A comparative study of combined surrogate models and evolutionary algorithms for ship resistance optimization

The optimal parameterization of ship resistance poses a complex engineering problem involving issues such as solution space sampling, to address this problem, this study presents a solution by comparing six optimization frameworks formed by two surrogate models and three evolutionary algorithms. The proposed solution involves the DE-RBF combination. Building upon an existing sample set, surrogate models, including Radial Basis Function (RBF) neural network and Kriging model, were employed to establish mathematical relationships between input design variables and resistance performance. Using these surrogate models as a foundation, the study applied Particle Swarm Optimization (PSO), Differential Evolution (DE), and Artificial Bee Colony (ABC) to optimize water resistance. Experimental results confirm the significant optimization effect of the surrogate model-evolutionary algorithm framework for ship resistance. Among them, the DE-RBF combination exhibits superior performance in finding the optimal parameter combination for ship resistance performance. Numerical experiments demonstrate that this frame effectively reduces ship resistance, with a 56% reduction compared to the initial resistance value of the original ship model.