A study on generalized optimization of energy distribution in electric vehicle hybrid energy storage system for personalized driving style scores

Abstract

Balancing battery capacity degradation and system energy loss while optimizing supercapacitor utilization remains a key challenge in hybrid energy storage system (HESS) for electric vehicle (EV). This study proposes a generalized optimization strategy based on personalized driving style scores. Using real-world EV data, Lasso regression identifies key energy consumption parameters for energy distribution control. Principal component analysis (PCA) and K-means clustering are then applied to classify the sample conditions into three driving styles: cautious, standard, and aggressive. The comprehensive scores for sample conditions are normalized to obtain individualized driving style scores, which serve as core indicators of personalized driving characteristics. A significant linear correlation is observed between the driving style scores and both battery capacity degradation and system energy loss. Based on this, a piecewise linear fitting model is constructed to calculate the contributions of battery capacity degradation and system energy loss to the total operational loss for each sample condition. These contributions are mapped to range of optimization weights, generating personalized weights for battery capacity degradation and system energy loss. A generalized weighted optimization cost function is then formulated, which is compatible with various optimization algorithms. Using GWO as an example, the results demonstrate that the proposed weighted optimization strategy reduces the average battery capacity degradation for aggressive driving styles by an additional 8.43% and decreases the average system energy loss for cautious driving styles by an additional 5.09% compared to non-weighted optimization. This enhances the flexibility and applicability of energy distribution in HESS.