Project Details
Projekt
XDP-Opt: Experience-Based Design Process Optimization for Industrial Manufacturing
Subject Area
Engineering Design, Machine Elements, Product Development
Term
since 2024
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 543073350
The XDP-Opt project aims at optimizing the product design process in existing industrial manufacturing systems by a combination of Artificial Intelligence (AI) methods as the foundation of a hybrid Interactive Design Decision Support System (IDDSS). The investigated approach relies on the formalism of the digital life cycle record, which is based on the concept of the asset administration shell, to collect all data and information about the production process. XDP-Opt considers the entire product design process and derives possible optimization potentials in the respective production phases, which in turn are presented to the product designer on an ad-hoc basis via a recommendation system as decision option (hybrid decision support). An important element of decision support is the consideration of circularity, sustainable production, and the resource-efficient use of production units. XDP-Opt investigates a combination of different AI methods as foundation for decision support. Federated learning from extreme sensor data (in particular, quality control data) is investigated to capture experience from data coming from a wide variety of production domains to overcome potential data protection concerns in industry. Using Case-Based Reasoning (CBR) as another AI method, this experience shall be made available in the IDDSS system to support new design decisions based on previous similar decision cases. We expect that the use of the CBR methodology enables new problems to be solved efficiently and old errors to be avoided, while at the same time, explainability of the results is provided. As a third AI method, symbolic reasoning methods, in particular AI planning and constraint-based approaches, are investigated to enable a more comprehensive exploration of the design space beyond the experience that is available from the production data. This means that new, previously unexplored design options can also be considered based on the available information about the manufacturing domain. The investigated AI methods are experimentally implemented to combined into the hybrid IDDSS, which suggests design options to the product designer as a human in the loop in real time and provides information about their effects. The proposed research is centered around a tangible application use case, specifically a truck assembly production line, utilizing industrial hardware provided by our associated partner, SmartFactoryKL. This partnership facilitates the creation of a demonstrator and enables the experimental assessment of the proposed approach. Moreover, the data underpinning the IDDSS will be curated and disseminated in an accessible format, thereby benefiting both members of the SPP and the wider research community.
DFG Programme
Priority Programmes
Subproject of
SPP 2443:
Hybrid decision support in product creation
