물리정보신경망을 활용한 다이나믹 전기차 주행 충전 상태 추정 프레임워크

초록

As electric vehicles (EVs) continue to gain widespread adoption, the need to accurately monitor and manage their internal systems has become increasingly critical. Among the various factors influencing battery performance, the State of Charge (SoC) stands out as a key variable requiring precise estimation and control. This study proposes a physics-based approach for estimating the dynamic behavior of the SoC by considering real-life operating scenarios and phases. The estimation is performed using a Physics-Informed Neural Network (PINN), a recent advancement in machine learning that integrates physical laws with data-driven modeling, offering superior capability in capturing complex system dynamics compared to traditional deep learning methods. The proposed PINN model’s performance is rigorously evaluated against several conventional machine learning algorithms, demonstrating the effectiveness of the proposed framework under the operational scenarios.

키워드