This work is conducted in the context of the development of a decision support tool for real time rescheduling in dense railway systems, which was applied at SNCF Transilien. More precisely, the objective is to propose a new selection process that provides the decision maker with a subset of non-dominated solutions while taking into account both the objectives and decision types. This means that the selected solutions must be sufficiently diversified to offer the decision maker multiple alternatives with different types of decisions. To do this, a clustering approach is proposed to select K solutions from a Pareto front that is obtained by solving an industrial instance. Numerical results show that the developed approach identifies interesting solutions.