Sürdürülebilir Kalkınma Amaçları
Yayınlar
- ArticleAçık Erişim2025Enhancing hydrological time series forecasting with a hybrid Bayesian-ConvLSTM model optimized by particle swarm optimization
Kilinc, Huseyin Cagan; Apak, Sina; Ergin, Mahmut Esad; Ozkan, Furkan; Katipoğlu, Okan Mert; Yurtsever, Adem
Hydrological time series forecasting often relies on addressing the inherent uncertainties and complex temporal dependencies embedded in the data. This study presents an innovative hybrid framework, the Bayesian-ConvLSTM-PSO model, specifically designed to tackle these challenges. The framework synergistically combines 1D convolutional neural networks (CNNs), a convolutional Bayesian network, multi-head attention, and long short-term memory (LSTM) networks, with parameters optimized through particle swarm optimization (PSO). The fusion of the convolutional Bayesian network and 1D convolutional neural networks enhances feature robustness by capturing both probabilistic uncertainties and spatial patterns effectively. The multi-head attention model further amplifies this by focusing on the most relevant features, improving the learning process and ensuring better representation of complex temporal dependencies. The proposed model is rigorously tested on daily streamflow data from three flow measurement stations (FMS): Ahullu (D14A014), Kızıllı (D14A080), and Erenkaya (D14A127). Experimental results reveal that the Bayesian-ConvLSTM-PSO model achieves significant performance gains across various evaluation metrics, including root mean square error (RMSE), mean absolute error (MAE), determination coefficient (R2), Kling–Gupta efficiency (KGE), and bias factor (BF). Notably, the model demonstrates exceptional accuracy with an R2 of 0.9950, a KGE of 0.9950, and a bias factor of 0.0003, surpassing the results of PSO-1D CNN-LSTM and benchmark models, such as DNN, DNN-LSTM, and 1D ConvLSTM. These compelling findings underscore the potential of the Bayesian-ConvLSTM-PSO framework as a robust and effective tool for applications in river engineering and hydrological time series forecasting.
RDL Kalıcı Kaydıİncele - ArticleYalnız künye2024A comparative study of daily streamflow forecasting using firefly, artificial bee colony, and genetic algorithm-based artificial neural network
Kilinc, Huseyin Cagan; Haznedar, Bulent; Katipoğlu, Okan Mert; Ozkan, Furkan
The management of water resources and the modeling of river flow have a prominent position within environmental research. They form a critical bridge between human societies and the delicate ecosystems they inhabit. Scholars have focused on benefiting more efficient methods based on the use of artificial intelligence for river flow forecasting, notably because modeling hydrological systems is quite challenging. This study primarily centered on exploring the predictive capacities of hybrid models in establishing a link between daily flow data and prospective data. In the study, the mentioned algorithms, firefly algorithm (FFA), artificial bee colony (ABC), genetic algorithm (GA), were hybridized with the artificial neural network (ANN) model and data analyzes were examined with the stations in the Konya Closed Basin. A comparative analysis of FFA–ANN, GA–ANN, ABC–ANN, and long short-term memory (LSTM) models was conducted for daily flow forecasting for daily flow forecasting according to a range of graphical and statistical metrics. The outcomes indicate that the FFA–ANN hybrid model generally performed better than other models and the deep learning algorithm.
RDL Kalıcı Kaydıİncele - ArticleYalnız künye2024An evolutionary hybrid method based on particle swarm optimization algorithm and extreme gradient boosting for short-term streamflow forecasting
Kilinc, Huseyin Cagan; Haznedar, Bulent; Ozkan, Furkan; Katipoğlu, Okan Mert
Streamflow estimation is necessary to develop sustainable water management strategies that balance the needs of various water users while protecting the ecological health of the watershed. This study presents a comprehensive analysis of river flow data in Turkey's Meriç basin using long-term daily data spanning the years 2001–2011. Within the scope of forecasting of daily river flows, 80% of dataset was used for training stage and the 20% of dataset was used for testing. The performance of four different models, including Extreme Gradient Boosting (XGBoost) benchmark model, Linear Regression, Long Short-Term Memory Network (LSTM), and proposed Particle Swarm Optimization (PSO)—XGBoost hybrid model, was evaluated using several statistical evaluation criteria such as RMSE, MAPE, MAE, SD, NSE, KGE, and model success was assessed based on determination coefficient (R2) values. The forecasting results of the models were evaluated with statistical metrics. The proposed PSO-XGBoost model outperformed all models with R2 values of 0.7460 at the Babaeski FMS station, 0.9582 at the Meriç FMS station, and 0.8116 at the Hayrabolu FMS station, respectively. It has been found that when the parameters of the XGBoost model are optimized with the meta-heuristic PSO algorithm, the daily flow prediction accuracy is significantly increased in the Meriç River basin, which has a pluviozonal River regime. The results demonstrate that the proposed hybrid model produces precise and reliable estimations for daily river flow analysis in the Meriç basin. The findings derived from the proposed model demonstrate its promising potential for water resource management and decision-making in the region.
RDL Kalıcı Kaydıİncele