There were many works for solving the constrained optimization problems. Among them, the methods which apply Evolutionary Algorithms have shown promising results. However, most of the previous works would produce the solutions which violate given constraints. Other methods would need heuristic knowledge or complex parameter tuning to be applied on the real world circumstances. In this thesis we propose a new scheme for solving the constrained optimization problem with evolutionary algorithms. This scheme is mostly concerned on how to find the feasible region without user intervention. We apply the Pareto ranking with niching algorithm to constrained optimization problem. The Pareto ranking algorithm deals every constraints equally, not preferring some constraints randomly. And the niching algorithm spreads individuals out the whole feasible region. Niching is accomplished by fitness sharing on decision variable space. For computing sharing parameter on decision variable space, an adaptive method is proposed. After finding the feasible regions by Pareto ranking and niching method, the proposed scheme optimizes for the given objective function. Experimental results show that this scheme shows capable of producing feasible solutions with reasonably less modeling parameters.