This study delves into a comparative analysis of knee point prediction algorithms for electric vehicle (EV) battery health monitoring. Knee points are pivotal markers of battery degradation, crucial for precise state-of-health (SOH) prediction. We evaluate the performance of four machine learning (ML) models of support vector machine (SVM), AdaBoost, neural network (NN), and a novel hybrid XGBoost-random forest (XG-RF) using a real-world battery data. While all models consistently identify knee points, disparate mean absolute error (MAE) values suggest potential algorithmic variations. We emphasize the significance of comprehending these disparities, particularly through residual distribution and quantile plots, to enhance the reliability and robustness of battery health monitoring systems. This analysis offers valuable insights for optimizing EV battery health monitoring strategies and advancing the effectiveness of predictive maintenance techniques.
Effect of Algorithms in Accurate Knee Point Prediction of Lithium-ion Batteries in Electric Vehicles
2024-07-31
599491 byte
Conference paper
Electronic Resource
English
Adaptive algorithms for monitoring of lithium-ion batteries in electric vehicles
Tema Archive | 2014
|Lithium Ion batteries in electric drive vehicles
SLUB | 2016
|Lithium ion batteries in electric drive vehicles
TIBKAT | 2016
|