Classification of Power Quality Disturbances Using Convolutional Neural Networks with Wavelet Scalogram Images

Abstract

Accurate and fast detection of power quality disturbances (PQDs) is of critical importance for the reliability and efficiency of electric power systems. In this study, a time-frequency based approach is proposed for the classification of power quality disturbances. In this context, signals belonging to 15 different PQD classes were generated through mathematical models, and a dataset was created using a sampling frequency of 12.8 kHz. The generated signals were converted into scalogram images using Continuous Wavelet Transform (CWT) to obtain a suitable input format for deep learning models. In the classification phase, the AlexNet, SqueezeNet, and ResNet-18 architectures were trained using a transfer learning approach. The performance of the models was evaluated under noise-free, 40 dB, and 30 dB SNR (signal to noise ratio) levels. The results showed that all models achieved high accuracy rates in the noise-free condition, but performance differences became significant with increasing noise levels. In particular, the ResNet-18 architecture was the most successful model, maintaining high classification accuracy even under low SNR conditions. The study results show that scalogram-based data generation and deep learning methods offer an effective and reliable approach for classifying power quality disturbances.

Keywords

• Power quality disturbances
• Classification
• Scalogram
• Deep learning

Dalgacık Skalogram Görüntüleri Kullanılarak Evrişimli Sinir Ağları İle Güç Kalitesi Bozulmalarının Sınıflandırılması

Abstract

Güç kalitesi bozulmalarının (PQD) doğru ve hızlı bir şekilde tespit edilmesi, elektrik güç sistemlerinin güvenilirliği ve verimliliği açısından ciddi bir öneme sahiptir. Bu çalışmada, güç kalitesi bozulmalarının sınıflandırılması amacıyla zaman–frekans temelli bir yaklaşım önerilmiştir. Bu kapsamda, 15 farklı PQD sınıfına ait sinyaller matematiksel modelleri aracılığıyla üretilmiş ve 12,8 kHz örnekleme frekansı kullanılarak veri kümesi oluşturulmuştur. Üretilen sinyaller, Sürekli Dalgacık Dönüşümü (CWT) yardımıyla skalogram görüntülerine dönüştürülerek derin öğrenme modelleri için uygun bir giriş formatı elde edilmiştir. Sınıflandırma aşamasında transfer öğrenme yaklaşımı kullanılarak AlexNet, SqueezeNet ve ResNet-18 mimarileri eğitilmiştir. Modellerin performansı gürültüsüz, 40 dB ve 30 dB SNR (sinyal gürültü oranı) seviyeleri altında değerlendirilmiştir. Elde edilen sonuçlar, tüm modellerin gürültüsüz durumda yüksek doğruluk oranlarına ulaştığını, ancak gürültü oranının artmasıyla performans farklılıklarının belirgin hale geldiğini göstermiştir. Özellikle ResNet-18 mimarisi, düşük SNR koşullarında dahi yüksek sınıflandırma doğruluğunu koruyarak en başarılı model olmuştur. Çalışma sonuçları, skalogram tabanlı veri üretiminin ve derin öğrenme yöntemlerinin güç kalitesi bozulmalarının sınıflandırılmasında etkili ve güvenilir bir yaklaşım sunduğunu ortaya koymaktadır.

Keywords

• Güç kalitesi bozuklukları
• Sınıflandırma
• Skalogram
• Derin öğrenme

References

1. Akkaya, S., & Dümen, S. (2025). A novel vision transformer-based power quality disturbance classification method. Ain Shams Engineering Journal, 16(11), 103718. https://doi.org/10.1016/j.asej.2025.103718
2. Coban, M., & Tezcan, S. S. (2021). Detection and classification of short-circuit faults on a transmission line using current signal. Bulletin of the Polish Academy of Sciences: Technical Sciences, 69(4). https://doi.org/10.24425/bpasts.2021.137630
3. Gümüş, B., Coban, M., & Tezcan. S Suleyman. (2024). Metasezgisel Tabanlı Özellik Seçim Algoritmalarının Güç Kalitesi Olaylarının Sınıflandırılmasına Etkisinin İncelenmesi. Uluslararası Mühendislik Araştırma ve Geliştirme Dergisi, 16(02), 646–658.
4. He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2016-Decem, 770–778. https://doi.org/10.1109/CVPR.2016.90
5. Iandola, F. N., Han, S., Moskewicz, M. W., Ashraf, K., Dally, W. J., & Keutzer, K. (2017). SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0.5 MB model size. arXiv preprint arXiv:1602.07360. 1–13.
6. Karasu, S., & Saraç, Z. (2022). The effects on classifier performance of 2D discrete wavelet transform analysis and whale optimization algorithm for recognition of power quality disturbances. Cognitive Systems Research, 75(February 2019), 1–15. https://doi.org/10.1016/j.cogsys.2022.05.001
7. Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). ImageNet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84–90. https://doi.org/10.1145/3065386
8. Mallat, S. G. (1989). A Theory for Multiresolution Signal Decomposition: The Wavelet Representation. In IEEE Transactions On Pattern Analysis And Machine Intelligence: Vol. I (Issue 7).

9. Mishra, M. (2019). Power quality disturbance detection and classification using signal processing and soft computing techniques : A comprehensive review. International Transactions on Electrical Energy Systems, 29(8), 1–42. https://doi.org/10.1002/2050-7038.12008
10. Radhakrishnan, P., Ramaiyan, K., & Vinayagam, A. (2021). A stacking ensemble classification model for detection and classification of power quality disturbances in PV integrated power network. Measurement, 175(October 2020), 109025. https://doi.org/10.1016/j.measurement.2021.109025
11. Ranjan, R., Mishra, M., Nayak, J., Shanmuganathan, V., Naik, B., & Jung, Y. (2022). A power quality detection and classification algorithm based on FDST and hyper-parameter tuned light-GBM using memetic firefly algorithm. Measurement, 187(October 2021), 110260. https://doi.org/10.1016/j.measurement.2021.110260
12. Salles, R. S., & Ribeiro, P. F. (2023). The use of deep learning and 2-D wavelet scalograms for power quality disturbances classification. Electric Power Systems Research, 214(PA), 108834. https://doi.org/10.1016/j.epsr.2022.108834
13. Sindi, H., Nour, M., Rawa, M., & Polat, K. (2021). A novel hybrid deep learning approach including combination of 1D power signals and 2D signal images for power quality disturbance classification. Expert Systems with Applications, 174(November 2020), 1–13. https://doi.org/10.1016/j.eswa.2021.114785
14. Sipai, U., Jadeja, R., & Member, S. (2024). Performance Evaluation of Discrete Wavelet Transform and Machine Learning Based Techniques for Classifying Power Quality Disturbances. IEEE Access, 12(June), 95472–95486.
15. Subudhi, U., & Dash, S. (2021). Detection and classification of power quality disturbances using GWO ELM. Journal of Industrial Information Integration, 22, 1–11. https://doi.org/10.1016/j.jii.2021.100204
16. Vinayagam, A., Veerasamy, V., Tariq, M., & Aziz, A. (2022). Heterogeneous learning method of ensemble classifiers for identification and classification of power quality events and fault transients in wind power integrated microgrid. Sustainable Energy, Grids and Networks, 31, 100752. https://doi.org/10.1016/j.segan.2022.100752
17. Wang, J., Zhang, D., & Zhou, Y. (2022). Ensemble deep learning for automated classification of power quality disturbances signals. Electric Power Systems Research, 213(August), 108695. https://doi.org/10.1016/j.epsr.2022.108695
18. Wei, K. (2021). A simple gated recurrent network for detection of power quality. IET Generation, Transmission &Distribution, 15(4), 751–761. https://doi.org/10.1049/gtd2.12056
19. Xu, W., Wang, R., Zhang, Y., Wang, J., Wang, Z., Wu, X., Li, W., Li, X., Zhang, M., & Sun, L. (2026). A power quality disturbance classification method using a hybrid transformer and discrete wavelet transform model. Electric Power Systems Research, 253(November 2025), 112547. https://doi.org/10.1016/j.epsr.2025.112547
20. Zhang, Y., Zhang, Y., & Zhou, X. (2022). Classification of power quality disturbances using visual attention mechanism and feed-forward neural network. Measurement, 188, 110390.