First important question to ask about any evolutionary algorithm experiment:
Did you check your experiment against a simulated annealing process?
It's simpler to implement, converges faster and gives better results. So far I am unaware of any optimizing problem that was solved better with an evolutionary/genetic algorithm than simulated annealing. That is, unless the evolution part (crossover, mainly) is inextricably connected to the problem at hand and therefore other algorithms are simply not suitable to even try.
Did you check your experiment against a simulated annealing process?
It's simpler to implement, converges faster and gives better results. So far I am unaware of any optimizing problem that was solved better with an evolutionary/genetic algorithm than simulated annealing. That is, unless the evolution part (crossover, mainly) is inextricably connected to the problem at hand and therefore other algorithms are simply not suitable to even try.