Knowledge Discovery in Automotive Catalysts Using Microstructural Characterizations

Abstract

Automotive catalysts are regularly examined using several methodologies such as X-ray fluorescence (XRF), X-ray powder diffraction (XRD), wavelength dispersive spectroscopy (WDS), and analytical electron microscopy (AEM) since no single analytical technique is capable of fully characterizing and measuring changes in morphology and chemistry that occur in emission system materials with use. Each analytical technique provides a specific piece of information, such as elemental weight percentages, phase identification, or spatial distribution of elements. However, in aggregate too much information from too many different specimens is often present, making assimilation and processing difficult. A computerized method using data mining techniques could greatly improve the quality of the interpretation and allow discovery of trends that are not readily apparent. However, finding an appropriate method of combining the disparate types of information is a challenge

The first step was to collect and archive the data in a relational database. This task comprised a large percentage of the development process since the data are initially created in several locations and must be collected, cleaned, and stored.