Adaptive modeling of nonlinear dynamic systems by integration of physics and data

Abstract

This proposal concerns the development of a robust family of techniques for accurate mod-eling of practical engineering systems. The state of the art modeling approaches leave muchto be desired as they are not accurate enough for modern systems such as those used bythe Department of Defense. Moreover, as systems change with time, the models do not,and hence have limited usability for behavior prediction, performance determination andcontrol.Practical systems exhibit behavior which can only be predicted using nonlinear models, andhence, this proposal will focus on nonlinear models, primarily, those expressed by ordinarydi -erential equations. Physics-based models will be used to develop generic descriptionsand sensor data will be integrated to create hybrid models that are more accurate thaneither of the two approaches. In general, models can be inaccurate to begin with. Inaddition, they become usually more inaccurate because (a) the systems often change, (b)the operating conditions change, and (c) environmental inuences - including interactionwith other connected systems - change. Hence, the resulting hybrid models that we createneed to change in a way as to be always accurate in their predictions.We will develop and explore statistical and machine learning methods such as recurrent neu-ral networks, Hop?eld networks and Gaussian mixture models. We will hone the algorithmsfor a variety of mechanical, electrical and electromechanical systems. Candidate systemsinclude: nonlinear oscillators, machinery systems, electrohydraulic servovalve actuators,pumps, and motors.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 07, 2019
Source ID
N000141912070

Entities

People

  • C. Nataraj

Organizations

  • Office of Naval Research
  • United States Navy
  • Villanova University

Tags

Readers

  • Computational Modeling and Simulation
  • Distributed Systems and Data Platform Development
  • Military Logistics and Supply Chain Management

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • Microelectronics
  • Microelectronics - Microelectromechanical Systems