Machine Learning Techniques for the Prediction of the Magnetic and Electric Field of Electrostatic Discharges
Georgios Fotis, Vasiliki Vita, Lambros Ekonomou
Abstract
The magnetic and electric fields of electrostatic discharges are assessed using the Naïve Bayes algorithm, a machine learning technique. Laboratory data from electrostatic discharge generators were used for the implementation of this algorithm. The applied machine learning algorithm can be used to predict the radiated field knowing the discharge current. The results of the Naïve Bayes algorithm are compared to a previous software tool derived by Artificial Neural Networks, proving its better outcome. The Naïve Bayes algorithm has excellent performance on most classification tasks, despite its simplicity, and usually is more accurate than many sophisticated methods. The proposed algorithm can be used by laboratories that conduct electrostatic discharge tests on electronic equipment. It will be a useful software tool, since they will be able to predict the radiating electromagnetic field by simply measuring the discharge current from the electrostatic discharge generators.