An evolutionary optimization-learning hybrid algorithm for energy resource management

Rui Qi

doi:10.1016/j.swevo.2024.101831

An evolutionary optimization-learning hybrid algorithm for energy resource management

2025-02-01

0

PRE

AI

DOI：10.1016/j.swevo.2024.101831

Original

Original request for help

Save

Abstract

En 中文

Energy resource management (ERM) is important to an energy system. Effective management is hard to achieve because of the ubiquitous uncertainty of distributed energy resources and the massive number of participants, especially small storage devices (SSDs). Evolutionary computation algorithms have been applied to the ERM problem, but the high-dimensional nature of this problem makes them inefficient. In this paper, we propose an evolutionary optimization-learning hybrid algorithm to solve the ERM problem effectively and efficiently. A novel hybrid encoding scheme is proposed with two parts, optimization and learning. In the optimization part, an SSD integration strategy is designed to treat all SSDs as a whole, thereby significantly reducing the dimensions related to SSDs. In the learning part, the genetic programming algorithm is adopted to learn SSD state allocation rules automatically. Based on the hybrid encoding scheme, a delicately orchestrated evolution process is proposed to evolve these two parts simultaneously. Comparisons on a real-world distribution network located in Spain show that the proposed algorithm has outperformed the state-of-the-art algorithms.

Keywords:

Energy resources management

Genetic programming

Heuristic learning

Large-scale global optimization

An evolutionary optimization-learning hybrid algorithm for energy resource management

Abstract

Journal

Organization