ESTIMATION OF STATE OF CHARGE IN A SOLAR CHARGING SYSTEM USING ARTIFICIAL NEURAL NETWORKS FOR LEAD-ACID BATTERIES

Solar panel, as a renewable energy source, require batteries to store the generated energy. Continuous use of batteries can lead to capacity reduction and performance decline. To address this issue, it is important to have a system that can estimate the State of Charge (SOC) of the battery to control the charging process and maintain optimal battery performance. This research develops an SOC estimation system for lead-acid batteries by applying the Artificial Neural Network (ANN) algorithm. The ANN method has several advantages, such as more efficient iteration processes, increased speed in parameter updates, and the ability to achieve convergence more quickly. Meanwhile, the ANN algorithm also offers advantages in ease of implementation and better interpretability. The SOC estimation results for a 48V, 12Ah lead-acid battery using the ANN algorithm show that the training phase reveals a compelling regression plot, demonstrating an impressive R value of 0.99979. The minuscule error rate in the prediction system unequivocally affirms the exceptional quality of this ANN model.