Wardani, Firly Setya (2025) WIND SPEED PREDICTION USING TEMPORAL CONVOLUTIONAL NETWORK (TCN): A CASE STUDY AT BMKG JUANDA. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 1)
21083010093_BAB 1.pdf Download (226kB) |
|
Text (Bab 2)
21083010093_BAB 2.pdf Restricted to Repository staff only until 5 December 2027. Download (531kB) |
|
Text (Bab 3)
21083010093_BAB 3.pdf Restricted to Repository staff only until 5 December 2027. Download (497kB) |
|
Text (Bab 4)
21083010093_BAB 4.pdf Restricted to Repository staff only until 5 December 2027. Download (4MB) |
|
Text (Bab 5)
21083010093_BAB 5.pdf Download (152kB) |
|
Text (Daftar Pustaka)
21083010093_DAFTAR PUSTAKA.pdf Download (164kB) |
|
Text (Lampiran)
21083010093_LAMPIRAN.pdf Download (395kB) |
|
Text (Cover)
21083010093_Cover.pdf Download (1MB) |
Abstract
Wind speed is a meteorological phenomenon that plays a critical role in transportation safety, energy optimization, and disaster mitigation, yet it is highly fluctuating and difficult to predict accurately. This study aims to forecast wind speed around BMKG Juanda using two deep learning models, Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM), and compare their performance with the conventional ARIMA method. Daily wind speed data were divided by season to handle missing values more representatively. Hyperparameter optimization was conducted using Optuna to obtain the optimal model configuration. Prediction results show that TCN achieved the best performance with lower and more stable errors compared to LSTM and ARIMA, particularly in capturing extreme wind speed fluctuations. TCN’s 30-day forecast ranged from 5.079 to 5.556 m/s, while LSTM ranged from 4.926 to 6.108 m/s, and ARIMA from 8.188 to 18.265 m/s. These findings indicate that TCN is more effective in learning complex temporal dynamics in wind speed data and provides more accurate and realistic predictions. This study is expected to support the development of a reliable weather forecasting system, enhance extreme weather risk mitigation, and optimize wind energy utilization in densely populated urban areas.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA76.6 Computer Programming Q Science > QA Mathematics > QA76.87 Neural computers |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Firly Wardani | ||||||||||||
| Date Deposited: | 05 Dec 2025 08:37 | ||||||||||||
| Last Modified: | 05 Dec 2025 08:37 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/47811 |
Actions (login required)
 |
View Item |