Radar high-resolution one-dimensional range profile target recognition method and system based on dual parallel networks

By adopting a dual parallel network structure in radar target recognition, the timing characteristics of HRRP data are extracted using LSTM and GRU networks, and weighted fusion is carried out through dynamic gate structures, the problem of low recognition accuracy in the prior art is solved, and higher recognition accuracy and robustness are achieved.

CN114594440BActive Publication Date: 2025-03-25Chinese People's Liberation Army Cyberspace Force Information Engineering University
2 Cites 0 Cited by

Patent Information

Application Number
CN202210072547.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-01-21
Publication Date
2025-03-25
Estimated Expiration
2042-01-21

AI Technical Summary

Technical Problem

When using HRRP data for target recognition, the data dimensions are high and information is redundant, resulting in low recognition accuracy and inability to fully grasp the characteristics of the target, affecting the recognition effect.

Method used

The HRRP sequence data is encoded and extracted features through multi-channel encoding and weight adjustment, and the robustness of the network model is improved. The timing characteristics of the data sequence are extracted using LSTM and GRU network structures, and the dynamic gate structure is weighted and fusion is carried out to improve the recognition accuracy.

Benefits of technology

It improves the accuracy and robustness of target recognition, can effectively identify it when the amount of training data is small, and is suitable for practical scenario applications.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention belongs to the field of radar target recognition, and particularly relates to a radar high-resolution one-dimensional range image target recognition method and system based on dual parallel networks, by constructing a target recognition model and using the measured data of a preset band broadband radar test field as sample data to train and optimize the model, wherein the model comprises: two parallel branch networks for extracting and mapping the correlation features of the input radar high-resolution one-dimensional range image data sequence respectively, a fusion layer for performing data weighted fusion on the outputs of the two branch networks, and an output layer for classifying and identifying the output sequence after weighted fusion; the trained and optimized target recognition model is used to identify high-resolution one-dimensional range image targets within the detection range. The present invention extracts features by performing multi-channel encoding on HRRP sequence data, and can further improve the robustness of the network model and the target recognition accuracy through weight adjustment, which is convenient for application in actual scenarios.
Need to check novelty before this filing date? Find Prior Art