Firefly algorithm-based estimation of the dynamics of lithium-ion battery used in wireless sensors

Wireless sensors are powered by small-sized batteries with limited energy and lifespan; hence, proper battery management is essential for communication reliability and network performance. Thus, this paper presents a firefly algorithm (FA) – based model for estimating the battery's behaviour. A single resistor-capacitor (RC) circuit representing the battery was simulated to obtain the voltage response of the battery to charge and discharge current pulses. The battery was first fully charged, and discharge current pulse of 30 mA was applied for 5 seconds within a duration on one hour. The measured voltage responses were stored, and the procedure was repeated for each 10% State-of-Charge (SoC) interval until the battery was discharged. The transfer function of the circuit was transformed into coefficients that were optimized by the FA for estimating the terminal voltage, Open-Circuit Voltage (OCV) and SoC of the battery. The proposed FA model estimated the terminal voltage with an average estimation error of 8.2 mV. The FA model gave Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) of 0.0479 and 0.0864, respectively, which outperforms the conventional Ampere-hour (Ah) Counting method that incurred MAE and RMSE of 0.1115 and 0.1361, respectively. The results reveal that the FA model outperforms the Ah Counting method in OCV-SoC estimation by 76.72 %. This study has shown that the proposed FA model is suitable for providing improved battery level estimation for wireless sensors; thus, Wireless Sensor Networks (WSN) can be better monitored and operated.