EMI Radiation Prediction of Packages Based on Convolution Neural Network

With the increase of the current operating frequency, the electromagnetic interference (EMI) problem of extra-highspeed electronic equipment and systems becomes more and more complex, and the traditional analysis and design methods cannot efficiently deal with the current EMI problems. Therefore, this paper proposes a method of predicting electromagnetic radiation using a two-dimensional (2-D) convolutional neural network (CNN) in deep learning. This method can quickly and accurately predict the maximum 3-m pin map (PINMAP) package radiation electric field. By comparing the prediction results with the full-wave simulation results, both results are in good agreement, and the CNN prediction time is at the millisecond level. Therefore, the method that can accurately and effectively predict the radiation value has been verified.