Hybrid CNN–ML System for Wind Speed Level Identification in Complex Terrain: A Case Study from Maden, Turkey
Balkan Journal of Electrical and Computer Engineering, cilt.14, ss.33-40, 2026 (TRDizin)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 14
- Basım Tarihi: 2026
- Dergi Adı: Balkan Journal of Electrical and Computer Engineering
- Derginin Tarandığı İndeksler: TR DİZİN (ULAKBİM)
- Sayfa Sayıları: ss.33-40
- Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
- Bursa Uludağ Üniversitesi Adresli: Evet
Özet
This study proposes a hybrid deep and traditional learning framework for wind speed level classification in the complex terrain of the Maden region, Turkey. A one-dimensional convolutional neural network (1D-CNN) was employed for automatic feature extraction from a 30-day meteorological window, followed by classification using multiple machine learning algorithms. Among them, K-Nearest Neighbor (KNN) achieved the highest accuracy (98.75%) when applied to features extracted from the global average pooling (GAP) layer. The hybrid CNN–KNN model significantly outperformed standalone CNN and KNN baselines. The study highlights the effectiveness of combining deep feature representations with interpretable classifiers in data-scarce, topographically challenging regions, offering a transparent and high-performance alternative for wind energy assessment.