The short answer: NO

To guarantee optimality of a solution (say single-objective) you have to base it on a mathematical model. GA's are in general not based on mathematical models, hence the solutions to the problems cannot be proven optimal. The same thing goes for Multi-Objective problems.

I would like to stress that this is NOT a critique of meta-heuristics. For many many interesting problems, there is no alternative. This problem is even more pronounced in Multi-Objective optimization, where the methods which guarantee optimal Pareto Fronts are even weaker ! Today there are NO general solvers for multi-objective or even two-objective problems.

I totally agree with Thomas, I would like to add that metaheuristics are designed to avoid as much as possible local optima, but do not garantee that the obtained ones are global.

Thank your answers 

One of the main reasons in using a meta-heuristics is to avoid being trapped in local optima. However, to get more reliable results, I recommend to solve your model with different meta-heuristics and compare the obtained results.   

There is a contradiction in your proposition already: The idea of a "global optimum" is, imho, to be a single point that dominates all others. The vast majority of multi-target problems does not have such.

If you are asking for the Pareto set of a given problem I would go along with the other commenters. But please mind: Your found solutions, even if they are "true ones" can only form a subset of all Pareto solutions, at least if your problem is not a discrete one with countable solutions. So you will always have to reflect yourself on the meaning of the identified solutions w/r/to the discussed pragmatic task to be solved.

Thank you very much for your answers