A Method for Identifying Space Objects

Abstract

The present invention provides a method for identifying space targets, comprising: preprocessing HRRP sample data envelope alignment, eliminating redundancy for HRRP sample data after envelope alignment preprocessing; wrapping HRRP sample data after eliminating redundancy Network amplitude preprocessing, the data set and the label set corresponding to the data set constitute the training set; set the network parameters, construct the GRU neural network; use the training set to train the GRU neural network, and reserve part of the data in the training set for After verification, the trained GRU neural network model is obtained; the HRRP data of the space target to be recognized is preprocessed and input into the GRU neural network model, and the one with the highest probability is taken as the space target recognition result. This method can automatically extract the deep essential features of radar HRRP data by constructing RNN neural network, and then complete the recognition of space targets. Eliminate the impact of manual selection of features on the recognition results, greatly reduce the time and effort of feature extraction, and improve the recognition accuracy of space targets.

Description

technical field [0001] The present invention relates to the technical field of radar target identification, in particular to a method for identifying space targets, in particular to a space target identification method for wideband radar one-dimensional high-precision range profile (High Resolution Range Profile, HRRP). Background technique [0002] As countries all over the world attach great importance to spaceflight and the rapid development of civil spaceflight, more and more satellites are launched into space, and space becomes increasingly crowded, more competitive and confrontational. How to effectively perceive the space situation and then control the space has become a major problem faced by all aerospace powers. As an important means of effectively sensing the space situation, broadband radar has the characteristics of all-day, all-weather, and high resolution. It plays an important role in the space target recognition system and has been widely used. [0003] The...

AI Technical Summary

Problems solved by technology

However, these features are all designed and selected based on the way people are in the loop. It takes a lot of time and energy, and there is a lot of uncertainty at the same time. The generalization ability for different radars and different target types is weak.

In addition, some scholars have proposed the use of BP neural network and multi-layer perceptron for radar target recognition, but these are shallow networks, and it is difficult to obtain deep feature information of data.

Images