This research article introduces a design of a meander line incorporated L-band microstrip patch antenna aimed at reducing its size and volume while enhancing its performance characteristics. The proposed design, furthermore is optimized using popular machine learning techniques. The initial simulation presents a conventional rectangular patch antenna characterized by a gain of 2.23 dB and a resonant frequency of 1.38 GHz, with an impedance bandwidth of 20 MHz. The specified dimensions of this design are 150 × 150 × 1.6 mm3. This study aims to develop a smaller prototype antenna by incorporating symmetric slots and meandering lines into the patch structure. Consequently, the overall dimensions of the prototype are reduced to 40 × 45 × 1.6 mm3, resulting in an impressive volume reduction of 92%. Furthermore, this innovative design yields a substantial increase in bandwidth of 300% and a gain enhancement of 124%, while maintaining the original resonant frequency i.e. 1.38 GHz. Then, four popular machine learning techniques namely Regression Trees, Support Vector Machines under Regression Model (SVR), Kernel Least Squares, and Artificial Neural Networks using Bayesian optimization are compared for antenna parameter optimization. It is observed that the incorporation of different machine-learning approaches in parameter optimization allows us to evaluate potential calculation errors and provide comparatively accurate results.
扫码关注我们
求助内容:
应助结果提醒方式:
