On Line Prediction Method of Several-Hours-Ahead Inflow Rate based on Rainfall Information

Abstract

A method for on line several-hours-ahead prediction of inflow rate into a dam was developed based on rainfall information. We analyzed actual data of inflow rate and rainfall during past several years at Midorikawa dam. Based on the analysis, we derived the inflow rate model and the stational rainfall model which expressed appropriately the relationship between the inflow rate and the rainfall at the dam area. By using a recursive least-squares estimation of these model parameters, we obtained the on line several-hours-ahead prediction of inflow rate. The results were evaluated based on the actual data in comparison with other three methods.
We can conclude that the proposed method gives an accurate prediction of inflow rate by a simple algorithm, adapts to slow changes of characteristics of the inflow rate and the rainfall, because the models are updated on line. The proposed method will be widely accepted in actual fields as an accurate and simple method for predicting inflow rate into dam.