Operating a reservoir system based on the shark machine learning algorithm
Document Type
Article
Publication Date
1-1-2018
Abstract
The operating process of a multi-purpose reservoir needs to develop models that have the ability to overcome the challenges facing the decision makers. Therefore, the development of a mathematical optimization model is crucial for selecting the optimal policies for the reservoir operation. In the current study, the shark machine learning algorithm (SMLA) is proposed to develop an optimal rule for operating the reservoir. The SMLA began with a group of randomly produced potential solutions and later interactively executed the search for the optimal solution. The procedure for the SMLA is suitable to be applied to a reservoir system due to its ability to tackle the stochastic features of dam and reservoir systems. The major purpose of the proposed models is to generate an operation rule that could minimize the absolute value of the differences between water release and water demand. The proposed model has been examined using the data of the Aswan High Dam, Egypt as the case study. The performance of the SMLA was compared with the performance of the most widespread evolutionary algorithms, namely, the genetic algorithm (GA). Comprehensive analysis of the results was performed using three performance indicators, namely, resilience, reliability, and vulnerability. This work concluded that the performance of the SMLA model was better than the GA model in generating the optimal policy for reservoir operation. The result showed that the SMLA succeeded in providing high reliability (99.72%), significant resilience (1) and minimum vulnerability (20.7% of demand).
Keywords
Aswan High Dam, Semi-arid region, Water deficit, Water release
Divisions
fac_eng
Funders
“Bold Grant” of University Tenaga Nasional (10289176/B/9/2017/57),University of Malaya Research Grant “UMRG” (RP025A-18SUS)
Publication Title
Environmental Earth Sciences
Volume
77
Issue
10
Publisher
Springer Verlag (Germany)