Training

AI and Machine Learning for Simulation and Digital Twins

In order to capture the benefits of digital engineering, digital transformation initiatives have become important to companies and organizations in all industries. Digital transformation calls for greater use of simulation and the integration of simulation models in creating digital twins, unlocking benefits such as improved decision-making, enhanced design confidence, and more efficient engineering processes.

To achieve the aims of digital transformation modern AI and machine learning (ML) tools are needed to handle challenges such as:

• Long simulation run times which can render statistical or iterative tasks like optimization, sensitivity analysis, and uncertainty quantification infeasible.
• Difficulty extracting relevant information and quickly gaining insights from large, correlated data sets, for example from sensor readings.
• Inaccuracy of models due to the presence of uncertainties arising from sources including initial conditions, model parameters, model form, manufacturing, and operating environment.
• Difficulty in obtaining fast, accurate digital twins.

To address these challenges and more, this course provides an introduction to important topics including:

• How to optimally collect or select ML model training data using design of experiments (DOEs) and data sampling.
• How to use ML to train surrogate models for accelerating simulation analyses.
• Use of surrogate models for sensitivity analysis, uncertainty propagation, and optimization under uncertainty.
• Statistical calibration for achieving better agreement between models and physical data and as a tool for creating digital twins.

Points will be illustrated with SmartUQ customer use cases and example problems demonstrated in SmartUQ. Use cases and examples can be tailored to be relevant to the needs and industry of the attendees.