流出・洪水 (2)

[II-158] Trial of rainfall runoff prediction by LSTM of deep learning on time series data

*西本 吉伸¹ (1. 開発電子技術(株))

Keywords:precipitation, flow rate prediction, Deep learning, LSTM

It is important to predict the inflow to the reservoir in the flood discharge operation of a large reservoir. As a method of forecasting rainfall inflow, we tried to apply Long and Short Memory 'LSTM' which is one of time series deep learning, using time series data of precipitation in the basin area. As a result, we confirmed that it was possible to create a highly reproducible prediction model by setting the learning conditions appropriately.