Session: 01-05-01: Applications of Artificial Intelligence/Machine Learning for Aerospace Structures 1

Paper Number: 162291

162291 - Cpp Digital Twin and Digital Shadow for a Damaged Plate Demonstration 

Digital Twin is a concept of high interest to the aerospace community, including the US Air Force and private companies. For our definition of the digital twin, we refer to the AIAA position paper: “A digital twin is a set of virtual information constructs that mimics the structure, context, and behavior of an individual/unique physical asset, is dynamically updated with data from its physical twin throughout its life cycle, and informs decisions that realize value” [1]. This definition distinguishes a digital twin from complex simulations. However, the digital twin has been used with different meanings in literature: detailed simulations with no data exchange (digital models), with one-way data exchange (digital shadows), and with two-way data exchange (true digital twins). Digital twins and digital shadows enable proactive solutions before problems arise by predicting and identifying potential issues early. A digital twin is a tail number-specific computational model of an individual aircraft (or other physical assets). The digital twin will be updated based on sensory data over the life of the physical asset. A digital twin decreases maintenance costs by acting as a live health monitoring system. It also helps engineers and operators make better decisions based on running what-if analysis on the digital twin.

The CPP-DiTTA goal is to create a technology demonstrator for a digital twin for educational purposes. This presentation presents a digital twin and a digital shadow framework for a damaged plate. The digital shadow predicts the location of the damage from a few sensor data. The digital twin provides decision-making information about the maximum applied load possible for the detected damaged plate. We implemented the entire process on a website as a benchmark demonstrator of a digital twin.

The physical asset is a rectangular aluminum plate. However, a detailed finite element model is used instead of the physical asset for ease of use. This approach allows us to host the entire experience on a digital platform. In addition, we have three digital subcomponents: Data Gathering, Machine Learning (ML) models, and Visualization. Data Gathering involves cleaning the data sent by the sensors attached to the detailed digital model (the replacement of the digital platform), ensuring that the data forwarded to the ML model resembles the training data.

The training data are gathered using many ANSYS runs. We trained a neural network to act as the predictive model. Then, a k-nearest model is used to provide feedback to the user.

One of our key goals is to provide an intuitive environment for users to understand the simulation results in a 3D virtual environment. We use the 3D game engine Unity. Using this game engine, we create a 3D environment containing the physical asset as a 3D model that users can inspect from any side and angle and interact with intuitively.

The presentation includes a brief introduction of each component of the digital twin and a demonstration of the DiTTA in action through our DiTTA website.

Acknowledgment: This work is partially sponsored by AFRL under agreement number FA8650-24-2-2403. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes, notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of AFRL or the U.S. Government.

Presenting Author: Zahra Sotoudeh California State Polytechnic University, Pomona

Presenting Author Biography: Zahra Sotoudeh is a professor of aerospace engineering.