End-to-End Traceability and Defect Prediction in Automotive Production Using Blockchain and Machine Learning

Abstract

With the increasing number of digital technologies in production, more and more detailed product data can be collected, measured, and analyzed. Coupled, this forms the basis for future data services during the product life cycle. In this way, suppliers, OEMs, and vehicle users achieve a new quality and transparency of their respective internal value chains and enable completely new improvements. However, many of the current product data services only serve specific services or partial aspects of a vehicle with a digital twin but do not enable an overall continuous digital profile. One reason is the multitude of different systems in the automotive data landscape and along the entire product line, from product design to the end of life.

In addition, the vulnerability of the data increases with each touchpoint in the value chain and thus in the field use. Missing transparency and trust of data are also major challenges for companies and users beyond the vulnerability of blockchain-based in-vehicle data use and thus also hamper the potential of a new era of collaboration and cooperation structures in the use of new digital technologies in production. This research aims to conceptualize a design approach for continuous and reliable data for the overall digital product profile along the entire product life cycle. In doing so, a design and validation approach for an overall digital product profile based on the integrated automation of blockchain technology and machine learning is developed and the individual areas of action are demonstrated in a concrete demonstrator. Virtual product twins in combination with structural digital twins should enable the new quality of end-to-end traceability and defect prediction as part of the overall digital product profile for automotive OEM data services, supplier services, and customers.