镍基多晶高温合金作为航空发动机及燃气轮机热端部件的关键材料，其高温疲劳行为直接决定了构件的服役安全与寿命。 本文综述了镍基多晶高温合金疲劳力学响应、微观组织演化、损伤机制与失效分析、寿命预测与数值模拟等方面的研究现状和进展。 研究表明，合金在高温下的宏观力学响应展现出强烈的路径依赖性、非比例效应及相位角敏感性，其根源在于热-力载荷与微观组织的深度耦合。 在微观层面，疲劳过程伴随着 γ′/γ″强化相的动态演化及位错机制的竞争，这些演化直接导致了循环硬化 / 软化等不可逆性能退化。 在损伤与失效层面，载荷相位通过调控应力 -温度场，主导了沿晶与穿晶断裂模式的竞争，并可能诱发寿命交叉现象。 在机理溯源层面，内部缺陷是裂纹萌生的主导因素，而氧化与环境效应则显著加速裂纹扩展。 在预测模型层面，研究范式正从宏观唯象模型向融合物理机制的跨尺度集成模型与数据驱动智能方法演进。 宏观模型提供了工程可用的寿命预测框架；而基于晶体塑性理论、机器学习及疲劳 /蠕变双指标参数的先进方法，提升了对损伤演化量化表征与全寿命预测的精度。 本文最后展望了通过深度融合多尺度物理机理与人工智能技术，实现从经验模型向机理与数据双驱动的精准寿命预测的未来趋势。

Nickel-based polycrystalline superalloys are critical materials for hot-section components in aeroengines and gasturbines, whose high-temperature fatigue behavior directly affects the service safety and lifespan of these components. Thisarticle systematically reviews the current research status and advances in fatigue mechanical response, microstructuralevolution, damage mechanisms and failure analysis, lifetime prediction, and numerical simulation of nickel-basedpolycrystalline superalloys. Studies have shown that the macroscopic mechanical response of these alloys at hightemperatures exhibits strong path dependence, nonproportional effects, and phase-angle sensitivity, which originate from thedeep coupling between thermomechanical loading and the microstructure. At the microscale, the fatigue process involvesthe dynamic evolution of γ′/γ″ strengthening phases and competition among dislocation mechanisms, directly leading toirreversible performance degradation, such as cyclic hardening/softening. With respect to damage and failure, load phasing, by modulating the stress-temperature field, dominates the competition between intergranular and transgranular fracturemodes and may induce lifetime crossover phenomena. In terms of mechanism tracing, internal defects are the dominantfactors in crack initiation, whereas oxidation and environmental effects significantly accelerate crack propagation. Inpredictive modelling, the research paradigm is evolving from macroscopic phenomenological models to multiscale integratedmodels that incorporate physical mechanisms and data-driven intelligent approaches. Macroscopic models provide anengineering-usable framework for lifetime prediction, whereas advanced methods based on crystal plasticity theory, machinelearning, and fatigue-creep dual-parameter indicators improve the accuracy of damage evolution quantification and full-lifeprediction. Finally, a future trend toward precise lifetime prediction driven by both mechanisms and data, achieved throughthe deep integration of multiscale physical insights and artificial intelligence technologies, is proposed.