Analysis of Electromagnetic Decoupling of an Isolation Barrier Using Machine Learning

In this paper, a machine learning technique is used to analyze the decoupling performance of an isolation barrier that prevents electromagnetic interference between cables in nuclear power plants. To train the machine learning model, the mode-matching method is also employed to generate a training dataset including the decoupling capability corresponding to design parameters. After confirming that the deep neural network model is well-trained, we utilize the resulting model to predict the decoupling capability of an isolation barrier in the variation of design parameters. We conclude that the machine learning model can be possibly used to obtain information on precautions against electromagnetic coupling problems in cabling applications in nuclear power plants.