Akademik Veri Yönetim Sistemi
7th International Congress on Human-Computer Interaction, Optimization and Robotic Applications, ICHORA 2025, Ankara, Türkiye, 23 - 24 Mayıs 2025, (Tam Metin Bildiri)
The accurate prediction of battery health is crucial for the safe, reliable, and efficient operation of applications such as electric vehicles, renewable energy storage, and portable electronics. Battery State of Health (SoH) prediction enables maintenance strategies, preventing unexpected failures and extending battery lifespan. This study has performed the SoH prediction of lithium-ion batteries using discharge voltage data through machine learning techniques. Utilizing the NASA battery dataset, a comparative analysis is conducted among Multi-Layer Perceptron (MLP), Random Forest (RF), and Support Vector Regression (SVR) algorithms. In this study, unlike other works in the literature, discharge data has been synchronized over time. Furthermore, the effect of normalized data on battery State of Health (SoH) estimation has been investigated by testing across various data formats. The results obtained reveal that the SVR method achived superior performance across all data formats, with an average MSE error rate of 0.000021. Normalized data yielded better results across all models.