ENHANCING BATTERY PERFORMANCE PREDICTION FOR SPORTS-RELATED MOBILE ENERGY DEVICES USING ADVANCED DRAGONFLY-OPTIMIZED SPATIAL BAYESIAN MODELS

Abstract

Lithium batteries power a wide range of sports and fitness devices, from wearable technology to equipment used in athletic training and competitions. Accurate prediction of state-of-charge (SOC) and battery life is essential for the optimal performance and reliability of these devices. Traditional SOC prediction methods often fall short in accuracy, particularly under the variable conditions encountered in sports environments. This study introduces an advanced predictive model, the Improved Dragonfly Optimized Spatial Bayesian Network (IDO-SBN), which enhances prediction accuracy by integrating Improved Dragonfly Optimization (IDO) with Spatial Bayesian Networks (SBN). The methodology involves extensive data collection from experimental battery tests, including hybrid pulse power characterization (HPPC) and static discharge tests (SDT), tailored to the specific demands of sports applications. The IDO algorithm optimizes hyper-parameters to tailor the SOC predictions to the unique usage patterns seen in various sports scenarios, such as varying intensities and durations of athletic activities. The performance of the IDO-SBN model is rigorously validated across different battery chemistries and under diverse operating conditions relevant to sports, such as temperature fluctuations and mechanical vibrations. The results demonstrate low Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) in SOC estimations, maintaining high accuracy even amidst noise and battery aging—factors commonly encountered in sports settings. This improved prediction model offers significant advancements in battery management for sports technology, ensuring athletes and coaches can rely on their electronic devices for training and performance monitoring without unexpected disruptions.