抄録
The transition toward inverter-dominated power systems challenges conventional reliability theory, which was developed largely for synchronous-generator-dominated systems and cannot fully capture the coupled effects of fast inverter controls, multi-timescale dynamics, operating uncertainty, and component degradation. This review examines how reliability emerges across control, dynamic, thermal, and aging processes and how these processes can be integrated into a consistent analytical framework. Existing knowledge is synthesized across four dimensions: inverter control behavior, reliability mechanisms, assessment methodologies, and enhancement strategies, with particular emphasis on their physical basis and characteristic timescales. The synthesis shows that grid-following and grid-forming inverters create distinct reliability pathways through interactions among electromagnetic, electromechanical, thermal, protection, and degradation processes, while divergent findings often arise from differences in operating conditions, model fidelity, and timescale representation. Current assessment approaches span probabilistic analysis, dynamic simulation, multi-timescale modeling, multiphysics and physics-of-failure methods, and data-driven artificial intelligence, each involving trade-offs among physical fidelity, computational efficiency, and online applicability. Reliability enhancement is also evolving from isolated converter-level measures toward coordinated control, system-level support, condition-aware operation, and reliability-oriented hardware design, yet these approaches remain weakly integrated with lifecycle assessment and operational decision-making. Key gaps include the absence of a unified reliability theory, incomplete cross-timescale and multiphysics coupling, limited integration of control dynamics with component degradation, and weak links between reliability assessment and real-time operation. Future research should prioritize lifecycle-aware modeling, digital-twin-enabled online assessment, physics-informed data fusion, and adaptive reliability-aware control. By reframing reliability as a cross-timescale process linking mechanisms, assessment, and intervention, this review clarifies the theoretical and methodological basis for reliability-oriented operation and design of inverter-dominated power systems.
