Many real-world decision problems involve multiple and conflicting objectives, which need to be optimized simultaneously while respecting various complex constraints. In this paper, we investigated how to solve these kinds of problems by using genetic algorithms. A new fitness assignment method for multiple objective optimization problems - the adaptive hyperplane-based fitness assignment method - was proposed in order to give a genetic algorithm the search pressure toward the positive ideal point in the objective space. An adaptive penalty function was used together with the adaptive hyperplane method in order to let genetic search explore the optima through both feasible and infeasible areas in the solution space. A Pareto solution reserving method was incorporated into normal genetic algorithm loop in order to maintain a set of Pareto solutions during the evolutionary process.