3D EM data driven surrogate based design optimization of traveling wave antennas for beam scanning in X-band: an application example

Abstract

In this paper, design and optimization of a microstrip elliptic traveling wave antenna (TWA) are presented for the beam scanning in X-band as an application of 3D data driven surrogate based design optimization technique. A novel Modified Multi-Layer Perceptron (M2LP) algorithm is utilized as a fast and accurate black-box modeling and compared to the alternative Multi-Layer Perceptron (MLP), Support Vector Regression Machine (SVRM), Gradient Boosted Tree algorithms for the generation of surrogate model of the TWA design. In order to have a computationally efficient modeling, Latin-Hyper Cube Sampling LHS method is utilized to obtain the training and test data from 3D CST Microwave numerical computation tool. A novel meta-heuristic, population based optimization algorithm, Invasive Weed Optimization (IWO) is applied to build up M2LP model for the determination of the optimal geometric design parameters. The optimum TWA model has total physical size of 100 mm x 20 mm with the operation frequency band between 8.5 and 12 GHz and measured overall gain of 7.1-11.8 dBi. The steerable radiation pattern characteristic is measured to be between -70 and 25 degrees. Thus, the proposed TWA design points out superior radiation performance for X band radar applications.