그래픽 데이터를 활용한 기계학습 적용 파랑 예측 기초 연구

Abstract

The Convolution Neutral Network(CNN) has been widely utilized to classify the image categories in machine learning. We initially conducted some numerical trainings in order to figure out the application validity of the machine learning in the overall sea condition prediction. A long-crested wave condition was firstly producted with Airy wave superpositions. We constructed the suitable the CNN structure with proper training parameters. Various image filtering techniques were adopted and evaluated in wave field prediction. Various short-crested waves were also modelled and trained in wave height classification. While the middle range of wave height categories were not predictable, the highest and lowest wave height categories showed the good prediction results. Some variations on training parameters and structures were lastly performed in order to increase the prediction accuracy in short-crested wave trainings while their effectiveness was not significant.