Toward an Intelligent Diagnosis and Prognostic Health Management System for Autonomous Electric Vehicle Powertrains: A Novel Distributed Intelligent Digital Twin-Based Architecture

In the era of rapidly advancing technological landscapes, the fusion of cutting-edge sensor technology, innovations in electric vehicles, and artificial intelligence are driving the growth of diagnosis and prognostics health management techniques tailored for electric vehicle applications. Such a system plays a pivotal role in ensuring the reliable, efficient, and safe operation of complex engineering systems like electric vehicles by facilitating fault detection, diagnosis, recovery, and prognostics. The main challenge in the field of diagnosis, prognostics and health management is to pinpoint the optimal methods and areas of application in electric vehicle to extract the most pertinent information. In this paper, a novel intelligent digital twin-based architecture for intelligent diagnosis, prognostics and health management of electric vehicle powertrains is proposed, it aims to address the aforementioned challenges. The proposed architecture is based on a multilayered approach that integrates multiple intelligent digital twin models at different abstraction levels and leverages transfer learning strategies to enable individualization and servitization of the proposed system. Furthermore, pivotal technologies enabling this approach are delved into, along with a discussion of inherent challenges and emerging opportunities. The findings underscore the benefit of using distributed computing and real-time prognostics and health management information to enhance the reliability, efficiency, and availability of the electric vehicles, concurrently ensuring driver safety. The proposal is enriched with a case study that illustrates the practical application of our architecture in a real-world electric vehicle monitoring system scenario, showcasing its potential for customization and service-oriented features, and provides foresight into vehicle parameter trends.