Generation of Rule-Based Variance Schemes Towards a Data-Driven Development of High-Variant Product Portfolios
DS 134: Proceedings of the 26th International DSM Conference (DSM 2024), Stuttgart, Germany
Year: 2024
Editor: Harold (Mike) Stowe; Christopher Langner; Matthias Kreimeyer; Tyson R. Browning; Steven D. Eppinger; Ali A. Yassine
Author: Thorsten Schmidt; Steffen Marbach; Frank Mantwill
Series: DSM
Institution: Helmut-Schmidt-Universitat, Hamburg, Germany
Page(s): 079-088
DOI number: 10.35199/dsm2024.09
Abstract
The management of product portfolios with high variance is complex due to the underlying constraints as well as prone to errors due to uncertainty and dynamics in the historical data. The documentation of these product portfolios is usually done in so called rule-based variance schemes, which are specified in Conjunctive Normal Form (CNF). For better planning, feasibility assessments and simulation of mentioned product portfolios as well as for the generation of a synthetic reference dataset for downstream analytic processes and training of Artificial Intelligence applications, this paper considers the development of a CNF-Generator for the creation of rule-based variance schemes as a contribution to data-driven product development. In this paper, particular consideration is given to satisfiability, reasoning, model counting and plausibility of the generated data. The implementation and evaluation are embedded in an use case of the mass customizable variant configuration being found in the automotive industry.
Keywords: Model Checking and Counting, Data-Driven Product Design, Rule-based Product Configuration, CNF-Generator, Automotive Industry