Improved simulated annealing for electromagnetic multimodal continuous problems
Salvatore Alfonzetti · 2004
The simulated annealing (SA) algorithm, is often used in the optimisation of electromagnetic devices because of its, simplicity of implementation and ability to find the global extremum of objective functions, especially in the presence of local extrema. Electromagnetic optimisation problems concern a continuous control space; hence SA must use a suitable generator of random changes in the system configuration. A modification of the downhill simplex method is multimodal when this algorithm is affected by premature convergence. An improvement on SA for continuous problems, which is able to prevent the algorithm from becoming trapped in local minima, is proposed.