An Intelligent Method to Solve the Eigenvalues of Characteristic Modes Using PointNet
- Resource Type
- Conference
- Authors
- Kong, De-Hua; Zhang, Wen-Wei; He, Xiao-Yang; Cao, Jia-Ning; Xia, Ming-Yao
- Source
- 2023 International Applied Computational Electromagnetics Society Symposium (ACES-China) Applied Computational Electromagnetics Society Symposium (ACES-China), 2023 International. :1-2 Aug, 2023
- Subject
- Engineering Profession
Fields, Waves and Electromagnetics
Point cloud compression
Three-dimensional displays
Neural networks
Memory management
Scattering
Computational electromagnetics
Eigenvalues and eigenfunctions
Artificial Intelligence
Point Cloud
Characteristic Mode Analysis
- Language
This paper presents an artificial intelligent method that can be used to solve the eigenvalues of characteristic modes (CM) of a PEC target. This method uses PointNet neural network with point cloud as input and eigenvalues as output. The process of meshing and matrix generation in the traditional method is skipped, and the eigenvalues are directly calculated, thus accelerating the process of CM analysis. A dataset consisting of PEC quadrangular platforms is constructed. Numerical experiments demonstrate the feasibility of the proposed method. This method is useful in the field of antenna design and target scattering analysis.