Hybrid Algorithm of Particle Swarm Optimization and Grey Wolf Optimizer for Reservoir Operation Management

  • PDF / 1,175,698 Bytes
  • 16 Pages / 439.642 x 666.49 pts Page_size
  • 78 Downloads / 247 Views

DOWNLOAD

REPORT


Hybrid Algorithm of Particle Swarm Optimization and Grey Wolf Optimizer for Reservoir Operation Management Saad Dahmani1

· Djilali Yebdri1

Received: 24 April 2019 / Accepted: 14 October 2019 / © Springer Nature B.V. 2020

Abstract Metaheuristics are highly efficient optimization methods that are widely used today. However, the performance of one method cannot be generalized and must be examined in each class of problems. The hybrid algorithm of particle swarm optimization and grey wolf optimizer (HPSOGWO) is new swarm-based metaheuristic with several advantages, such as simple implementation and low memory consumption. This study uses HPSOGWO for reservoir operation optimization. Real-coded genetic algorithm (RGA) and gravitational search algorithm (GSA) have been used as efficient methods in reservoir optimization management for comparative analysis between algorithms through two case studies. In the first case study, four benchmark functions were minimized, in which results revealed that HPSOGWO was more competitive compared with other algorithms and can produce high-quality solutions. The second case study involved minimizing the deficit between downstream demand and release from the Hammam Boughrara reservoir located in Northwest Algeria. A constrained optimization model with non-linear objective function was applied. Based on the average solutions, HPSOGWO performed better compared with RGA and was highly competitive with GSA. In addition, the reliability, resiliency, and vulnerability indices of the reservoir operation, which was derived from the three algorithms, were nearly similar to one another, which justified the usability of HPSOGWO in this field. Keywords Reservoir optimization model · Evolutionary algorithm · Genetic algorithm · Gravitational search algorithm · Constrained optimization

 Saad Dahmani

[email protected]; [email protected] Djilali Yebdri [email protected]; [email protected] 1

Laboratoire de Gestion et Traitement des Eaux (LGTE), Universit´e des Sciences et de la Technologie d’Oran Mohamed Boudiaf (USTO-MB), BP 1505, El M’naouer, 31000 Oran, Alg´erie

S. Dahmani, D. Yebdri

1 Introduction Optimization algorithms aim to determine values of decision variables that maximize or minimize an objective function with or without constraints. Several algorithms, such as linear programming (LP; Loucks 1968), non-linear programming (NLP; Arunkumar and Jothiprakash 2012), and stochastic dynamic programming (SDP; Stedinger et al. 1984), have been used to solve problems in reservoir optimization management for a single or a system of reservoirs. Despite the performance and wide usability of the classical methods, they suffer from several disadvantages, such as inability to solve problems with non-linear and non-convex objective functions (i.e., in the case of LP) and the curse of dimensionality (Sharif and Swamy 2014). Evolutionary optimization algorithms (EOAs) are new methods for searching the approached solutions for optimization problems with reasonable computation ti