Improved particle swarm optimization algorithm solving optimization problems with mixed variables and constraints

Aiming at the optimization problems with mixed variables and constraints in engineering design, a particle swarm optimization based on simulated annealing was proposed. By introducing the simulated annealing algorithm, the locations of the particles, which had stopped the evolution, were regenerated in order to enhance the global search ability. In view of the characteristics of optimal solution in the border of feasible region, combined with a strategy of adaptively maintaining the proportion of unfeasible solutions, the constraints were dealt with by using individual comparative norms. Considering the characteristics of mixed-variable optimization problem, the algorithm could search in the discrete space through the transfer function, to ensure the feasibility of solution. Simulation results show that the algorithm can find the optimal solution quickly and accurately, with good stability.