Optimal Power Allocation of Hybrid Energy Storage Systems with Proximal Policy Optimization

Lithium-ion batteries (LIBs) are widely used in electric scooters for their high power density and other characteristics. Nonetheless, challenges arise from peak currents and battery heating, which significantly diminish the battery's cycle life. To address this issue, we present a power allocation strategy for battery-supercapacitor hybrid energy storage systems and utilize real-world data to train the proximal policy optimization (PPO) reinforcement learning method. Results demonstrate that our proposed method outperforms the low pass filter (LPF) techniques by improving efficiency by 4.3% to 16.8%.