The Accuracy and Error of Ground Penetrating Radar System with Machine Learning Support Vector Regression Technique

Abstract

Ground penetrating radar (GPR) is a non-destructive evaluation technique which involve knowledge of electromagnetic theory. Basically, there are three types of radar systems that are often applied in radar applications such as Monostatic, Bistatic, and Multi-static radar. Besides, in order to detect and locate the underground object, various technique has been implemented to cater issues in GPR such as clutter issues, inaccuracy in detect and locate the target object, signal loss, properties of soil and etc. In this paper, machine learning (ML) with support vector regression (SVR) is applied in GPR system using copper plat as buried object. Evaluation and validation on this method was carried out in term of S-Parameter and operating frequency. The scope of this work focuses on data analysis for the accuracy of object detection, validation graph and the error signal processing of Machine Learning in GPR system. The result of the experiment was shows low error, the validation point fit to hyperplane line (validation graph). Therefore, the output that expected for this research is validate the low false alarm rate of machine learning in GPR system.