PITPS: Balancing Local and Global Profits for Multiple Charging Stations Management

The rapid growth of electric vehicles (EVs) has created a challenge for the EV charging problem due to unpredictable user patterns and fluctuations in demand. To maximize profits for charging stations (CSs), previous research has focused on making management strategies for individual CSs. However, it remains a challenge to maximize the profit for diverse local environments while maintaining high overall profit for the entire system. To address this challenge, we present PITPS, a Personalized Federated Reinforcement Learning method, which consists of two key components: personalized reinforcement learning models that maximize local profits and a global aggregator model that balances loads and optimizes overall profit by pricing electricity. Our experiments validate the effectiveness of PITPS in achieving high local and global profit while filling the valley and reducing the peak loads.