Reverse Engineering-Based Modeling of an EV Motor Drive for Digital Twin Development

Digital twins (DTs) play crucial roles in various applications, with their significance extending to critical domains like electric vehicles (EVs). Essentially, a DT is a virtual representation or digital counterpart of a physical object, system, or process, crafted through sensor technology, data collection, modeling, and simulation. Modeling a real system with uncertain parameters is a challenging task. This paper presents a comprehensive methodology for reverse engineering the drive system of an EV Permanent Magnet Synchronous Motor (PMSM) with an uncertain control strategy parameter. The motor was modeled based on d-q mathematical representation. Extensive tests on the EV motor drive under diverse conditions and scenarios were conducted, with performance data acquired using a precision data acquisition system (DAS). The collected data were preprocessed and refined for the modeling process. In response to feedback on input mechanisms, a data-driven model employing a Long Short-Term Memory (LSTM) network was developed to emulate the drive system's behavior. The LSTM-based drive system model underwent rigorous training with the actual data collected from the physical system. The overall accuracy of the LSTM model in replicating the motor drive system was 96.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>, affirming its high performance and reliability in mimicking the intricate drive system. The LSTM model was then combined with the motor simulation model, creating a hybrid-driven model. The hybrid model's success in closely aligning with the actual motor drive outputs underscores its comprehensive ability to effectively reflect both the nuanced dynamics and the operational intricacies of the EV propulsion system.