Discrete Differential Dynamic Programming and Neural Network on Deriving a General Operating Policy of a Multiple Reservoir System : A Case Study in the Mae Klong System, Thailand

概要

論文の詳細を見る
A general operating policy for a multiple reservoir system operation is derived in this study using the combination of a discrete differential dynamic programming (DDDP) and a neural network (NN). The combination model is divided into three stages that consist of a DDDP algorithm, a NN algorithm, and a simulation model. In the first stage the optimal operating policies are derived from the DDDP in association of the genetic algorithm (GA) using the first set of data. The obtained optimization results are then supplied as the training patterns to the NN to derive a general operating policy. During the training process proceeds, the networks are evaluated by a simulation model to investigate their performance for another set of data. The demonstration is carried out through application to the Mae Klong system in Thailand. The objective of the optimization is to minimize the total irrigation deficits during the operation period. The results obtained in this study show that the combination model performs satisfactorily on deriving a reservoir general operating policy.
日本雨水資源化システム学会の論文