Optimizing Constrained Mixed-Integer Nonlinear Programming Problems Using Nature Selection

Many practical engineering optimization problems involving real and integer/discrete design variables have been drawing much more attention from researchers. In this paper, an effective adaptive real-parameter simulated annealing genetic algorithm (ARSAGA) was proposed, applied to cope with constrained mixed-integer nonlinear programming problems. The performances of this proposed algorithm, including reliability and convergence speed are demonstrated by examples. It is noted that the intrinsic parameters of this novel hybrid algorithm, i.e. population size and frozen number, were discussed and appropriate parametric combinations of both parameters were also suggested in this paper. These illustrative simulations demonstrate that the results through the proposed method are very reliable and reasonable.