Cylindrical Dielectric Resonator Antenna Optimization: A Machine Learning Perspective

Designing a Cylindrical Dielectric Resonator Antenna (CDRA) is complicated by its many physical characteristics and ideal values. Using real-world data to calculate a precise numerical response from experimental investigations is tricky. Machine learning methods solve this issue. After proper training, the machine learning approach eliminates the need for antenna design iterations, resulting in a fast and precise solution. The S11 parameter of a cylindrical dielectric resonator antenna is determined in this study by using a frequency range of 4-12 GHz, a height range of 9-11 mm, and a radius range of 7-9 mm. Additionally, multiple machine learning algorithms are employed to find the parameter. This research presents several different types of algorithms, such as Decision Tree, KNN, Random Forest, XGBoost, and Deep Neural Networks. The HFSS software is used to generate the data set that will be utilized in the training and validation of these algorithms. In addition, the findings of the proposed methodologies are compared to the values that were simulated, and it is shown that both sets of results are in agreement. Nevertheless, the outcomes of the KNN, Random Forest, and Deep Neural Network procedures are noticeably superior to those of the other methods.