My work in applications provides insight into scientific data and addresses specific questions posed by domain scientists. Many factors, occurring individually or in combination, make this endeavor challenging, including the size of the data set, its quality, its complexity, and how clearly the questions being addressed have been articulated by the domain scientists. Solving a real problem is far from trivial and often involves modifying and combining ideas and algorithms from different fields.
More details on my work in applications is available on the following pages:
- Identifying parameters for building high-density, additively-manufactured parts
- Data mining techniques for use in additive manufacturing
- WindSENSE – Integrating wind energy on the power grid
- Identifying human settlements in remotely-sensed data
- SBOR – Similarity-Based Object Retrieval
- Validation of computer simulations
- Identification of bent-double galaxies
- Analysis of coherent structures
- Classification of orbits in a Poincare plot
Identifying parameters for building high-density, additively-manufactured parts. It is a challenge to determine process parameters to use in building high-density metal parts with laser powder-bed fusion, especially for materials that have never been additively manufactured before. We show how simple simulations and experiments, combined with data analysis, can help guide the selection of optimal parameters.
Select publications (available from Google Scholar):
- Chandrika Kamath, Bassem El-dasher, Gilbert F. Gallegos, Wayne E. King, and Aaron Sisto, “Density of additively-manufactured, 316L SS parts using laser powder-bed fusion at powers up to 400 W,”. Int J Adv Manuf Technol. Volume 74, Issue 1 (2014), Page 65-78.
- C. Kamath, “On the use of data mining to build high-density, additively-manufactured parts,” invited book chapter, Information Science for Materials Discovery and Design, T. Lookman, F. Alexander, and K. Rajan (eds.) in Springer Series in Materials Science, 225, pp 141-155, 2016.
- C. Kamath, “Determination of Process Parameters for High-Density, Ti-6Al-4V Parts Using Additive Manufacturing,” Lawrence Livermore National Laboratory Technical report, LLNL-TR-736642, August 2017. [PDF] Available as https://e-reports-ext.llnl.gov/pdf/889406.pdf
Data mining techniques for use in additive manufacturing. Laser powder bed fusion is a rich source of problems that can be addressed using data analysis techniques. These include i) feature selection and parallel plots to identify important process parameters; ii) sampling and surrogates for exploring the input design space and for solving inverse problems; iii) self-organizing maps to understand the solution to the inverse problem; and iv) Gaussian process regression for prediction with associated uncertainties. By incorporating these techniques into an efficient iterative approach that starts with simple experiments and simulations to understand the design space, and moves to progressively more complex experiments and simulations as we get closer to a solution, we can reduce the time it takes to address problems of interest.
Select publications (available from Google Scholar):
- C.Kamath, “Data Mining and Statistical Inference in Selective Laser Melting”, International Journal of Advanced Manufacturing Technology, Volume 86, Issue 5, pp 1659–1677, September 2016. Appeared online 11 January, 2016. http://dx.doi.org/10.1007/s00170-015-8289-2
- C. Kamath and Y.J. Fan, “Regression with Small Data Sets: A Case Study using Code Surrogates in Additive Manufacturing,” Knowledge and Information Systems Journal, Volume 57, Number 2, November 2018, pp. 475-493.
- Chandrika Kamath, Juliette Franzman, and Ravi Ponmalai, “Data mining for faster, interpretable solutions to inverse problems: A case study using additive manufacturing,” Machine Learning with Applications, Volume 6, 15 December 2021, https://doi.org/10.1016/j.mlwa.2021.100122.