A comparative study of ARIMA-GARCH model and artificial neural network model for wind speed forecasting
Wind energy is a noteworthy alternate energy in times of energy crisis. An accurate wind speed forecasting model are essential in determining the suitable location for wind energy harvesting. However, the intermittency and nonlinearity of a wind speed make it difficult to obtain an accurate predic...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English |
| Published: |
2023
|
| Online Access: | http://eprints.utem.edu.my/id/eprint/28028/1/A%20comparative%20study%20of%20ARIMA-GARCH%20model%20and%20artificial%20neural%20network%20model%20for%20wind%20speed%20forecasting.pdf http://eprints.utem.edu.my/id/eprint/28028/ https://pubs.aip.org/aip/acp/article-abstract/2500/1/020009/2875195/A-comparative-study-of-ARIMA-GARCH-model-and?redirectedFrom=fulltext |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Teknikal Malaysia Melaka |
| Language: | English |
| Summary: | Wind energy is a noteworthy alternate energy in times of energy crisis. An accurate wind speed forecasting
model are essential in determining the suitable location for wind energy harvesting. However, the intermittency and
nonlinearity of a wind speed make it difficult to obtain an accurate prediction and may cause several operational
challenges to grid interfaced the wind energy system. In this study, the time series model and artificial neural network
(ANN) model was applied on the daily wind speed data in Senai and Mersing, Johor to forecast future wind speed series.
For the time series model, the daily wind speed data was initially been modelled using theARIMA model. However, due
to the presence of heteroscedastic effect in the residuals of ARIMA model, GARCH model was introduced to handle the
nonlinearity criteria. On the other hand,the Multilayer Perceptron (MLP)model which is in the class of feed-forward
ANN was developed with different configurations based on selected hyperparameters. The MLP model configuration
with the lowest RMSE value was selected as the best MLP model. A comparison has been made between the ARIMA-
GARCH model and theMLPmodel. Results indicate that the MLP model was found to outperform the ARIMA-GARCH
model by providing the lowest value of root mean square error (RMSE), mean absolute error (MAE) and mean absolute
percentage error (MAPE) in both training and testing data sets. Thus, the artificial neural network can be concluded as the
best method to provide a good forecasting model in predicting the daily wind speed data. |
|---|