DataVar: Closed Loop Data-Driven Manufacturing Variation Management using Industry 4.0

Complex manufacturing systems, made up of a chain of individual processes, experience some level of variation at various stages. The DataVar project will utilize large amounts of data collected throughout these chains of processes to optimize the overall capability of the manufacturing systems. The data will serve as an input to machine learning algorithms and statistical procedures that will learn the hidden patterns between numerous parameters. The result of this data processing, namely data-driven models, will be the central artifacts to realize smart decision support system and multi-process manufacturing control system. The latter will allow to adjust parameters in a manufacturing process based on measurements in the processes before it. Furthermore, the data will be used when designing new product to develop products that are less sensitive to the manufacturing process variations. Within the reporting period the focus has been the mapping of the current state of the data collection for both industry partners and their related analysis. Due to Covid-19 restriction, the work had to be carried out as far as possible off-site which has drastically limited the necessary effort to achieve the anticipated target. In addition, automotive crisis due to lack of semiconductors has resulted in reduced production and constitutes another burden for the project which is depending on production data.

The DataVar project will develop data-driven methods for a holistic manufacturing variation management with the aim for improving process capability and reducing variations of the existing multi-process production systems, and thus minimizing cost and resource usage. The project will utilize the large amount of production data collected at Benteler Automotive Raufoss AS and Hexagon Ragasco AS to allow for decision support with respect to optimising manufacturing variations and realisation of multi-process monitoring and control functionality. Artificial intelligence and statistical methods will be used to develop data-driven models that will serve as a basis for the variation-optimising decision-support technology. The two major value creation drivers from the developed methods will be the ability to perform intelligent manufacturing project cost prediction grounded in data and reduced time for readjustment of the production equipment.