http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
일기 예보를 이용한 기계 학습 기반의 레일 온도 예측 모델 개발
홍성욱(SungUk Hong),이현우(Hyunwoo Lee),박찬(Chan Park),정현석(Hyunsuk Jung),박철정(CheolJeong Park),김형욱,홍성경(SeongKyung Hong),조성진(Seong J Cho) 대한기계학회 2018 대한기계학회 춘추학술대회 Vol.2018 No.12
In railway safety management, rail-temperature is directly affects the buckling of rails is very important. The previous rail-temperature prediction models have been developed for predicting the rail-temperature, but lack of versatility and performance. In this study, we suggest a novel rai-temperature model based on machine learning with high performance and universally available. We constructed a monitoring station to measure rail-temperature and local weather conditions, and then performed the measurements for 10 months. Using measured data, we found new features that affect rail-temperature. The model was constructed using XGboost, one of the machine learning methods, and the optimal hyperparameter was selected through k-fold cross validation. Our novel rail-temperature model can be used anywhere in the world because rail-temperature can be predicted using only weather forecast data. In particular, the novel rail-temperature prediction model can predict the local rail temperature after 67 hours from the weather forecast of Korea Meteorological Administration. It is expected to help planning railway management plan.