A New Method for Pointing Direction Recognition of Spacecraft Parabolic Antenna Payload Based on the Measured RCS Sequences

A pointing direction recognition approach for spacecraft parabolic antenna payload based on measured RCS sequences is presented in this paper. An electromagnetic simulation model is built for analyzing the RCS characteristic of spacecrafts who carry parabolic antenna payload. The simulated RCS shows that the step effect of RCS sequence measured by radar could be used as the criterion to recognize whether the line of sight (LOS) of earth radar is in the aperture of spacecraft parabolic antenna. Aiming at classifying whether the LOS of earth radar in the aperture of spacecraft parabolic antenna, this paper designed a hybrid-dimensional convolutional neural network (HCNN). Then the pointing direction recognition problem of spacecraft parabolic antenna payload is converted to a dichotomous problem where the RCS sequence is the input signal and the LOS of earth radar in the aperture of spacecraft parabolic antenna payload is the expected classified result. The proposed method was trained and validated on measured datasets. The results show that the proposed method yielded the classification accuracy as 90.5%. Moreover, it can tell that the proposed method has the limit of over fitting, which should be solved later in the engineering application.