Study of Short-Term Load Forecasting Techniques
Electric demand forecasting is increasingly challenging in the modern grid system, with emerging technologies like rooftop solar photovoltaics and vehicle electrification. Multiple utilities generate load forecasting independently, leading to suboptimal resource allocation and inefficiency. The chal...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Proceeding |
| Language: | en |
| Published: |
2024
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/47829/3/Study%20of%20Short-term.pdf http://ir.unimas.my/id/eprint/47829/ https://ieeexplore.ieee.org/abstract/document/10474795 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Electric demand forecasting is increasingly challenging in the modern grid system, with emerging technologies like rooftop solar photovoltaics and vehicle electrification. Multiple utilities generate load forecasting independently, leading to suboptimal resource allocation and inefficiency. The challenge lies in capturing non-linear power system characteristics associated with emerging technologies. This study has investigated several load forecasting techniques for short-term forecasting in the context of dynamic conditions and consolidates the essential components to devising an alternative solutions. This study presents a novel approach that utilizes an ensemble model as an alternative technique for short-term demand forecasting, which offers the advantage of least complicated and best-performing forecasting models. The data from the Sabah state power utility company and the Red Eléctrica de España were used as case studies to analyze the effectiveness of these techniques. The accuracy of univariate and multivariate methods is evaluated in terms of their ability to accurately forecast recent patterns of demand. The proposed alternative method using weighted ensemble model which employs Multilayer Perceptron (MLP), Decision Tree Regression and Gradient Boosting has produced an average mean absolute percentage error (MAPE) performance of 0.83% for the Sabah Grid dataset and 4.47% for the Spanish dataset |
|---|