Research On Key Technologies For Innovative Design Based On Patent Mining

Knowledge is the basis of innovation, and patent is a crucial knowledge resource to promote product innovation and enterprise development. TRIZ is patent knowledge mining oriented to technology innovative problem solved, which is one of the most important theories in innovative design field at present. The technology maturity of product forecasting based on TRIZ and principles of solving technical contradictions aided producing innovative schemes, make R & D decision-making of enterprises be more scientific, and product innovation have methodologies and evidences to support. On the basis of TRIZ and amazing at the key technologies of product innovation - technology maturity forecasting and Inventive Principles, this thesis focuses on computer supported innovation methods relying on the patents with text mining, natural language understanding and other advanced technologies, to broaden the designer's capabilities of decisions and innovation. This thesis has the following contributions:A method of product innovative design based on patent knowledge is proposed, by extracting knowledge and rules from patent documents, to increase the predictability of innovative design.With the application of technology maturity forecasting and principles of solving contradictions, patents are analyzed on the basis of text mining technology. This is a kind of new way for product innovation. It not only extends views and methods of patent analysis, making patent analysis and application more simple and widespread, but also promotes the construction of patent knowledge base corresponding to Inventive Principles, which can promote the develop -ment and perfect of TRIZ theory itself.The existing methods of technology maturity forecasting are analyzed. Based on Altshuller's patent review patterns, a method of technology maturity forecasting with text mining for patents is presented in the modern information society. According to the characteristics of patent literature, a suit of new indexes and models on technology forecasting of product maturity is put forward, and a software system is developed to extract the indexes data, which can be processing system before the technology evolution module of Invention Tool 3.0. The technology maturity of American butterfly valve is forecasted by using the method.Patent classification as technology in organizing and processing the huge amount of patent data, the common patent classification standards are described. From the perspective of product innovative design, the limitations of the current patent classification system supported innovative design is pointed out, the patent classification standard according to Inventive Principle is put forward. It can help TRIZ user obtain knowledge from patent literature in various fields, develop design ideas in product conceptual design phase, break the inertia of thinking, and enhance the ability of innovative design. Pretreatment technique of English patent and the method of establishing mathematical model are studied. In patent classification oriented for Inventive Principles, some feature selection methods are introduced and tested, the traditional TF-IDF feature weight calculation method is improved, and the classification methods are studied. On this basis, 10 Inventive Principles are selected, VC++ programming language is used for software code, visualization software for patent classification according to these Inventive Principles is developed as the post-processing innovation support system of principles of solving contradictions model in Invention Tool 3.0. And the paper machine is applied and verified as an example for innovative design with the reference in corresponding patents after principles determined to produce design schemes.