A Knowledge Discovery Method of Product Design Requirements Based on Pattern Matching

Abstract

In conventional product design process, the designers firstly formulate a general design assignment according to the users’ demands, design experience, product running environment and so on, then professional designers from different departments carry out detailed module designs according to those design requirements in the assignment. This kind of design method is suitable for small-batch and single-variety production. However, with the emergence of mass customization production, reading lengthy design assignments has gradually become a difficulty that restricts the production efficiency for enterprises, as a result of which extracting the requirement knowledge in the design assignment efficiently and well-targeted is becoming more and more important. In this paper an expert system (ES) framework for product design requirements knowledge discovery is proposed Firstly, a knowledge base is established by analyzing the types of requirements. Secondly, the requirements extraction program of the reasoning program is proposed based on the information extraction method of regular expression and named entity library, combined with manual-assisted positioning. Thirdly, the requirement similarity evaluating method of reasoning program is established combined with the characteristic of demand attributes. Finally, the design assignment books of screw conveyor are taken as an example to evaluate the proposed method and it turns out 90% of coverage and precision rate within 50 design assignment books.