Simulated Annealing with Adaptive Neighborhood using Fuzzy Logic Controller
Mohamed Lalaoui, Abdellatif El Afia, Raddouane Chiheb · 2018
Simulated annealing (SA) is a local search algorithm. It mimics the annealing process used in the metallurgy to approximate the global optimum of an optimization problem and uses the temperature to control the search. Unfortunately, the effectiveness of simulated annealing drops drastically when dealing with a large-scale optimization problem. This is due in general to a premature convergence or a stagnation. The both phenomenon's can be avoided by a good balance between exploitation and exploration. This paper focuses on this problem encountered by simulated annealing and try to heal it using the fuzzy logic controller by dynamically adapting the neighborhood structure of the simulated annealing. Empirical analysis was conducted on many numerical benchmark functions. The experimental results show that the dynamic controlling of neighborhood structure improves the quality of solution compared with the classical simulated annealing.