Precision Improvement of Dam Inflow Prediction by Prediction Learning and Deep Learning Yachiyo Engineering Co., Ltd.

Abstract

It is an urgent task to improve the precision of the dam inflow prediction that can support the effective use of dams in the increasing trend of rain as the external force by the climate change. We have made the dam inflow prediction models so far, whose parameters are optimized based on observation data. But in actual management, we have considered input data as prediction data and used those models. We can't expect the effectiveness of the current process that pursues the phenomenon reproducibility in actual management because there is a difference of characteristics between observation data and prediction data. This thesis shows that prediction learning causes the dam inflow prediction to get high precision and needs deep learning.