진화전략과 쌍 선형 회귀성 신경망을 이용한 새로운 태풍 예측 시스템의 개발 = A Development of the New Typoon Prediction System Using the Evolutionary Strategy and Bilinear Reccurent Neural Network

다국어 초록 (Multilingual Abstract)

In this paper, we predict the typhoon's track and center pressure after 12 hours using the evolution strategy and the bilinear recurrent neural network(ES_BLRNN). We plot the predicted track of the typhoon along with the predicted center pressure in the 3-dimensional space and we compare the predicted positions and pressures of a given typhoon with its real position and pressure, The dying position of the typhoon is predicted under the conditions that we discover the regularity of the variation of the center pressure of the typhoon. Comparing the predicted track and the real track of a given typhoons, it is shown that the prediction system using the evolution strategy and bilinear recurrent neural network is superior to the existing prediction system such as the numerical approach, the neural network approach and the cliper equation.