【<strong>4.27</strong>】Prof. Mirna Urquidi-Macdonald<br/>Topic：Data Mining and ...----中国科学院金属研究所

Data Mining and an example on Topological Ordered Maps of General Corrosion of Most Commonly UsedAlloys

Abstract

When problems are difficult, hypothesis on the phenomenon are tested by the means of collected data. Data mining comprises techniques used to extract information from data and to test the hypothesis advanced on the phenomenon. This talk will introduce a number of data mining techniques and will give an example on the design of new Alloys using one of these techniques. Metal alloys are important to nearly every engineering industry today. This study was developed to provide the optimum method for designing metal alloys. In order to do so, information on general corrosion of the most commonly used alloys, namely nickel-, iron-, aluminum-, stainless steel-, copper- and carbon-based alloys, was obtained from the National Institute of Standards and Technology (NIST). The pH, conductivity, and major ions in the electrolytes used in each experiment for which data were collected (alloy in contact with an environment); each alloy composition, as given by its UNS number; and the corrosion rates of each alloy/ electrolyte combination were collected in a single data vector. 　To efficiently cluster the systems by similarity, the data vectors were constructed and a web-based, publicly available, Kohonen mapping software was used to perform the clustering analysis. Kohonen maps have the ability to preserve the topological properties of the data; similar vectors cluster together while different vectors get separated on the map. The manner in which these vectors group allows for the analysis of the detailed effect of the minor alloy elements, as well as the composition and concentration of the environment, on the corrosion rates of each alloy. We draw each one of the parameters on the vectors as a function of the (X,Y) map coordinates.　“If-Then” scenarios are explored by drawing 3-dimentional maps indicating in a two of the dimensions the position on the topological map and in the third dimension the parameter intensity explored.

CV-Prof. Mirna Urquidi-Macdonald

Over the past twenty-five years, Dr. Urquidi-Macdonald’s research interest have revolved around energy-related and systems reliability issues.　In 1983, Dr. Urquidi-Macdonald moved to Ohio State University as a Research Associate in the Fontana Corrosion Center, Department of Metallurgical Engineering. She joined Stanford Research Institute (SRI) International in 1984, where she assumed the position of Senior Research Scientist. In 1991, Dr. Urquidi-Macdonald moved to Penn State University, where she assumed the position of Associate Professor in the Department of Engineering Science and Mechanics, College of Engineering. She was awarded tenure in June 1997. In 1994, she was appointed by Secretary Federico Peña to serve for six years as an advisor to the Department of Transportation's Technical Pipeline Safety Standards Committee. She served as a consultant in a program on Activity Transport in Water Cooled Reactors, at the International Atomic Energy Agency, located in Vienna, Austria. She is an Adjunct Professor at Massey University, in New Zealand, and an Adjunct Professor at KFUPM in Saudi Arabia. She has been recognized as a Fellow by NACE International (National Association of Corrosion Engineers) and by ASM International (American Society of Materials). Her research interest at Penn State have revolved around the development of advanced batteries and fuel cells, detection and assessment of corrosion, and on the application of ANNs to model physico-chemical systems, which are, by their very nature, complicated. She is a collaborator with INSA-Lyon, Ecole Central at Lyon, France and with the Biocorys Program in Europe where, this summer, she will be presenting courses and lectures with colleagues in Leuven, Belgium; Milan, Italy; Toulouse, France; and Plymouth, England in assessing the impact of bacterial corrosion in industrial and pipeline systems.