Scattering response modeling scheme based on combined neural network inspired by the equivalent scattering center

Area of Science:

• Electromagnetics and Radar Systems Engineering
• Computational Intelligence and Machine Learning

Background:

• Modeling complex radar target scattering patterns is challenging, especially with limited training data.
• Existing methods like the Geometrical Theory of Diffraction (GTD) and polynomial scattering models have limitations in accuracy and efficiency.

Purpose of the Study:

• To propose a novel scheme for modeling complex radar target scattering patterns using a small training dataset.
• To decouple and independently model the aspect and frequency domain dependencies of radar target scattering.

Main Methods:

• Employing an ideal equivalent scattering center as a transfer function to represent frequency domain response.
• Utilizing neural networks to model the complex aspect dependency.
• Designing a parameter extraction algorithm based on Orthogonal Matching Pursuit (OMP) for parameter continuity.

Main Results:

• The proposed scheme effectively models complex scattering patterns with limited data.
• Decoupling aspect and frequency dependencies leads to independent and accurate modeling.
• The new model demonstrates significantly better performance compared to GTD-based and polynomial scattering center models.

Conclusions:

• The novel scheme offers a more efficient and accurate approach to radar target scattering modeling.
• The method's ability to handle small datasets and its faster training speed make it a valuable advancement.
• This technique provides a robust solution for complex electromagnetic scattering analysis.