Ontology Versioning for Managing Inconsistencies in Engineering Models Arising From Model Changes in the Design of Intralogistics Systems

Abstract

The interdisciplinary design of intralogistics systems (ILS) involves engineers from various disciplines, resulting in the generation of discipline-specific model files with overlapping information. For instance, a conveyor system can be represented from various perspectives, such as 3D-CAD models that capture its geometric information and discrete-event simulation models that depict the system's dynamic material flow performance. The growing demands for flexible reconfigurability and adaptability in intralogistics systems necessitate frequent updates to engineering models. However, these updates often result in potential model inconsistencies due to insufficient stakeholder communication.Detecting the impact of model changes and related inconsistencies is challenging in practice due to data heterogeneity and complex inter-model relations. To address these challenges, we propose an ontology-versioning approach that automates the identification of inconsistencies resulting from model changes. Our approach facilitates the integration of heterogeneous model data, enables database versioning, detects inconsistencies caused by model updates, and provides traceability for identified issues. The concept is evaluated utilizing models from a prototypical implementation on a lab-sized demonstrator.Note to Practitioners-In the industry, the current development of intralogistics systems often lacks automated synchronization of overlapping model information and consistent model interfaces, frequently leading to contradictions among the models. This has been identified as a significant source of errors in the design of both industrial and academic intralogistics systems, as revealed by a study involving intralogistics experts from different technical disciplines. Effectively managing model inconsistencies is crucial for project success, particularly when frequent model changes occur. A promising approach to tackle this issue is to systematically link model data from different disciplines, through which model inconsistencies caused by inadequate com-