Large antennas are key front-end components in communication, tracking, and observation systems, and their structural reliability is essential to long-term safe and reliable operation. In service, structural monitoring data reflect the combined effects, including servo motion, task-dependent working modes, temperature variation, wind loading, and potential structural degradation. However, applying fixed thresholds to such mixed monitoring data may lead to false alarms or missed anomalies. This work proposes a data-decoupling framework for structural health monitoring of large antennas based on measured multimodal sensor data. First, servo-related signals, including motor speed, motor current, command angle, measured angle, and tracking error, are used to characterize the antenna operation, and a two-level segmentation method combines operating-mode information with servo features extracted from sliding windows to identify distinct operating states. Second, state-dependent monitoring features and threshold boundaries are constructed separately for different antenna axes and motion states. Finally, screened standby samples are used to identify temperature-dominated structural responses, and a temperature-bin compensation method is established to obtain residual responses for wind-load analysis. The results show that the azimuth and elevation axes exhibit different motion and load-response patterns, and that temperature compensation significantly reduces the dominant thermal trend in standby strain responses. Overall, the proposed data-decoupling framework provides an interpretable basis for state-aware monitoring, environmental disturbance separation, and subsequent anomaly warning of large antenna structures.

Large antenna, Structural health monitoring, Multimodal data, Operating state segmentation, Temperature compensation