A Satellite Target Recognition Method Based on Broadband Radar Data and Gru Neural Network

Abstract

The invention provides a satellite target recognition method based on broadband radar data and GRU neural network, the method mainly includes two parts of data division and deep learning model, by introducing orbital height and radar observation angle information into satellite target recognition, realizing The division of broadband radar HRRP training data corresponds to the matching of test data; GRU neural network is used to extract the effective features of HRRP data, and the divided HRRP training data is used as input to determine the weight space of GRU neural network through network training in order to extract HRRP test The deep essential features of the data are input into the classifier to realize satellite target recognition. The present invention can make full use of the existing broadband radar data, and use the deep learning model of GRU neural network to extract the deep-level features of the radar HRRP data, which is beneficial to extract the effective features of the training data and reduce the search range and calculation amount of the test data , improving the accuracy of satellite target recognition.

Description

technical field [0001] The invention belongs to the technical field of radar automatic target recognition, in particular to a satellite target recognition method based on broadband radar data and a GRU neural network. Background technique [0002] Space is an important part of modern life style, national security and modern warfare. All countries in the world attach great importance to the development of their space capabilities. Therefore, aerospace technology has developed rapidly, and more and more advanced satellites have been launched into space, making space increasingly crowded, more competitive and confrontational. How to effectively perceive the space situation and improve space monitoring capabilities has become a major problem faced by major aerospace countries. Among them, satellite target recognition is an important function of the space monitoring information system. As an effective and important means to perceive the space situation, broadband radar has the c...

AI Technical Summary

Problems solved by technology

The design and selection of the above features takes a lot of time and energy, and there is a lot of uncertainty at the same time, and the generalization ability for different radars and different target types is weak; the third is to use machine learning algorithms to extract the dimensionality reduction features of HRRP, such as the main Component analysis (Principal Component Analysis, PCA), dictionary learning and manifold learning, etc.

But they are all shallow networks, it is difficult to effectively describe the HRRP characteristics of the radar, and obtain the deep feature information of the data

Images