Implementasi Battery Management System pada Monitoring Patok Perbatasan Berbasis Solar Panel Menggunakan Artificial Neural Network

Abstract

Monitoring border markers is essential for maintaining the sovereignty and security of national border areas. However, conventional monitoring systems are constrained by limited electrical infrastructure and communication networks in remote locations. This study proposes the implementation of a Battery Management System (BMS) powered by a solar panel and integrated with an Artificial Neural Network (ANN) to estimate the State of Charge (SOC) of a LiFePO₄ battery used in a border marker monitoring system. The system consists of a solar panel, Solar Charge Controller (SCC), LiFePO₄ 4S 15 Ah battery, INA219 current and voltage sensors, ESP32 microcontroller, and LoRa communication module. The ANN model was trained using battery voltage and current data to improve SOC estimation accuracy compared with the conventional voltage-based method. Experimental results showed that the INA219 sensor provided stable measurements with low error, while the power supply system achieved an efficiency of 68.4%. The battery was capable of supplying the monitoring system continuously for approximately 9.6 days without solar charging. Furthermore, the proposed ANN method achieved an average SOC estimation error of 4.86%, outperforming the conventional method with an average error of 12.16%. These results demonstrate that the proposed system is capable of operating independently, efficiently, and in real time while providing more accurate battery condition monitoring. Therefore, the proposed approach is suitable for long-term border marker monitoring applications in remote areas with limited access to electrical power and communication infrastructure.