Feature Based Associative Memory for Design Retrieval System

点击翻译按钮获取中文摘要

This grant provides funding for research on a feature-based associative memory for engineering design retrieval. Engineering designs are typically stored as computer-aided design (CAD) files. A database of such files may contain many thousands of designs. Searching the database for similar designs instead of identical designs poses a challenging problem. This research will seek a design description that represents the geometric shape of a design by a set of machining form features such as counter-hole, pocket or slot and by a set of crucial technological attributes that affect machining processes such as dimension, tolerance, surface roughness and finish and material. Searches may use a fuzzy logic techniques to retrieve similar designs. The research will include development and experimentation with algorithms to find similar relationships among individual form features, methods to embody technological attributes, approaches to incorporate the ranking weights for design attributes, possibilities to represent spatial relationship as a retrieval criterion, and processes to integrate fuzzy linguistic expressions and interval values in query inputs.If successful, this research will contribute to the conceptual design stage of industrial components. By enabling designers to locate similar CAD files, they can use these to "cut and paste" new designs together from existing designs, greatly reducing the effort needed to create CAD files for new components.