طالبات ماجستير الذكاء الاصطناعي وتحليل البيانات أسل موسى، وهلا شاهين، وإسراء زواهرة ينشرن دراسة بعنوان “تصنيف أنواع الأعطال في خطوط نقل الطاقة الكهربائية”
ملخص البحث
This work proposes machine learning (ML) and deep learning (DL) algorithms. Applying SMOTE and class weighting techniques optimizes the process through grid search in combination with Bayesian optimization (Optuna). Accuracy levels improved significantly with the addition of more features and data-balancing techniques. After applying SMOTE to the CNN model, its accuracy increased from 86.56% to 87.58%, while the decision tree and random forest maintained their accuracy levels at 97%. According to the research, model performance significantly improved when our team integrated custom neural network protocols with batch normalization, dropout regularization, and cosine annealing learning rate scheduling techniques. These methods demonstrated their ability to enhance network reliability through efficient fault detection systems. Future research will expand existing datasets and implement improvements to adapt the model for integrating renewable energy systems and smart grids.
كيفية الاستشهاد للبحث:
Mousa, A., Zraqou, J., Shaheen, H., & Zawahra, I. (2025, April). Fault Type Classification in Electrical Transmission Lines. In 2025 1st International Conference on Computational Intelligence Approaches and Applications (ICCIAA) (pp. 1-8). IEEE.
رابط البحث