International Research Journal of Education and Technology

Mobile batteries, primarily composed of lithium-ion or lithium olymer cells, power modern smartphones' advanced functionalities. Designed to be lightweight and resilient to multiple charge cycles, these batteries are essential for portable devices. However, factors like charge cycle frequency, temperature fluctuations, user behaviour, and environmental conditions affect battery health. Over time, wear on battery components reduces charge retention and performance. Accurately predicting a mobile battery's Remaining Useful Life (RUL) is critical to meeting user expectations and enabling proactive maintenance. Traditional estimation methods, such as charge cycle counts and rule-based projections, are often static and imprecise, offering limited insights into degradation processes. In contrast, machine learning provides a more adaptive solution, integrating real-time data to account for unique usage patterns and environmental factors.This paper presents a machine-learning approach to predict mobile battery RUL using Long Short-Term Memory (LSTM) networks and Random Forest algorithms. These models analyse key metrics like charge cycles, temperature, and usage patterns to deliver personalized, accurate predictions. Comparative analysis with traditional methods shows improvements in prediction accuracy, efficiency, and responsiveness. Machine-learning-based battery life prediction enhances device reliability, enables timely maintenance interventions, and improves user satisfaction.