The fast-paced advancement of electric vehicle (EV) technology highlights a need for new methods to calibrate control systems that allow for flexible responses that improve performance, safety, and energy efficiency. A new digital twin (DT) architecture is proposed to enable continuous calibration of on-road EV control systems in development and production. Existing calibration methods are historically often compartmentalized and broken apart by relying on offline unsuspected simulations and engineering reality prototypes. Both of which inevitably limit iteration and flexibility once in use. In the case of EVs, the proposed architecture assimilates near real-time sensors with predictive analytics and provides a closed-loop feedback of control parameters in the field that gap integrated between the virtual vehicle and the physical vehicle. A modular architecture including a cloud-based DT platform, an edge for low-latency, and embedded vehicle controllers was developed to validate using a case study of torque distribution and battery management systems and showed to lessen the time to calibrate by 22% in development, and improved energy efficiency by 15% when in real-world operation. The result shows promise for the potential of continuous calibration to substantially improve the vehicle development cycle and ensure a better operational performance once on the road. Implications for the study include automated and scalable manufacturing processes, and automated control opportunities for heterogeneous EV fleets. Next work will look to add an AI anomaly detection system, with multi-vehicle DT synchronization software interactions.
Vijayachandar Sanikal (Mon,) studied this question.