Artificial Intelligence for Optimal Sitting of Individual and Networks of Wind Farms

Abstract

Renewable Energy (RE)-based power production often comes with certain challenges in variability and uncertainty of generated electricity. One promising solution to tackle these challenges is developing a network of RE power plants with sites located far enough from each other that experience different weather patterns. Most of the site selection-related literature use Geographical Information Systems to determine the studied site RE suitability. This work converts the site selection into a numerical problem through a novel Networked Renewable Power Plant Site Selection model and solves it by employing optimization techniques. To enhance the accuracy of the results, it compares a set of criteria for individual and network of sites at different regions to determine the exact locations for RE plant developments. The Analytical Hierarchy Process is used for criteria weighing. The state-of-the-art meta-heuristic Bare Bones of Fireworks algorithm offer a simple, fast, yet accurate approach to solve the optimization. The proposed method is applied on North Carolina wind farms for both individual and a network of sites. The results identified the areas with the highest wind capacity potential for individual or a network of wind farms in North Carolina. The identified suitable areas were verified with Amazon Wind Farm US East.