The early 1990s onset trend towards customized production led to the development of production strategies such as Computer Integrated Manufacturing, Total Quality Management, Lean Management and Agile Manufacturing. Agile Manufacturing aims mainly to a higher flexibility of production, ie the ability of a production system to adapt to unforeseen changes in production situations. On the other hand, the rapid development of IT has led to an improvement of the computer-based production planning and control systems and the development of knowledge-based approaches. This work starts at this point. The aim is to contribute to the solution of research questions that deal with knowledge-based approaches to the analysis and improvement of production systems. The focus is on flexibility-oriented production systems in the process industry, which have been considered less in the research. The core of this work is to answer the question which knowledge-based approaches are suitable to design and improve the operational production management in a flexibility-oriented process industry. After a characterization of the flexibility-oriented process industry a method for collaborative MRP is developed that is based on existing supply chain management approaches. By improving the transfer of knowledge delivery reliability was significantly increased by the example of a supply chain in the aluminum industry. Furthermore, this work deals with data mining and explains the use of data mining methods to discover potentially useful knowledge that is implicitly present in existing data of production planning systems and can help to derive scheduling rules and improve logistical targets. The application of the concept results in an approach to a knowledge-based production planning system and brought a significant reduction in processing time and resources for a defined process route in a flexibility-oriented integrated aluminum rolling mill. With the developed so-called lean data mining approach wastes in production can be discovered using data mining with the aim to derive improvement actions and to design processes with minimum waste, like short lead times, that means lean.