You are here

Physics-Based Simulation & Machine Learning Fusion for Sensor Network Design, Optimization, and Digital Twin Applications

Paul R. Ohodnicki Jr., University of Pittsburgh

Abhishek Venketeswaran, National Energy Technology Laboratory

Sandeep Reddy Bukka, Max Planck Institute for Dynamics of Complex Technical Systems 

A broad spectrum of industries (aerospace, automotive, biomedical, civil, manufacturing, and oil & gas) are embracing rapid automation and digitalization, commonly referred to as the Industrial Internet of Things (IIoT). A sensor network coupled with machine learning (ML) tools, constitutes the core intelligence of an IIoT system. Sensor data can be harnessed by ML tools to develop smart operational capabilities such as enhancing the interconnectedness of the IIoT components, streamlining supply chain logistics, and enabling predictive maintenance of assets. Recent demand for the development of digital twins of industrial assets has also motivated research into computational methods to integrate real-time sensor data with asset simulation models.

In recent years, scientific machine learning (SciML) and multi-physics simulation tools have been utilized to model sensor response, optimize sensor deployment strategies and develop intelligent sensor operation capabilities. This mini-symposium aims to bring together practitioners of computational mechanics and SciML focused on developing hybrid methods that harness sensor data and multi-physics simulation models for predictive modeling tasks. These include, but are not limited to efforts on:

Multi-Physics simulation for sensor modeling

Optimization methods for sensor design & deployment

Reduced-order modeling & SciML methods for digital twin applications

SciML methods for sensor data analytics, structural health monitoring & prognostics

Sensor-asset interaction modeling

Physics-based machine learning methods for sensor materials design

Deep learning, transfer learning, & uncertainty quantification of sensor response

Multi-sensor fusion & data assimilation