Research On Similarity-based Reasoning By Extenics Of Function-tree In Conceptual Design

Innovative reasoning of product is the key to succeed in market competition. The innovation of product mainly depends on the phase of conceptual design in product design. Function model is the core object of conceptual design. Consequently how to process innovative reasoning of function model efficiently is the key problem of conceptual design. Presently, varieties of function models are used in conceptual design. Function-tree is a typical function model and widely used. Similarity extension is a key step in innovative reasoning of function tree. The modeling of function-tree is mostly based on the professional knowledge of domain experts, while it's hard to consider all aspects of problem by human brain. So the information in function-tree modeled by experts is always incomplete. In addition, the existing method to calculate similarity degree of function-trees has some shortages, and is unable to calculate similarity degree with any form of and-or-not function-trees. As a result, similarity extension method of and-or-not function-tree is researched in this dissertation. The main contents go as follow:(1) Definition and solutions of disjunctive normal form treeDisjunctive normal form tree is defined based on Boolean algebra. And two algorithms are presented to get disjunctive normal form tree from any and-or-not function-tree. Then all structures of function-trees can be normalized. And it makes the foundation for similarity degree calculation between any function-trees.(2) The method to calculate similarity degree of matter-elementsAs the existence of contrast association, the characteristic value of function may be negative. There are defects of the formula to calculate similarity degree in similarity theory. A new formula to calculate similarity degree of characteristics is proposed. It overcomes the defects of the original one, and is able to calculate with interval-values. So the universality of new formula is improved. Based on the new formula, the methods to calculate ordinary similarity degree and contrast similarity degree of matter-elements are given.(3) Similarity-based Extension of function-treeOrdinary similarity functions and contrast similarity functions of selected function-tree's node are searched in knowledge base. Then extension of function-tree is executed by adding search results into function-tree. So function-tree may have some new functions as before. Consequently designers can get more information related to design, and develop design ideas. And the possibility to get more innovative design solutions is enhanced.