Production research is currently experiencing the rise of a paradigm-shift from time and cost optimization towards flexibility of processes and customization. Modern organizational forms and global market factors in the manufacturing industry require higher flexibility and greater dynamics of production systems to be able to compete in today's highly competitive and globalized market environment. For this reason, new concepts have emerged, including rapid reconfiguration, mass customization and holonic enterprise. This flexibility makes it possible to respond to customers' needs while preserving the advantages of current mass production systems in performance, cost and quality. But to respond to increasing individualization of products and their adjacent processes, decentralized, flexible processes are needed for production environments that are characterised by a high amount of variants as well as a high degree of automation. The main goal is therefore to integrate static structural models and dynamic flow models into a joint model for object-oriented processes in production. While object-oriented and semantic concepts have been researched in-depth for programming languages and static models, object-orientation in workflows has not been extended to runtime model execution, meta-modelling and deep inheritance of processes. Hence, this thesis introduces an integrated concept for process modelling that builds on - and incorporates - concepts of the object-oriented paradigm like ontologies, inheritance, instantiation, aggregation, semantic relations and classification. These basic (meta-)modelling concepts are complemented by a runtime component and a method for distributed and decentralized application. This extended model achieves the goal of being able to model, execute, modify and adapt processes dynamically while the production is running. OMICRON, an object-oriented meta-model for interactive computer-based processes, introduces a generic, object-oriented component-relation meta-model which enables process models to be extended by new component types and new relation types that can be specialized using inheritance. It offers object-oriented concepts like inheritance, instantiation and aggregation. The thesis also solves the problem of cyclic relations, relation type definitions and their inheritance and instantiation. Complex components are created by aggregating atomic or even other complex components creating hierarchic structures. When a complex component, e.g. a variant process, is being specialized, the new specialized component inherits all steps and data parts of the above general/source component. Fundamental changes can be applied to all variants by modifying the basic process only. At runtime, object-oriented polymorphism enables the use of variants instead of the basic process using the type compatibility between the basic process and its variants. Processes can be instantiated on runtime in OMICRON. This means that every process instance is an object-oriented instance of its process. Changes in the process have direct influence on all its instances and specializations, while the instance in turn is executed and typed according to the process defined. The sequence of process steps is identified by relations of the relation type flow, connecting the process steps in the desired sequence. A token concept similar to that of the Unified Modelling Language 2 is used for activation of flows in execution. To benefit from these enhanced models and execute processes in decentralized environments, a message-based, decentralized infrastructure is created that enables a decentralized execution of the model in a dynamic, changing production environment. In this way, flexible and heterogeneous organisational forms like build-operate-transfer (BOT) and virtual enterprises can use common processes, share semantically defined datasets and use common semantic models and classifications. Process instances can flow from one peer to another carrying their semantic model with them that describes how they are interpreted and executed. An implementation scenario with existing partial systems, an application scenario at BSH Bosch und Siemens Hausgeräte GmbH and a detailed scenario on variants show the approach's applications, capabilities and limits of applicability. This fundamental work describes a broad concept from the basis to the application. It delivers a foundation for further work in research and system development of decentralized and flexible systems in modern production. This work highlights the need for further research and introduces new questions that arise from a different view on production systems and new possibilities created by advancing intelligent technologies. This includes decentralized scheduling, new organizational structures of systems that go beyond semantic models and - evolving from the Web 2.0 paradigm - new technologies and concepts for "Production 2.0".