Current Research

Online process monitoring and control of nanoparticle selfassembly processes

A self-assembly is considered as a promising method for producing nano particles in large-scale. The self-assembly refers to the process by which small-scale building blocks such as atoms and molecules are spontaneously arranged into order structures at nanoscale by thermodynamics and other underlying physics, which is highly dynamic and random. Due to the random nature of the self-assembly, nano particles produced from the self-assembly do not have the same size and shape but prone to have wide size and shape distributions. Producing nano particles with narrow size and shape distributions is a long-term desire of nano scientists.

For production of precisely shaped and sized particles, online process characterization is pressingly needed to monitor particle sizes and shapes in realtime of production. This is because poorly controllable self-assembly processes are still not well understood, and direct online measurements of the processes currently appear to be the only way of broadening the understanding and thus improving controllability. The online process characterization use the direct online measurement of intermediate products sampled from a process (1) to monitor the dynamic changes of particle sizes and shapes online, (2) to use the dynamic changes for deriving realtime dynamics of the self-assembly process, and (3) ultimately to control the dynamics for precisely controlling particle sizes and shapes. However, the direct online measurement is not available for all necessary time and length scales, due to physical limitation of online instrumentation capability. So, before the next innovations in online instrumentation tackle the physical limitation, the possible workaround is a data-driven approach that fully utilizes limitedly available online measurements to estimate the dynamic changes of particle sizes and shapes with some (ideally, small) degree of uncertainties. Developing the data-driven approach is the main objective of the proposed research.
  • Relevant publications:
    • Park, C., 2012, "Estimating multiple pathways of object growth using non-longitudinal image data," Submitted.
    • Park, C. and Shrivastava, A., 2012, "Multimode geometric profile monitoring with correlated image data and its application to nanoparticle self-assembly processes," Submitted.

Analysis of smartphone usage data

I am interested in analyzing a big data from smartphones. The data reveals movement patterns, phone usage patterns, application usage patterns and other information of smartphone users. When the data from multiple individuals are collectively analyzed, the analysis provides important information for social science research such as marketing. I am collaborating with a faculty in the Department of Geography at FSU for this project.

Previous Research

Nanoscale Measurement and Control

  • Motivation:
    • High correlation of nanoparticle's properties with its size and shape
    • Importance on observing new properties of nanomaterials having different scales and structures
    • Repeatable realization of the desired properties
  • Measurement equipments: High Resolution Transmission Electron Microscope, Scanning Electron Microscope, Scanning Tunneling Microscopy, Atomic Force Microscopy, X-ray scattering
  • Research Interests:
    • Efficient measurement of the morphology of nanoparticles synthesized by wet chemical processes
    • Production of nanoparticles having the uniform shapes and sizes by controlling their synthesis processes
  • Relevant publications:
    • Huitink, D., Kundu, S., Park, C., Mallick, B., Huang, J. Z. and Liang, H., 2010, "Nanoparticle Shape Evolution Identified through Multivariate Statistics," Journal of Physical Chemistry A. 114(17), 5596-5600.
    • Park, C., Huang, J. Z., Huitink, D., Kundu, S., Mallick, B., Liang, H. and Ding, Y., 2011, "A Multi-stage, Semi-automated procedure for analyzing the morphology of nanoparticles," IIE Transactions special issue on Nanomanufacturing. Accepted
    • Park, C., Huang, J. Z., Ji, J. and Ding, Y., 2011, "Segmenting, inference and classification of partially overlapping nanoparticles," Submitted.
    • Park, C., 2012, "Estimating Nanocrystal Growth Trajectories From Non-longitudinal Image Data via Nonparametric Bayesian Radial Growth Model," Submitted.

System Diagnosis with a Large Scale Sensor Network

  • Motivation:
    • Old infrastructure: 25% of nation's 601,411 bridges are either as structurally deficient or functionally obsolete, 2 million miles of natural gas lines
    • Hardly reachable systems: Off-shore wind turbine
  • Emerging Technology: Wireless sensor network (WSN) emerges as a key technology for efficient system diagnosis and maintenance.
  • Research Interests:
    • Reduction in power consumption of wireless sensors to avoid frequent battery replacements
    • Digest of a large amount of data from many sensors
    • Robust system diagnosis with some corruptions in sensor measurements
    • Detection of anomalies in system never happened before
  • Relevant publications:
    • Park, C., Tang, J., and Ding, Y., 2010, "Aggressive data reduction for damage detection in structural health monitoring," Structural Health Monitoring. [paper][ppt]
    • Park, C., Huang, J. Z., and Ding, Y., 2010, "A computable plug-in estimator of minimum volume sets for novelty detection," Operations Research.
    • Park, C., Ding, Y., and Byon, E., 2008, "Collaborative data reduction for energy efficient Sensor Networks," Proceedings of the IEEE Conference on Automation Science and Engineering, Washington, D.C. [paper][ppt] (Best Student Paper)
    • Ding, Y., Byon, E., Park, C., Tang, J., Lu, Y. and Wang, X., 2007, "Dynamic Data-Driven Fault Diagnosis of Wind Turbine Systems," Lecture Note in Computer Science. [paper]

Fast Computation of Gaussian Process Regression with Large scale Datasets

  • Motivation:
    • Prediction on unobservations locations using many noisy observations is one of inevitable steps for several application: environmental statistics and the quality test on the surface of manufacturing products.
    • Gaussian Process Regression, so called kriging, is powerful and flexible for such prediction, but it is computationally expensive.
  • Research Interests:
    • Fast approximation methods of Gaussian Process Regression
    • Domain Decomposition for faster approximation
    • Linkage model to bond heteogeneous random fields
  • Relevant publications:
    • Park, C., Huang, J. Z. and Ding, Y., 2011, "Domain Decomposition for Fast Gaussian Process Regression," Journal of Machine Learning Research. vol 12. pages 1697-1728.
    • Park, C., Huang, J. Z. and Ding, Y., 2012, "GPLP: A local and parallel computation tool box for Gaussian process regression," Journal of Machine Learning Research. vol 13. pages 775-779.