Enhanced simulated annealing for constrained design problems

Abstract

Real-world design problems such as welded beam design, pressure vessel design, and three-bar truss design were recognized as challenging tasks due to the associated constraints. This work aims to develop an Enhanced Simulated Annealing (ESA) optimizer that embeds the Q-learning algorithm in order to control its execution at run time. Specifically, the Q-learning algorithm is used to guide SA toward the best performing value of the annealing factor at run-time. To assess the performance of ESA, a total of four popular constrained engineering design problems were conducted. The outcomes reveal the ability of ESA to significantly overcome the standard SA as well as other optimization algorithms such as GWO, PSO, and CLPSO.