Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks
Wind speed forecasting is necessary to integrate wind farms into power systems. In the past ten years, the forecasting models have become increasingly complex due to the development of arti-ficial intelligence methods and computing power. Simultaneously, the robustness of models has decreased since...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Academy of Sciences of Moldova
2020-09-01
|
| Series: | Problems of the Regional Energetics |
| Subjects: | |
| Online Access: | https://journal.ie.asm.md/assets/files/07_03_47_2020.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849247635663224832 |
|---|---|
| author | Matrenin P.V. Manusov V.Z. Igumnova E.A. |
| author_facet | Matrenin P.V. Manusov V.Z. Igumnova E.A. |
| author_sort | Matrenin P.V. |
| collection | DOAJ |
| description | Wind speed forecasting is necessary to integrate wind farms into power systems. In the past ten years, the forecasting models have become increasingly complex due to the development of arti-ficial intelligence methods and computing power. Simultaneously, the robustness of models has decreased since complex models have a high risk of overfitting and decline in the accuracy if working conditions change significantly. This work aims to develop a machine learning model for short-term wind speed forecasting with acceptable accuracy but high robustness and the pos-sibility of automatic online retraining. A shallow multilayer perceptron, trained only on retro-spective data on wind speed, is proposed. The most significant results are combining simple neu-ral network architecture with ReLU activation function, Adam training method developed for deep neural networks; and the automatic hyper-parameters selection using Grid search with open upper bounds. The model was trained on the data of the autumn period and tested on the winter data. A comparison was made with the simplest and most robust adaptive forecasting methods: Brown and Holt models. The significance of the obtained results is that shallow neural networks using ReLU, Adam, and Grid search are practically not inferior to adaptive models in terms of tuning speed and the risk of subsequent differences in accuracy between training data and data supplied during operation. At the same time, shallow neural networks make it possible to obtain more accurate forecasts, and due to their small size, they are trained quickly; and retraining can be performed automatically when new data arrives. |
| format | Article |
| id | doaj-art-07159722f0dc40c9a34552b1ecadd028 |
| institution | Kabale University |
| issn | 1857-0070 |
| language | English |
| publishDate | 2020-09-01 |
| publisher | Academy of Sciences of Moldova |
| record_format | Article |
| series | Problems of the Regional Energetics |
| spelling | doaj-art-07159722f0dc40c9a34552b1ecadd0282025-08-20T03:58:10ZengAcademy of Sciences of MoldovaProblems of the Regional Energetics1857-00702020-09-01473698010.5281/zenodo.4018960Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural NetworksMatrenin P.V.0Manusov V.Z.1Igumnova E.A.2Novosibirsk State Technical University Novosibirsk, Russian FederationNovosibirsk State Technical University Novosibirsk, Russian FederationNovosibirsk State Technical University Novosibirsk, Russian FederationWind speed forecasting is necessary to integrate wind farms into power systems. In the past ten years, the forecasting models have become increasingly complex due to the development of arti-ficial intelligence methods and computing power. Simultaneously, the robustness of models has decreased since complex models have a high risk of overfitting and decline in the accuracy if working conditions change significantly. This work aims to develop a machine learning model for short-term wind speed forecasting with acceptable accuracy but high robustness and the pos-sibility of automatic online retraining. A shallow multilayer perceptron, trained only on retro-spective data on wind speed, is proposed. The most significant results are combining simple neu-ral network architecture with ReLU activation function, Adam training method developed for deep neural networks; and the automatic hyper-parameters selection using Grid search with open upper bounds. The model was trained on the data of the autumn period and tested on the winter data. A comparison was made with the simplest and most robust adaptive forecasting methods: Brown and Holt models. The significance of the obtained results is that shallow neural networks using ReLU, Adam, and Grid search are practically not inferior to adaptive models in terms of tuning speed and the risk of subsequent differences in accuracy between training data and data supplied during operation. At the same time, shallow neural networks make it possible to obtain more accurate forecasts, and due to their small size, they are trained quickly; and retraining can be performed automatically when new data arrives.https://journal.ie.asm.md/assets/files/07_03_47_2020.pdfshort-term forecastingawind energy |
| spellingShingle | Matrenin P.V. Manusov V.Z. Igumnova E.A. Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks Problems of the Regional Energetics short-term forecasting a wind energy |
| title | Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks |
| title_full | Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks |
| title_fullStr | Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks |
| title_full_unstemmed | Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks |
| title_short | Robust Short-Term Wind Speed Forecasting Using Adaptive Shallow Neural Networks |
| title_sort | robust short term wind speed forecasting using adaptive shallow neural networks |
| topic | short-term forecasting a wind energy |
| url | https://journal.ie.asm.md/assets/files/07_03_47_2020.pdf |
| work_keys_str_mv | AT matreninpv robustshorttermwindspeedforecastingusingadaptiveshallowneuralnetworks AT manusovvz robustshorttermwindspeedforecastingusingadaptiveshallowneuralnetworks AT igumnovaea robustshorttermwindspeedforecastingusingadaptiveshallowneuralnetworks |