Preliminary Study on Composite Lifecycle Management Using Digital Twin Technology

Jinglei Yang *

The Hong Kong University of Science and Technology, Hong Kong SAR, China

ABSTRACT: Digital Twin is a novel technology, which creates a virtual digital system for mapping the real physical system. This cyber-physical system communicates with the real physical world with the aid of advanced modeling techniques, big data and intelligent decision tools to satisfy the need for complex physical system design or optimization. It is essentially a diversified combination of multi-physics and multi-scale simulation models with artificial intelligence technology, covering the entire lifecycle of products from R&D, manufacturing to application till failure, providing real-time analysis and guidance. Since the digital twin technology was first introduced in NASA Technology Roadmap in 2010, it has been used in the manufacturing process of F35. With further advancement by the US Air Force Laboratory and GE, this technology has become a competitive asset and key strategy for the US to revitalize its manufacturing industry.

Digital manufacturing is a well-known technology trend, and the concept of "Digital Thread" used here is an essential part of the digital twin technology in the field of composite materials. Generally, the application of digital twins includes three layers: physical layer, data layer, and analytics and interaction layer. Taking composites as an example, the physical layer includes the full lifecycle behavior of the components from forming the composite structures under design principles and specific processing conditions, to serving in actual environment until functional failures. The data layer integrates the real-time data collected by the sensors in physical entity and historical related data. Besides, the virtual data obtained through theoretical and numerical analysis in the virtual twin system is also transmitted to the data layer. The analytics and interaction layer utilizes artificial intelligence as the primary method to understand the combined real and virtual data, following the first principles and basic physical laws applied to the target system as constraint input conditions, provides decision-making optimization and feedback to the physical entity system in real time. Key techniques that need to be focused: a fully digital twin virtual system with sufficient accuracy and efficiency; structured sensing technology to obtain the data from the physical system (e.g., performance, health situation and defect, etc.) and environments; a trust standard database that can be updated in real time; Artificial intelligence modeling for real-time pattern recognition and decision optimization for composites.

Our preliminary work includes: 1. Interpretable machine learning assisted digital feature extraction and performance characterization of microencapsulation process in self-healing composites; 2. Design method of composite structures based on machine learning. With the support of trust datasets, artificial intelligence methods can provide efficient and effective solutions to some key issues in digital twin technology, and eventually realize the full lifecycle management and behavior prediction of digital twin technology on composites, enabling our country’s leading position on composite genetic engineering, industrial IoT, intelligent manufacturing, artificial intelligence algorithms, etc., creating an application model for the composite industry 4.0, and instantly radiating to other industries.

Keywords: Composites; Digital Twin; Trust Database; Artificial Intelligence; Machine learning; Product lifecycle management; Smart Manufacturing

* Corresponding author: maeyang@ust.hk