Abstract : Interoperability between enterprise applications is a multifaceted problem. In response, we have investigated the semantic alignment of business process and object knowledge supported by enterprise applications. This knowledge is heterogeneous across business and is implemented in different ways, generating semantic conflicts due to different interpretations of the actors. We study the identification of correspondences and/or conflicts between semantic knowledge, especially in the field of Manufacturing Execution System (MES). We propose an approach to formalize knowledge as a standardized repository ISO 62264 compliant: the "MES repository", and an approach to enhance the MES repository by successive alignments with other standards and/or repositories. Thus, we formally define a meta-model that describes the semantic relationships between elements of the aligned repositories, and semantic control rules to keep MES repository consistency and its ISO compliance. Finally, we present an approach to align customer needs with the MES repository, in order to assist the implementation of enterprise applications. This PhD thesis was conducted as part of a research and development project supported by the economic Cluster Edit (Rhône-Alpes, France), and its results have been used in the new proposal "on demand MES".