A Data-Driven Approach for Electric Vehicle Powertrain Modeling

TL;DR

This work addresses the lack of system-level EV powertrain simulations by introducing a modular, data-driven framework that unifies battery, inverter, and motor models via standardized interfaces. It blends data-driven surrogates with physics-based components within a V-model development process, enabling independent development in Python and integration in Simulink, with FMI compatibility. Key contributions include the hybrid architecture, bottom-up integration with rule-based control, and a discussion of fidelity-versus-cost trade-offs, plus plans for validation and real-time deployment. The approach enables rapid virtual validation, scalable design space exploration, and preparation for hardware-in-the-loop testing in next-generation electric powertrains.

Abstract

Electrification in the automotive industry and increasing powertrain complexity demand accelerated, cost-effective development cycles. While data-driven models are recently investigated at component level, a gap exists in systematically integrating them into cohesive, system-level simulations for virtual validation. This paper addresses this gap by presenting a modular framework for developing powertrain simulations. By defining standardized interfaces for key components-the battery, inverter, and electric motor-our methodology enables independently developed models, whether data-driven, physics-based, or empirical, to be easily integrated. This approach facilitates scalable system-level modeling, aims to shorten development timelines and to meet the agile demands of the modern automotive industry.

Figures (4)

• Figure 1: Literature review and research gap. Although there are data-driven studies focusing on data-driven modeling and simulation for individual components, this research targets an integrated powertrain simulation composed by these individual components.
• Figure 2: Structure of model development. The proposed methodology suggests starting with higher-level systems by defining their required inputs and outputs. Then, a top-down approach is used to decompose the system to the smallest modelable element. Finally, a bottom-up approach is followed to integrate them.
• Figure 3: Architecture of model development. The proposed methodology can independently integrate conventional physics-based and data-driven component models into a cohesive powertrain simulation.
• Figure 4: Overview of the proposed framework for an EV drive system. This model integrates traditional rule-based logic for control components with data-driven plant models for the core powertrain. These plant models use artificial intelligence to accurately capture complex nonlinear behaviors, estimating the desired outputs.