MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641099282 A) filed by Srinivas Maganti on August 17, 2026, for A Generative Machine-Learning Method For Automated High-Dimensional Antenna Geometry Optimization.
Inventor includes Srinivas Maganti.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: ABSTRACT: Title: A Generative Machine-Learning Method for Automated High-Dimensional Antenna Geometry Optimization A generative machine-learning method for automated high-dimensional antenna geometry optimization is disclosed. The method receives an antenna geometry comprising a plurality of geometric parameters and electromagnetic performance data corresponding to previously evaluated antenna configurations. The geometric parameters and electromagnetic responses are encoded and used to train a generative machine-learning model configured to generate candidate antenna geometries according to predetermined electromagnetic targets. A machine-learning performance prediction model estimates electromagnetic characteristics of generated candidates, thereby reducing the number of candidates requiring computationally intensive electromagnetic simulation. Selected candidates are evaluated using electromagnetic simulation and a multi-objective optimization function considering parameters including reflection coefficient, resonant frequency, bandwidth, gain, efficiency, radiation characteristics, and physical constraints. Simulation results are fed back to update the learning models and guide subsequent candidate generation. The iterative process generates an optimized antenna geometry satisfying electromagnetic and manufacturing requirements while reducing computational effort associated with high-dimensional antenna design.
Disclaimer: Curated by HT Syndication.
