RNN-LSTM-TimeSeries

This is for my master project in UoG

LSTM TimeSeries prediction 时间序列预测

Data—Preprocess 数据预处理

数据预处理包含：时间序列的插值，流速的异常点检测与替换，移动事件的检测，对事件进行标签处理。
Data preprocessing includes: Interpolation of time series, Detection and replacement of abnormal points of streamwise_ve, Detection of moving events, and label the events.

Model-Building 模型建立

模型采用三层LSTM模型，优化器和LOSS函数参看代码 训练100次.维基百科介绍长短期记忆
The Model in this project is built on LSTM model,here is the wiki link LSTM-Wiki The parameters of the model are lr=0.001,nb_epoch=100, batch_size=1.

Effects 效果图

Contact 联系方式

Email： [email protected]