Study on the Inversion Method of Rendezvous Parameters Using Near-Field Echo Characteristics

In this paper, a rendezvous parameter inversion method based on one-dimensional convolutional neural network is proposed by using near-field rendezvous echo data. The simulated near-field rendezvous echo data of typical targets are preprocessed and data enhanced to form a training set. A one-dimensional convolutional neural network for inversion of rendezvous parameters is designed and trained. The influence of different network structures on the accuracy of inversion parameters is analyzed. The results show that the inversion method based on near-field echo characteristics has good robustness and accuracy.