Research On Design Scheme Generation Method Driven By Product Design History Data

With the advancement of technology and the improvement of living standards,the customer requirements(CRs)for products has become more personalized,diversified and complicated,which puts higher demands on the design of products.However,the product design is an original or improved development process that integrates various elements,such as requirements,intentions,functions,structures,levels,rules,etc.Therefore,it is necessary to develop a fast generation method of product scheme to achieve the quick response of product to market and changing requirements.Aiming at this problem,this dissertation proposes a product design scheme generation method based on multi-level decomposition of CRs and the data mining of product history design case,which realizes product rapid design based on CRs.The main contents are as follows:(1)Research on hierarchical mapping of CRs based on FAHP and improved quality function deployment(QFD).The weight of CRs was being determined by FAHP methods.A hierarchical functional mapping model based on FAHP and improved QFD CRs is established to transform performance characteristic indexs(PCIs)and functional characteristic indexs(FCIs)into corresponding functional characteristics.Then,the PCIs are divided into three levels according to the index weight value,and the function analysis system technique model is used to decompose the FCIs.(2)Research on the historical case data mining of task-driven patent texts.Firstly,the expression model of patent text knowledge is established.The text mining method is used to perform part-of-speech tagging and syntactic analysis on patent texts,to extract functional text and structure text data.On this basis,the design case sample data based on functional structure data is established,which is the foundation of the design model in later content.(3)Research on neural network model construction and design scheme generation based on case data.A multi-layer feedforward neural network is selected as the training model.The sample data is used to train the model.and the parameters are adjust to reach the optimum.At the same time,the custom matching data is determined by using the similar matching rule.The data is input into the trained neural network product design model.and the model automatically outputs the product design plan that meets the requirements which realize the rapid design of the product.