The objective of this research project is to create a framework, and to advance the associated theory, for design of materials at the microstructural level. The proposed work is significant because it will enable materials scientists to derive characterizations of microstructure that go substantially beyond statistical measures. Furthermore, by incorporating computational learning theory into a decision-theoretic setting, knowledge from the materials science and micromechanics viewpoints can be effectively incorporated into the materials design process. The theoretical foundation of the design method is general, and may well find application to design problems not directly involving microstructure. Bayesian classifiers, developed primarily for image analysis problems, will be used to evaluate the micromechanical response of candidate microstructures without time consuming, high fidelity mechanics analyses. By recasting the classification procedure into the framework of decision theory we transform the technique from one that is purely evaluative to one which can be used as part of the design optimization scheme critical to improved microstructural design. The proposed research will strengthen the link between materials science and component design, leading to higher performing materials using less materials resources, and safer and more durable designs with direct benefits for the consumer.
The proposed work will provide training for engineering graduate students in novel strategies for design and analysis of materials. Through the Ingenuity Project of the Baltimore City Schools the project will also provide experience in engineering research to under-represented high school students.
