Predicting Rectangular Patch Microstrip Antenna Dimension Using Machine Learning
Nazmia Kurniawati, Arif Fahmi, Syah Alam
Abstract
When designing a microstrip antenna, the designers determined the desired parameters. However, the simulation software can only give the parameters result based on the given dimension. Therefore, optimization is required to meet the desired parameters. The designers usually do the optimization by the trial-error process. This research conducts machine learning implementation to predict the microstrip antenna dimension. The study focused on rectangular patch microstrip antenna with resonant frequency ranged from 1-8 GHz. The dataset used to make the prediction is obtained from simulation with antenna width ranged from 19-63 mm and length 10-54 mm. There are four algorithms employed: decision tree, random forest, Support Vector Regression (SVR), and Artificial Neural Network (ANN). Among all algorithms, random forest with estimator 15 gives the best result with Mean Square Error (MSE) value is 3.45. From the obtained result, the researchers can estimate the rectangular patch microstrip antenna dimension based on the desired parameters, which can’t be done by the antenna simulation software before.