Abstract: A large phased array radar requires high-precision stable operation on a dynamic platform. The structural rigidity and strength of the radar directly affect its safety, reliability, and precision performance. This paper takes the structure of the radar as the research object, The finite element model was established, and static and dynamic mechanical analyses, including self-weight, ship motion, and vibration, were performed. Additionally, based on this finite element model, an order-reduced digital twin model for the radar structure was developed using a hybrid approach combining Singular Value Decomposition (SVD) and neural networks. The timeliness and accuracy of this reduced-order model were validated. This work effectively supports the structural design of the radar system and the deployment of digital twin technologies, offering valuable references for the structural design and digital twin modeling of similar equipment.