Smart Maintenance Ecosystems for Electric Vehicles: A Systematic Review of Artificial Intelligence and IoT-Based Approaches

Abstract

The global electric vehicle (EV) fleet now exceeds 40 million vehicles. Unplanned service represents 18-23% of the total cost of ownership for EVs. The individual applications of artificial intelligence (AI) and the internet of things (IoT) have both been shown to be capable of enabling predictive maintenance (PdM) for EVs. However, there is a currently an insufficient characterization of how they can be integrated together in EV specific architectures in the existing body of literature. Therefore, this systematic review will address three key gaps: first, it will address the gap of the absence of a common framework or architecture for the classification of AI-IoT architectures. Second, it will address the gap of the insufficient cross paradigm benchmarking of performance metrics such as detection rate, false positive rate, time to detect faults, etc. Third, it will address the gap related to the deployment barriers of using AI-IoT in PdM for EVs such as data privacy issues, latency constraints, and regulatory compliance requirements. Using the PRISMA guidelines from 2020, a comprehensive search was performed using IEEE Xplore, Scopus, Web of Science, and ACM digital library to obtain 117 peer reviewed studies published between 2018-2025. The results identified cross OEM federated learning, explainable AI for technician, decision support, and cybersecurity compliance (ISO 21434) as the top three priority areas for future research. Overall, the results provided a rigorous evidence based reference point for OEMS, infrastructure developers and policy makers who seek to operationalize AI-IoT PdM at scale.