Optimisation of reflection coefficient of microstrip antennas based on KBNN exploiting GPR model

• Published in: IET microwaves, antennas & propagation Vol. 12; no. 4; pp. 602 - 606
• Main Authors: Chen, Yi, Tian, Yu-Bo, Qiang, Zhe, Xu, Lan
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 28.03.2018
• Subjects: electromagnetic simulation; electromagnetic wave reflection; full‐wave electromagnetic simulation; Gaussian process regression; Gaussian processes; GPR model; KBNN; knowledge‐based neural network; microstrip antennas; neural nets; prior knowledge input method; reflection coefficient optimization; regression analysis; Research Article; telecommunication computing; reflection coefficient optimization; regression analysis; microstrip antennas; electromagnetic simulation; telecommunication computing; full-wave electromagnetic simulation; KBNN; GPR model; knowledge-based neural network; electromagnetic wave reflection; Gaussian processes; Gaussian process regression; prior knowledge input method; neural nets
• ISSN: 1751-8725; 1751-8733
• DOI: 10.1049/iet-map.2017.0282

When microstrip antennas (MSAs) are optimised, the full-wave electromagnetic simulation takes long time to get the result which is very time consuming. Hence, the knowledge-based neural network (KBNN) is used here to replace the electromagnetic simulation to shorten the calculating time and raise efficiency. However, prior knowledge of KBNN is always obtained by empirical formulas and neural networks, both of them are heavy and complicated. In this study, Gaussian process regression (GPR) is proposed to get the prior knowledge. The modelling of MSA is realised by the prior knowledge input method. The short optimal time and excellent optimal results prove that the proposed KBNN is a fast and effective method for the optimisation of MSAs.