Toward automated interpretation of integrated information: Managing "big data" for NDE
AIP Publishing LLC
Is Version Of
Aerospace EngineeringCenter for Nondestructive Evaluation
Large scale automation of NDE processes is rapidly maturing, thanks to recent improvements in robotics and the rapid growth of computer power over the last twenty years. It is fairly straightforward to automate NDE data collection itself, but the process of NDE remains largely manual. We will discuss three threads of technological needs that must be addressed before we are able to perform automated NDE. Spatial context, the first thread, means that each NDE measurement taken is accompanied by metadata that locates the measurement with respect to the 3D physical geometry of the specimen. In this way, the geometry of the specimen acts as a database key. Data context, the second thread, means that we record why the data was taken and how it was measured in addition to the NDE data itself. We will present our software tool that helps users interact with data in context, Databrowse. Condition estimation, the third thread, is maintaining the best possible knowledge of the condition (serviceability, degradation, etc.) of an object or part. In the NDE context, we can prospectively use Bayes' Theorem to integrate the data from each new NDE measurement with prior knowledge. These tools, combined with robotic measurements and automated defect analysis, will provide the information needed to make high-level life predictions and focus NDE measurements where they are needed most.
This proceeding may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This proceeding appeared in Gregory, Elizabeth, Tyler Lesthaeghe, and Stephen Holland. "Toward automated interpretation of integrated information: Managing" big data" for NDE." In AIP conference proceedings, vol. 1650, no. 1, pp. 1893-1897. American Institute of Physics, 2015, and may be found at DOI: 10.1063/1.4914815. Copyright 2015 AIP Publishing LLC. Posted with permission.