Study for integration of simulation – optimization – artificial intelligence to improve operational performance of reservoir system in Ba river in dry season

Abstract

     This study sets up methodology and resolves the problem to improve hydropower generation of reservoir system. Methodology to resolve the problem is an integration of mathematical models: (i) simulation using HEC-ResSim; (ii) optimization using Dynamic Programming (DP) with Discrete Differential DP algorithm programmed by author; and (iii) Artificial Neural Networks (ANN). The methodology is applied to Ba river reservoir system consisting 06 reservoirs, and the objective is to maximize power generation and to satisfy minimum downstream requirement stipulated in scheduling inter-connected reservoirs. Testing result for Song Hinh reservoir operation shows that ANN has output very close to DP and achieves about 2% power generation higher than actual operation.