 13.3% and 9.7%, respectively for the two input scenarios.
ISSN:2169-3536

Ähnliche Einträge