MAEO: Multiobjective Animorphic Ensemble Optimization for Scalable Large-scale Engineering Applications

The Multiobjective Animorphic Ensemble Optimization (MAEO) framework tackles the No Free Lunch theorem by combining four state-of-the-art evolutionary algorithms — NSGA-III, CTAEA, AGEMOEA2, and SPEA2 — within an island-based architecture. Each island runs a different optimizer, and a parameter-free hypervolume indicator assesses performance. Individual scoring uses strict Pareto-rank with crowding distance and nadir-point proximity to maintain selection pressure. Extensive benchmarking on 12 DTLZ/ZDT functions across 36 dimensionality settings, evaluated via Wilcoxon signed-rank tests with hypervolume and inverse generational distance metrics, shows that MAEO achieves balanced convergence-diversity performance, outperforming or matching leading multiobjective optimization algorithms.

To demonstrate real-world impact, the team applied MAEO to equilibrium-cycle optimization of a small modular nuclear reactor. Eight discrete design variables and three objectives (levelized cost of electricity, peak soluble boron concentration, fuel cycle length) were optimized under two safety constraints. After roughly 40,000 computer simulations, MAEO identified core designs that lower both levelized electricity cost and peak boron concentration while preserving fuel cycle length and meeting all safety constraints. This practical success highlights MAEO's scalability for large-scale engineering applications with multiple conflicting objectives.