Model-Based Controlling Approaches for Manufacturing Processes
Authors:
- Adrian Karl RüppelARAdrian Karl Rüppel
- Muzaffer AyMAMuzaffer Ay
- Benedikt BiernatBBBenedikt Biernat
- Ike KunzeIKIke Kunze
- Markus LandwehrMLMarkus Landwehr
- Samuel MannSMSamuel Mann
- Jan PennekampJPJan Pennekamp
- Pascal RabePRPascal Rabe
- Mark P. SandersMSMark P. Sanders
- Dominik ScheurenbergDSDominik Scheurenberg
- Sven SchillerSSSven Schiller
- Tiandong XiTXTiandong Xi
Open Access
- 30 December 2023
- book chapter
- Published by Springer Nature
- p. 221-246
Abstract
The main objectives in production technology are quality assurance, cost reduction, and guaranteed process safety and stability. Digital shadows enable a more comprehensive understanding and monitoring of processes on shop floor level. Thus, process information becomes available between decision levels, and the aforementioned criteria regarding quality, cost, or safety can be included in control decisions for production processes. The contextual data for digital shadows typically arises from heterogeneous sources. At shop floor level, the proximity to the process requires usage of available data as well as domain knowledge. Data sources need to be selected, synchronized, and processed. Especially high-frequency data requires algorithms for intelligent distribution and efficient filtering of the main information using real-time devices and in-network computing. Real-time data is enriched by simulations, metadata from product planning, and information across the whole process chain. Well-established analytical and empirical models serve as the base for new hybrid, gray box approaches. These models are then applied to optimize production process control by maximizing the productivity under given quality and safety constraints. To store and reuse the developed models, ontologies are developed and a data lake infrastructure is utilized and constantly enlarged laying the basis for a World Wide Lab (WWL). Finally, closing the control loop requires efficient quality assessment, immediately after the process and directly on the machine. This chapter addresses works in a connected job shop to acquire data, identify and optimize models, and automate systems and their deployment in the Internet of Production (IoP).Keywords
This publication has 48 references indexed in Scilit:
- Modelling and predictive control of a neutralisation reactor using sparse support vector machine Wiener modelsNeurocomputing, 2016
- XGBoost2016
- OntoSTEP: Enriching product model data using ontologiesComputer-Aided Design, 2012
- Reduction of fume and gas emissions using innovative gas metal arc welding variantsThe International Journal of Advanced Manufacturing Technology, 2010
- Lebenszykluskostenreduzierung durch zustandsorientierte InstandhaltungZeitschrift für wirtschaftlichen Fabrikbetrieb, 2009
- Machining Process Monitoring and Control: The State-of-the-ArtJournal of Manufacturing Science and Engineering, 2004
- Ontology of the Process Specification Language2004
- Tutorial overview of model predictive controlIEEE Control Systems, 2000
- An introduction to multisensor data fusionProceedings of the IEEE, 1997
- Industrial applications of model based predictive controlAutomatica, 1993