This paper aims at closing the gap between early phases (e.g. design) and later phases (e.g. procurement or production) of the Product Development Process (PDP) by proposing a Virtual Product Model (VPM) as a collection of individual components (VPMCs) without the need for a static structure. Based on an analysis of the requirements on product development in the automotive industry, the main problems we observe are limited transparency, limited continuity, and limited reusability throughout different phases of the PDP. Virtual Product Model Components (VPMCs) can be used in different products and allow the reflection of changes throughout the PDP as well as the derivation of domain-specific views on the overall product at runtime. We illustrate these concepts by use case scenarios derived from an analysis of automotive product development practices.

No abstract

Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über abrufbar.

Today, top-down processes, centralized IT infrastructures, and one-vendor strategies prevail in Product Lifecycle Management (PLM) of large multi-brand Original Equipment Manufacturer (OEM) groups. Given the usually decentralized organisation and structures and processes that emerge from cross-brand collaboration, these centralized approaches are challenging the adaptiveness and performance of the OEM groups.In this concept paper, we investigate challenges for cross-brand and cross-domain cooperation from the perspective of processes and IT systems. The main contribution of this paper is that we motivate and outline a novel technical architecture approach combining service-orientation with an event-driven software architecture and asynchronous event processing to support users from different brands and domains in their collaboration along the development process. We analyse related work on collaboration models as well as on event processing and discuss our approach before the background of the state of the art. Finally, we summarize our findings and give an outlook to future research venues.

Modern machine learning methods have the potential to supply industrial product lifecycle management (PLM) with automated classification of product components. However, there is only little work in the literature on this topic. We propose to apply supervised machine learning on component meta-data. By analysing an industrial case study, we identify requirements and opportunities for automating classification, e.g. in part numbers and product structures. We validate our novel approach through a classification experiment comparing four machine learning methods on a realistic component dataset.