An Innovative Hyperparametric Interface for Accurate and User-Friendly System Marginal Price Forecasting in Energy Markets
AKDENIZ 13th International Conference on Applied Sciences, Girne, Kıbrıs (Kktc), 3 - 05 Ocak 2025, ss.55, (Özet Bildiri)
- Yayın Türü: Bildiri / Özet Bildiri
- Basıldığı Şehir: Girne
- Basıldığı Ülke: Kıbrıs (Kktc)
- Sayfa Sayıları: ss.55
- Çukurova Üniversitesi Adresli: Evet
Özet
Accurate forecasting of the System Marginal Price (SMP) is crucial for efficient electricity market operations, particularly in dynamic environments like the Turkish energy market. This study introduces an innovative interface based on hyperparametric elasticity to enhance forecasting models' accuracy and reliability. It integrates advanced machine learning methods, including Multi-Layer Perceptrons (MLP), Long Short-Term Memory (LSTM), Bi-Directional Long Short-Term Memory (Bi-LSTM), Convolutional Long Short-Term Memory (ConvLSTM), Extreme Learning Machine (ELM), and eXtreme Gradient Boosting (XGBoost), as well as statistical approaches such as AutoRegressive Integrated Moving Average (ARIMA) and Seasonal AutoRegressive Integrated Moving Average (SARIMA). The interface offers flexibility in feature selection, parameter optimization, and data analysis, employing algorithms like minimum Redundancy Maximum Relevance (mRMR) and Maximum Likelihood Feature Selector (MLFS) to refine predictive models. By leveraging hyperparametric elasticity, users can systematically optimize configurations, evaluate performance using Mean Absolute Percentage Error (MAPE), and visualize predictions through interactive graphics. Empirical analysis on a dataset (January 2021–September 2023) from Energy Exchange Istanbul (EXIST) shows that models using this interface outperform traditional approaches, with ELM and XGBoost achieving the lowest MAPE values. The findings highlight the importance of tailored feature selection and parameter tuning in reducing errors and improving decision-making.