From Predictive to Prescriptive Maintenance in Power Generation: an Integrated Five-Layer Framework Evaluated through Systematic Comparative Analysis

Prescriptive Maintenance (RxM) has emerged as an important development in power generation maintenance by extending data-driven maintenance beyond failure prediction toward action-oriented decision support. Its potential lies in helping organizations move from anticipating equipment failures to considering what maintenance actions should be taken, under which operational conditions, and with what constraints. However, existing RxM frameworks remain fragmented, particularly in relation to modular design, execution mechanisms, feedback processes, and integration with complex power plant environments. This study conducts a systematic literature review guided by five research questions to examine the conceptual foundations of RxM, its enabling technologies, architectural components, implementation challenges, and validation approaches. Based on the synthesis, a modular five-layer RxM framework is proposed, consisting of data collection, analytics, decision-making, execution, and feedback. The framework was conceptually assessed through Systematic Comparative Analysis against seven benchmark studies. The comparison suggests that the proposed framework provides broader architectural coverage by linking cloud-edge infrastructure, analytical models, multi-criteria decision logic, execution support, and feedback-based learning. Rather than demonstrating empirical effectiveness, the study offers a structured pathway for future practical implementation in power generation, subject to simulation, expert validation, pilot testing, and field-based assessment. The findings may inform researchers and practitioners seeking to advance from predictive to prescriptive maintenance while maintaining attention to transparency, system integration, cybersecurity, and empirical validation.