Cryogenic distortion compensation in guided ultrasonic waves using a Physics-Informed Generative Adversarial Network

Reusable launch vehicles undergo severe thermal cycling, and cryogenic storage conditions introduce strong, nonlinear distortions in guided ultrasonic wave signals used for structural health monitoring. These temperature-induced effects obscure damage sensitive features and limit the robustness of both conventional compensation techniques and purely data driven learning models. This paper proposes a physics informed machine learning framework, termed the Guided Wave Physics-Informed Generative Adversarial Network, to compensate cryogenic guided ultrasonic wave distortions in a physically consistent manner. The artificial intelligence component is realized through generative adversarial learning and is applied to cryogenic guided-wave temperature compensation The model adopts a residual generative formulation in the Hilbert-envelope domain and integrates adversarial sequence learning with wave-physics priors through curvature based regularization and peak morphology constraints. These physics-informed components explicitly suppress nonphysical temporal artifacts and preserve the modal structure of guided-wave responses, enabling stable compensation across varying cryogenic conditions. Experiments conducted on an aluminum alloy plate under liquid-nitrogen cooling–heating cycles demonstrate that GW-PIGAN achieves high-fidelity reconstruction while maintaining consistent performance across varying reference temperatures. Compared with purely data-driven baselines, the proposed physics-informed formulation exhibits improved stability, physical plausibility, and generalization under extreme thermal conditions. By embedding guided-wave physics into generative learning, the proposed framework provides a practical and interpretable physics-informed machine learning approach for cryogenic structural health monitoring. It also establishes a methodological basis for future monitoring applications in reusable aerospace structures under extreme cryogenic conditions.