LSTM网络在台风路径预测中的应用

Application of LSTM in Typhoon Path Prediction

台风路径实质为二维平面上一段曲线,根据2条台风路径曲线的相似度可以判断其数值相似和形态相似,由此利用动态规整算法可以从历史台风数据库筛选出与目标台风相似的台风路径。同时考虑到台风路径信息的时间关联性,提出了长短时记忆网络预测模型。利用历史台风的经纬度信息,预测台风未来6小时位置信息,对比传统基于相似度预测台风路径方法,长短时记忆模型能够有效提高台风路径预测精度,模型更加稳定高效。

The typhoon paths are essentially curves on a two-dimensional plane. According to the similarity of the two typhoon path curves, the numerical similarity and morphology similarity can be judged. Therefore, the dynamic time warping algorithm can be used to select the typhoon path similar to the target typhoon from the historical typhoon database. Meanwhile, considering the time correlation of typhoon path information, we put forword a long short-term memory model. The paper uses the latitude and longitude information of historical typhoon to predict the position information of the typhoon in the next 6 hours. Compared with the traditional method of predicting typhoon path by similarity, the long short-term memory model can effectively improve the prediction accuracy of typhoon paths, and the model is more stable and efficient.