A novel Differential Evolution algorithm based on simulated annealing
Wang Peichong, Qian Xu, Yu Zhou, Ning Li · 2010
Differential Evolution (DE) which has been focused on computation intelligence is a new swarm intelligent algorithm by simulating intelligence of population after GA and PSO etc. It is more robust and efficient. Because the differential degree of individuals is minimized in the last, the diversity of population will be reduced and DE will converge ahead of schedule. It is well known that simulated annealing(SA) can accept both better solution and worse solution according to definite probability. This mechanism can maintain the diversity of the population so that can avoid appearing premature convergence. This paper proposes a novel hybrid DE (DESA) by combining original DE algorithm and simulated annealing strategy. At last, it is proved that the DESA algorithm is effective in solving global optimization problem by testing on five Benchmark functions.