Scientific Machine Learning Enabled Next Generation Design: application to bond lines in soft materi

Abstract

Approved for Public Release.We propose to employ machine learning within a revolutionary framework of model discovery that combines,clear box and black box modeling aspects in a way that uncovers new engineering theories from mechanistic insight into complex syste,ms. The new mathematical models, that the proposed method finds, will appeal to an engineers intuition and be amenable to the appli,cation of engineering judgment. Since these models will be expressed in the form of well understood mathematical constructions, they, can be employed as reduced order models, suitable for technological and design purposes. It is pointed out that one important goal,of the proposed work is to apply the later outlined methods to gain mechanistic insight and understanding regarding the complex phen,omenology observed within soft material (e.g. polymer) bond lines. We do this by learning partial differential equations (PDE), as w,ell as associated solution operators (e.g. Greens functions.) The structure of the PDE and Greens functions discovered with the pr,oposed method will offer deep and important clues regarding the mechanics of adhesive bond lines. Such mechanistic insight will be o,f great value and utility to domain experts when they design experiments and make performance predictions.

Document Details

Document Type
DoD Grant Award
Publication Date
Dec 10, 2021
Source ID
N000142212055

Entities

People

  • Christopher Earls

Organizations

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

Tags

Readers

  • Calculus or Mathematical Analysis
  • Neural Network Machine Learning.
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks