A deep learning approach for predicting the antenna pointing error caused by transmission faults with simulation data

Reflector antenna has been widely used in deep space exploration, radar warning, and other fields, all of which requires high pointing accuracy. The antenna elevation bearings are the key component that guarantees its pointing accuracy, while any degradation or fault can seriously affect the antenna's performance, leading to deviations in antenna pointing and instability during operation. However, the relationship between the antenna elevation bearing fault and its pointing accuracy remains unclear because there is insufficient experimental faulty transmission data and pointing error collected from the test-rig simultaneously. Therefore, this paper aims to establish a deep learning model-based relationship to reveal the underlying relationship between the antenna transmission faults and its pointing accuracy. By linking the two, transmission faults in key components can serve as a substitute for pointing accuracy as one of the criteria for antenna maintenance decisions, vibration signals, serving as a basis for fault diagnosis, can be collected and processed in real-time without the need for equipment shutdowns, undoubtedly bringing convenience to antenna maintenance providing a theoretical basis for the development of antenna maintenance strategies. In order to overcome the problem of insufficient data, this paper has established an antenna elevation system dynamic simulation model containing pre-defined transmission faults. Furthermore, to link antenna fault diagnosis with antenna pointing errors, a mathematical model for antenna axis error analysis has been established. Finally, labeled fault data and antenna pointing errors have been put into the deep neural network model for training to obtain the prediction model for predicting antenna axis error. The results showed that faults in the key transmission components have a significant impact on antenna pointing errors and the proposed deep neural network learning model exhibits a high predictive accuracy.