EXPERIMENTAL COMPARISON OF COULOMB COUNTING AND OPEN CIRCUIT VOLTAGE METHODS FOR STATE OF CHARGE ESTIMATION
Mohammad Rosyad Irfan Hasbullah, Triyogatama Wahyu Widodo, S.Kom, M.Kom
2026 | Skripsi | ELEKTRONIKA DAN INSTRUMENTASI
Accurate State of Charge (SOC) estimation is critical for the optimal management, safety, and longevity of lithium-ion batteries used in a variety of applications such as portable electronics and energy storage systems. Conventional SOC estimation methods, including Coulomb Counting and Open Circuit Voltage (OCV), face inherent challenges: Coulomb Counting suffers from cumulative error due to sensor inaccuracies and integration drift, while OCV requires prolonged resting periods to achieve voltage equilibrium, limiting real-time applicability, This research presents a comparative analysis of these two methods, each enhanced by Convolutional Neural Network (CNN) models to overcome their respective limitations.
Using the LG 18650HG2 lithium-ion battery dataset sourced from Mendeley Data, both CNN-augmented Coulomb Counting and OCV approaches were evaluated under varying load and temperature conditions. The CNN models were designed to extract complex nonlinear features from raw current and voltage data, enabling more accurate SOC predictions. Experimental results indicate that CNN integration reduces SOC estimation errors significantly compared to traditional methods alone. Specifically, the CNN-enhanced Coulomb Counting method showed superior performance in dynamic load scenarios, while the CNN-augmented OCV method provided more reliable estimates during stable conditions.
Kata Kunci : Estimasi State of Charge (SOC), Baterai Lithium-ion, Coulomb Counting, Open Circuit Voltage (OCV), Convolutional Neural Network (CNN)