In this paper we propose a new integration framework model for simplifying the feasible space exploration and product optimization in early design phases. Hence, modeling and optimizing tasks are core activities in this framework. Currently, system engineering problems are modeled and optimized using a wide range of domain-specific languages. One should not duplicate these languages by creating a new system engineering language capable of modeling and optimizing every aspect of a system. Thus we combine the UML2 language and the formalism of Constraint Optimization Problems (COPs). UML2 is a visual modeling language, which provides a set of diagrams and constructs for modeling the major aspects of a product. In order to optimize design parameters, we reformulate some of this modeling knowledge into a COP. A COP may be defined as a regular constraint satisfaction problem (CSP) augmented with a set of objective functions. Thus the optimization problem to be solved is stated declaratively with acausal constraints. Then, COP solvers are based on generic solving algorithms computing a set of optimal solutions. In this paper, generic concepts integrating variability modeling concepts and based on architecture description languages are introduced. We also briefly describe transformation strategy using ATL language to perform a bidirectional mapping between UML2 constructs and the corresponding COP models