| About this Abstract |
| Meeting |
Materials Science & Technology 2020
|
| Symposium
|
Micro- and Nano-Mechanical Behavior of Materials
|
| Presentation Title |
Quantitative Evaluation of Large Nanoindentation Data Sets |
| Author(s) |
Bernard Becker, Benjamin Stadnick, Eric Hintsala, Ude Hangen, Douglas D. Stauffer |
| On-Site Speaker (Planned) |
Douglas D. Stauffer |
| Abstract Scope |
Nanoindentation methods have progressed to the point where extremely large data sets, up to 106, are now practical. This then requires automated and statistical analysis, including clustering and the use of machine learning. This talk focuses on how to extract materials information from large arrays of nanoindentation data. Determining the necessary number of data points, how many clusters should be used, what cluster models, and the related questions related to data bootstrapping are discussed. A framework is presented that can be used to describe nanoindentation data with a vector, modeling that data, then simulating data of the model, and re-clustering of the simulated data to provide information related to the original clustering. The resulting properties are found to have a low bias (<1%) and relative uncertainty (<1%), indicating sufficiency in comparison to that of the original datasets. |