A Systematic Review and Unified Framework for Predictive Maintenance in the Railway Domain

Predictive maintenance has become an important strategy for improving asset reliability and safety, particularly with the increasing integration of artificial intelligence (AI) into maintenance processes. In the railway domain, where maintenance activities can account for up to 40% of the overall budget allocated across the V-cycle of rolling stock development, the adoption of smart maintenance strategies is increasingly important. However, an integrated understanding of predictive maintenance approaches across the railway domain remains lacking. This study provides an in-depth review of predictive maintenance research in the railway domain through a PRISMA-guided systematic search of three major scientific databases: Scopus, Web of Science, and ScienceDirect. A total of 61 full-text articles were retained and analyzed in detail. The reviewed studies are synthesized according to their methodological approaches, namely data-driven, model-based, and hybrid approaches, and their distribution across railway subsystems is examined. The analysis reveals a concentration of research on rolling-stock components, particularly wheelsets and switch machines, while comparatively less attention is devoted to less-instrumented but similarly safety-critical assets, such as traction, signaling, and door systems. Based on these findings, a conceptual hybrid predictive maintenance framework is proposed by synthesizing the approaches identified in the literature. Finally, the study highlights key research challenges and identifies open research questions and future research directions, including the need for more comprehensive, subsystem-specific, and practically validated predictive maintenance solutions.