Artificial neural network-assisted supercell thunderstorm algorithm for optimization of real-world engineering problems

Abstract

This study presents an artificial neural network (ANN)-assisted modified supercell thunderstorm optimizer (MSTO) for solving complex industrial component optimization problems. Inspired by the natural phenomena of spiral motion, tornado formation, and jet streams within supercell thunderstorms, the STO algorithm is enhanced with ANN integration to improve exploration, exploitation, and convergence rates. The algorithm is validated across five constrained engineering problems: cantilever beam optimization, industrial grinding cost optimization, tubular column design, diaphragm spring weight minimization, and fin and tube heat exchanger (FTHE) cost optimization. These results confirm MSTO's superior performance over recent metaheuristics, highlighting its potential for high-precision, stable, and efficient solutions across structural, thermal, and mechanical design domains.