Fast RCS Data Generation Based on InfoGAN
- Resource Type
- Conference
- Authors
- Ren, Yao; Dang, Xunwang; Wang, Chao; Han, Xiaosheng; Hao, Jinchuan; Yin, Hongcheng
- Source
- 2021 Photonics & Electromagnetics Research Symposium (PIERS) Photonics & Electromagnetics Research Symposium (PIERS), 2021. :331-336 Nov, 2021
- Subject
- Engineered Materials, Dielectrics and Plasmas
Fields, Waves and Electromagnetics
Geoscience
Photonics and Electrooptics
Training
Radar cross-sections
Electromagnetic scattering
Neural networks
Generative adversarial networks
Data models
High frequency
- Language
- ISSN
- 1559-9450
Radar cross section (RCS) is one of the electromagnetic scattering characteristic of targets, and it’s essential in modern military confrontation. This paper introduces a network model based on Information Maximizing GAN (InfoGAN), which is aimed to achieve fast generation of the target’s RCS data in a certain range of parameters. The InfoGAN is trained by RCS data obtained by high frequency asymptotic simulation algorithms. The training process is offline. Then the RCS data of targets with different parameter values is generated by the feed forward network in the online process. The statistical characteristics such as mean value and variance of the generated RCS data are close to the simulation data, which can prove the accuracy and efficiency of such method.