Smartphone Battery Remaining Runtime Prediction Model Based on Particle Filtering and Monte Carlo Simulation

Abstract

This paper proposes a closed-loop electro-thermal coupling modeling method for predicting the remaining battery life of smartphones. The method maps the screen, processor, network, GPS, and background tasks to system power consumption, and combines a second-order Thevenin equivalent circuit, dynamic SOC updates, effective capacity correction, and a first-order thermal model to characterize the feedback relationships among power consumption, current, voltage, temperature, and capacity degradation. Building on this foundation, a particle filter is introduced to perform individualized calibration of key parameters, a semi-Markov process is used to generate typical load sequences, and Monte Carlo simulations are employed to obtain the TTE distribution, confidence intervals, and the risk of premature shutdown. Validation results indicate that the model can adapt to different loads, temperatures, and battery aging states, providing a universally applicable modeling framework for battery state assessment and runtime prediction in mobile devices.