In-operando tracking and prediction of transition in material system using LSTM

Pranjal Sahu, Dantong Yu, Kevin Yager, Mallesham Dasari, Hong Qin

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations

Abstract

The structures of many material systems evolve as they are treated with physical processing. For instance, organic and inorganic crystalline materials frequently coarsen over time as they are thermally treated; with domains (grains) rotating and growing in size. When a material system undergoing the structural transformation is probed using x-ray scattering beams, the peaks in the scattering images will sharpen and intensify, and the scattering rings will become increasingly’textured’. Accurate identification of the transition frame in advance brings multiple benefits to the NSLS-II in-operando experiments of studying material systems such as minimal beamline damage to samples, reduced energy costs, and the optimal sampling of material properties. In this paper, we formulate the prediction and identification of the structural transition event as a classification problem and apply a novel LSTM model to identify sequences having transition event. The preliminary results of the experiments are encouraging and confirm the viability of the detection and prediction of transition in advance. Our ultimate goal is to deploy such a prediction system in the real-world environment at the selected beamline of NSLS-II for improving the efficiency of the experimental facility.

Original languageEnglish (US)
Title of host publicationProceedings of the 1st International Workshop on Autonomous Infrastructure for Science, AI-Science 2018 - In conjunction with HPDC
PublisherAssociation for Computing Machinery, Inc
ISBN (Electronic)9781450358620
DOIs
StatePublished - Jun 11 2018
Event1st International Workshop on Autonomous Infrastructure for Science, AI-Science 2018 - Tempe, United States
Duration: Jun 11 2018 → …

Publication series

NameProceedings of the 1st International Workshop on Autonomous Infrastructure for Science, AI-Science 2018 - In conjunction with HPDC

Other

Other1st International Workshop on Autonomous Infrastructure for Science, AI-Science 2018
CountryUnited States
CityTempe
Period6/11/18 → …

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Science Applications
  • Software

Keywords

  • Autonomous Infrastructure
  • Deep Learning
  • Future frame prediction
  • LSTM
  • X ray scattering

Fingerprint Dive into the research topics of 'In-operando tracking and prediction of transition in material system using LSTM'. Together they form a unique fingerprint.

Cite this