A novel self-adaptive learning charged system search algorithm for unit commitment problem

Abstract

Charged System Search (CSS) is a new evolutionary algorithm inspired by the interaction between charged particles. This paper presents a modified CSS (SACSS) algorithm, which highly improves the performance of CSS and applies it to solve the unit commitment (UC) problem. In order to achieve better performance and higher speed in solving the UC problem, a self-adaptive reformation technique with tree updated schemes has been implemented. The proposed algorithm has been tested on 10, 20, 40 and 100 unit systems for one-day and scheduling horizon. The results are compared with the solutions obtained from other methods such as Lagrangian Relaxation (LR), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Bacterial Foraging (BF) and Shuffled Frog Leaping Algorithm (SFLA). The results show the high performance and convergence speed of SACSS.