LSTM을 이용한 해양시정 예측 방법

국문 초록 (Abstract)

해양시정은 항공기 이착륙, 항해, 해양레저 활동에 직접적인 영향을 미친다. 본 연구는 시정을 예측하기 위해 머신러닝 방법인 LSTM(Long Short-Term Memory)를 이용한다. 기상 데이터는 2012년부터 2017년까지 서해 덕적도의 군사 분야와 민간 분야 기상 데이터를 수집하여 전처리한다. 전체 6년의 기간 중 5년은 LSTM 모델에 학습시키고, 남은 1년은 검증 데이터로 사용한다. 그리고, 기존 예측에서 사용하였던 인자인 시정, 풍속, 기압, 습도, 기온 인자에 해수면온도와 해기차를 추가하여 성능 차이를 비교한다. 실험 결과는 실제값과 예측값의 차이가 수치적으로 적은 값을 보인다.

다국어 초록 (Multilingual Abstract)

Sea visibility directly affects aircraft takeoff and landing, navigation, and marine leisure activities. In this paper, we used the Long Short-Term Memory (LSTM) of machine learning method to predict sea visibility. We collected and then preprocessed the weather data of the Deokjeok Island in the West Sea from 2012 to 2017 in the military and civilian sectors. The LSTM model was trained for five years of the entire six-year period, and the remaining one year was used as verification data. We compared performance differences by adding sea surface temperature and air-sea temperature difference to factors used in existing predictions such as visibility, wind speed, atmospheric pressure, humidity, and temperature. The experimental results showed that the difference between the actual and predicted values was numerically small.

목차 (Table of Contents)

• 요약
• Abstract
• 1. 서론
• 2. 관련연구
• 3. LSTM을 이용한 시정예측 방법
• 4. 실험 및 평가
• 5. 결론 및 향후연구
• References

참고문헌 (Reference)

• 1 D. Bari, "Visibility Prediction based on kilometric NWP Model Outputs using Machine-learning Regression" 2018
• 2 T. Deng, "Visibility Forecast for airport operations by LSTM Neural Network" 2 : 466-473, 2019
• 3 L. Zhu, "The Application of Deep Learning in Airport Visibility Forecast" 314-322, 2017
• 4 K. Cho, "PM10 Particulate Matters Concentration Prediction using LSTM" 632-634, 2019
• 5 Korea Maritime Safety Tribunal, "Marine Accidents Statistics" Ministry of Oceans and Fisheries 13-25, 2021
• 6 Forecast Division, "Manual for Aviation Meteorological Observation" Aviation Meteorological Office 11-12, 2020
• 7 X. Li, "Long short-term memory neural network for air pollutant concentration predictions: Method development and evaluation" 231 : 997-1004, 2017
• 8 S. Hochreiter, "LONG SHORTTERM MEMORY" 9 (9): 1735-1780, 1997
• 9 J. Jonnalagadda, "Forecasting Atmospheric Visibility using Auto Regressive Recurrent Neural Network" 2020
• 10 R. Dewi, Prawito, "Fog prediction using artificial intelligence: A case study in Wamena Airport" 2020
• 11 Forecast Bureau, "Fog Analysis Technology and Prediction Method" 27 : 2015
• 12 J. Choi, "Evaluation of Contribution of Meteorological Observation Data to Weather Forecasts" KMA 43-44, 2018
• 13 W. Cho, "Estimation Method of River Water Level Using LSTM" 439-441, 2020
• 14 K. Miao, "Application of LSTM for short term fog forecasting based on meteorological elements" 408 : 285-291, 2020
• 15 H. Son, "Analysis of Long-Term Fluctuation Characteristics of Fog Generation in Korea" 229-230, 2010
• 16 S. Jo, "Analysis of Correlation of Wind Direction/Speed and Particulate Matter (PM10) and Prediction of Particulate Matter Using LSTM" 1649-1651, 2020
• 17 M. Krishan, "Air quality modeling using long short-term memory (LSTM) over NCT-Delhi" 12 : 899-908, 2019