An adaptive replacement strategy-incorporated particle swarm optimizer for wind farmlayout optimization

The wind farm layout optimization (WFLO) aims to maximize power generation of a wind farm by optimizing the location of wind turbines. Traditional mathematical methods cannot provide a satisfactory solution for a wind farm due to the high complexity of the problem. Meta-heuristic algorithms have been used to optimize it. Particularly, genetic algorithms (GA) have been widely used and obtained success in WFLO problems. However, GA still suffers from the issues of insufficient optimization efficiency. In this study, a genetic learning particle swarm optimization with an adaptive strategy, termed AGPSO, is proposed to optimize WFLO problems. The strategy adaptively adjusts the location of the worst turbine to improve the conversion efficiency of a wind farm. Four wind scenarios, including single wind speed with single wind direction, single wind speed with uniform multiple wind directions, single wind speed with nonuniform multiple directions, and multiple wind speeds with multiple wind directions scenarios ones, are utilized to verify the effectiveness of AGPSO and the effect of different wind scenarios for it. Twelve constraints and three different scales are used to further verify the robustness of AGPSO and the effect of wind turbine location on WFLO problems. Extensive numerical experiment results demonstrate that AGPSO performs significantly better than other twelve state-of-the-art competitors in terms of conversion efficiency under different wind farms, wind scenarios, and constraints. AGPSO obtains the best average of 89.92%, 92.90%, 95.39%, and 90.75% conversion efficiency in a wind farm with 25 wind turbines under four wind scenarios, respectively.