Analysis and Synthesis of Multiband Sierpinski Carpet Fractal Antenna Using Hybrid Neuro-Fuzzy Model

Abstract

The paper presents the application of the hybrid neuro-fuzzy model for the analysis and synthesis of a square multiband Sierpinski carpet fractal antenna. For the analysis model, the antenna geometrical parameters were taken as the input, and the resonant frequencies were obtained as the output while for the synthesis model, the resonant frequencies were taken as the input, and geometrical parameters were obtained as the output. Also, a model was trained to obtain the return loss characteristics for the given set of geometrical parameters. The developed model was validated by comparing the resonant frequencies and radiation pattern of the simulated and fabricated antennas.

CORRELATION between MORPHOLOGY OF POLYMER COATING’S with TEXTURED SURFACE OF METAL AND ITS CORROSION RESISTANCE

“…The polymer surface is usually represented by fractal functions [8][9][10][11][12][13]. The most adequate method for modeling the surface morphology of a textured polymer coating is the diffusion limited aggregation method [14].…”

Ensemble‐based surrogate modeling of microwave antennas using XGBoost algorithm

“…Some of the mentioned methods can be named as; adjoints sensitivities into gradient-based optimization algorithms, 11 accelerating local procedures via sparse sensitivity updates, 12,13 the employment of machine learning methods, 14 as well as surrogate-assisted procedures involving both data-driven, [15][16][17][18] physics-based surrogates, 19 kriging, 20,21 radial basis functions (RBF), 16 Gaussian process regression (GPR), 22 neural networks, [23][24][25][26][27] support vector regression, [28][29][30] polynomial response surfaces, 31 or fuzzy models. 32 Ensemble learning is the mechanism by which many models (often called "weak learners") are strategically created and combined to solve a specific computer intelligence challenge and primarily employed to boost the efficiency of a model (classification, prediction, approximation of functions, etc.) or to lower the risk of a weak learner collection.…”

Artificial intelligence–based design optimization of nonuniform microstrip line band pass filter

“…This method is based on approximating sampled simulation data which provides advantages such as no need of expert knowledge of the studied problem, ability of adaptation between problems and applications, high computational efficiency, and so forth 8‐10 . Some of the commonly applied approximation/data‐driven models can be named as kriging, 11,12 radial basis functions (RBF), 13 Gaussian process regression (GPR), 14 neural networks, 6,15‐17 support vector regression, 18,19 polynomial response surfaces, 20 or fuzzy models 21 …”