Improving the Computational Efficiency for Optimization of Offshore Wind Turbine Jacket Substructure by Hybrid Algorithms

Abstract

When solving real-world problems with complex simulations, utilizing stochastic algorithms integrated with a simulation model appears inefficient. In this study, we compare several hybrid algorithms for optimizing an offshore jacket substructure (JSS). Moreover, we propose a novel hybrid algorithm called the divisional model genetic algorithm (DMGA) to improve efficiency. By adding different methods, namely particle swarm optimization (PSO), pattern search (PS) and targeted mutation (TM) in three subpopulations to become “divisions,” each division has unique functionalities. With the collaboration of these three divisions, this method is considerably more efficient in solving multiple benchmark problems compared with other hybrid algorithms. These results reveal the superiority of DMGA in solving structural optimization problems.