Examining different approaches for short-term load demand forecasting in microgrid management: a case study of a university in Nigeria
In remote regions, microgrids are increasingly recognized as dependable electricity sources, underscoring the necessity for precise short-term load demand forecasts to ensure efficient microgrid management. This study assessed three forecasting methodologies—Ridge Regression (RG), Autoreg...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Academia.edu Journals
2024-05-01
|
| Series: | Academia Green Energy |
| Online Access: | https://www.academia.edu/118710636/Examining_different_approaches_for_short_load_demand_forecasting_in_microgrid_management_a_case_study_of_a_university_in_Nigeria |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In remote regions, microgrids are increasingly recognized as dependable electricity sources, underscoring the necessity for precise short-term load demand forecasts to ensure efficient microgrid management. This study assessed three forecasting methodologies—Ridge Regression (RG), Autoregressive Integrated Moving Average (ARIMA), and Support Vector Machine (SVM) regression—for predicting electricity load demand in a microgrid at the University of Jos in Nigeria. The objective was to achieve accurate load demand forecasts within a specified timeframe. Leveraging 6 years of historical load data from the Jos Electricity Distribution Company (JEDC) and weather data from the Nigerian Meteorological Agency (NIMET), the research demonstrated that both RG and ARIMA outperformed SVM, exhibiting lower Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE) values. Specifically, RG recorded an RMSE of 256.414 (representing a reduction of 11.1% compared to SVM), MSE of 65,748.170 (a decrease of 8.9% compared to SVM), and MAPE of 37.730 (a decrease of 12.0% compared to SVM), while ARIMA displayed an RMSE of 215.820 (a decrease of 20.5% compared to SVM), MSE of 46,580.220 (a decrease of 21.4% compared to SVM), and MAPE of 30.210 (a decrease of 32.9% compared to SVM). In contrast, SVM yielded an RMSE of 289.419, MSE of 83,763.112, and MAPE of 45.482. The study concluded that ARIMA outperforms both RG and SVM in accurate microgrid load forecasting, emphasizing the importance of selecting suitable forecasting techniques in energy resource allocation decisions. |
|---|---|
| ISSN: | 2998-3665 |