Ground-Space Consistency Study of Large-Aperture Spaceborne SAR Antennas Based on Dynamics-Dimentionality Joint Analysis Method

Large-aperture parabolic satellite-borne SAR antennas represent a key future development direction for high-resolution Earth observation. Their in-orbit assembly technology is crucial for overcoming launch vehicle envelope constraints and enabling the construction of ultra-large space structures. However, traditional similarity criteria methods suffer from limitations in prediction accuracy and robustness due to distortion issues caused by inconsistent scaling of member diameters and lengths in scaled-down tests. This paper proposes a Least One-Out Cross-Validation Variable Power Law Similarity Modeling method (LOOCV-VPM). By introducing a diameter distortion factor and integrating LOOCV optimization, this approach effectively mitigates the impact of distorted models on predictions while enhancing the model's generalization capability for small samples and high-distortion scenarios. Numerical simulations and eight sets of scaled experiments validate that LOOCV-VPM demonstrates significant advantages in predicting natural frequencies and maximum deflections: Under optimal conditions, the minimum error can be reduced to below 1%, with extreme errors significantly converging. This provides an efficient, reliable, and scalable approach to addressing the similarity criterion issue in large-aperture parabolic satellite SAR distortion models, offering important reference value for ground testing and in-orbit applications of future space-based ultra-large SAR antennas.