This thesis is concerned with the foundations of the planning and control logistics systems and shows the interest of the PDS (Product-driven Systems). The development of such systems must consider the coherence between its different components. Centralized systems offer long term plans aiming optimal cost, but also stability and nervousness reduction. Distributed systems allow a fast reaction to perturbations occuring in the physical system (on the shop-floor or in the supply chain). Then, the hybridization of these two control systems is a way to increase productivity for logistic and industrial systems. The first chapter of the thesis describes the evolution of the production planning and control systems, to identify strengths and weaknesses of the different approaches proposed until our days and allow to define the general objective of the thesis. Chapter two analyses the state of the art concerning the tools for modeling such hybrid centralized/ distributed systems and also the concept of product-driven systems (PDS). This chapter serves as a basis to define the specific objectives of the thesis. Chapter three introduces the proposed modeling framework. This framework is based on a cybernetic approach, and more specifically on the Viable System Model (VSM). The chapter starts with a general presentation of VSM and then introduces a generic framework to model PDS. Finally, this chapter describes an application for production planning and control. Chapter four defines the different decision methods developed for the implementation of the generic model defined in the chapter three. At centralized and distributed levels these methods are based on mathematical programming models. Chapter five shows the main results of an application based on an industrial case which required the development of a simulation tool which considers short, medium and long terms variables for the different optimization models. These results show the interest of this type of hybridization.