Ansatz für ein prozessintegriertes Qualitätsregelungssystem für nicht stabile Prozesse

As production processes become more complex it is necessary to gain knowledge about the causalities between input and output parameters. Additionally an in-time problem solving is essential for a quality oriented production. These accounts require the implementation of a process integrated quality control system. The goals are the recognition of the causalities, inspection of the process’ parameters and taking action in case of irregularities within the process. In addition the quality control system has to be able to act predictively in arranging the following process steps. SPC is a standard tool for statistical process control but requires a stable process. Anyhow tools for quality control are needed which base on non stable processes. The following dissertation describes an approach for a quality control system in a non stable process environment. The example is a casting process in a light metal foundry. The developed quality control system is made up of two contrarily directed quality control loops. A reactive quality control loop is used for determination of the causalities and for taking action for optimizing the production process. In addition a predictive quality control loop is implemented which initiates the following activities in the process chain. The necessary inspection efforts for the parts’ delivery are included in the predictive loop. The essential part of both quality control loops are the analytical methods for model building. As appropriate tools Data Mining methods have been established. The methodical analytic approach has been approved on several analyses in the BMW light metal foundry and fully been integrated into the problem solving process. The following approach is applied: Learning from the past for designing the present to improve the future.