Simulation-based generation of robust heuristics for self-control of manual production processes: A hybrid approach on the way to industry 4.0.

点击翻译按钮获取中文摘要

In the sense of Industry 4.0, complete self-configuration and self-control of production planning and control (PPC) is proposed to compensate for disturbances while at the same time meeting high logistical targets. Numerous concepts exist for the implementation of such a PPC, which predominantly require permanent data availability as in future cyber-physical systems (CPS). This permanent data availability (real-time data and real-time interaction) is currently only partially feasible in industries with mainly manual production processes and will remain the subject of research and development for the coming years. On the one hand, the technological implementation and permissibility of a continuous digital recording of human work is unclear. On the other hand, interfaces for application programming of machines are simply not freely accessible. In order to take advantage of the self-control of production processes with regard to the improvement of logistical target variables under process uncertainties even in the transition and development phase from existing production systems to CPS, the partial automation of PPC functions becomes necessary, which makes it possible to speak of a hybrid PPC.As a concept approach, the project pursues the central and knowledge-based configuration of robust self-control for sequencing and resource allocation based on existing information or information that is easy to collect in the production system for a defined period (similar to rolling-wave planning). This approach is intended to be a first step in the implementation of the Industry 4.0 vision in relation to PPC, especially for industries with the characteristics of a high proportion of manual work and high process instability. In order to distinguish the planned research project from existing research in the field of self-control, it is based on significantly lower data quality and availability, as will still be the same conditions in this industrial sector in the future. Under these conditions, completely new methods and procedures for implementing the conceptual approach must be developed.Due to the limited real-time data and data exchange possibilities, the focus of the project is thus on the development of robust heuristics for self-control in order to make the best possible sequence and resource allocation decisions without permanent reconfiguration. Disturbances must therefore be included prospectively in the generation of these heuristics. Furthermore, implicit knowledge from simulation and production through machine learning is to be used for the continuous improvement of self-control.