Research Of Quality Function Deployment Based On Artificial Immune System

Quality function deployment (QFD) is a customer-driven design tool that aims to satisfythe desires of customers while maximizing enterprises goals. And QFD is a systematicpractice for product design based on methodology of system engineering. The process of QFDthat usually consists of product planning, part deployment, process planning and productionplanning is a collaborative decision making problem of multiobjective and multidisciplinary.Artificial immune system (AIS) that is a powerful information processing and problemssolving paradigm based on bionic mechanism offers new intelligent supports for QFD.According to the characteristics of QFD and AIS, AIS based QFD is presented by integratingQFD with AIS. So, the proposed AIS based QFD can be used to improve the level ofsystematic design, dynamic response and intelligent decision making for the process of QFD.In the dissertation, the methodology of AIS based QFD is investigated comprehensivelyby using fuzzy set, AIS, linear physical programming (LPP), grey regression theory and leastsquare support vector machine. Because the research of QFD based on AIS covers a broadscope, it is impossible to carry out investigation for each aspect. Thus, the paper will focus onsome aspects such as the framework of AIS based QFD, structure of immune house of quality,determination of importance weights of customer requirements, forecast of importanceweights of customer requirements, decision making of immune house of quality andevaluation of design concept. The main content is as follows:Firstly, by using AIS and multidisciplinary design optimization (MDO), the model ofAIS based QFD is proposed. The model is composed of optimization decision making ofMDO, immune house of quality (IHOQ), immune part deployment, immune process planning,immune production planning, analysis of immune house of quality, analysis of immune partdeployment, analysis of immune process planning and analysis of immune productionplanning. After the modeling method of traditional QFD is enriched and expanded through thetechnologies of fusion of multi-granularity information and2-tuple linguistic representation,the model of immune house of quality is also proposed based on fusion of multi-granularityinformation. Further, the antigens of perceiving customer requirements (CRs), relationship of engineering characteristics (ECs), relationship between CRs and ECs are constructed, thedecision making algorithm is developed based on multiple objective programming for IHOQand the antibodies responding to configuration result of IHOQ are illustrated.Secondly, subjective analysis method of importance weights of customer requirements ispresented by using triangular fuzzy numbers, analytic network process (ANP) anddecision-making trial and evaluation laboratory (DEMATEL). Objective analysis method ofimportance weights of CRs is developed based on methods of entropy and mean square. Themethod for determining the importance weights of CRs is developed based on maximaldeviation approach for competitive priority ratings of CRs. And based on Kano model, theimportance weights of CRs are determined. The combination model is developed fordetermination of importance weights of CRs based on minimal deviation approach.Thirdly, with view to capturing the better parameters of least square support vectormachine (LSSVM), the AIS based method is proposed to optimize the parameters of theestablished LSSVM model. The model of AIS based LSSVM (AISVM) is developed forforecast of importance weights of CRs. And the forecast models of importance weights ofCRs based on grey model of double residual error corrections (GMDREC) and neuralnetwork (NN) are developed. The performance of forecast model of CRs based on AISVMwas investigated by comparisons with the models based on GMDREC and neural network.Fourthly, the methodology of multiple objectives optimization is proposed based on AISand LPP. To satisfy customer, minimize cost and maximize extensibility of product, theoptimization model of IHOQ which integrates mixed integer linear programming model andKano model, alternatives cost model and alternative extensibility model is established. Theoptimization model is calculated by the proposed methodology of multiobjective optimization.The comparisons between the proposed method and Lingo and GA were provided to illustratethe effectiveness and feasibility of the proposed methodology.Fifthly, the method of design concept evaluation is proposed based on intuitionistic fuzzyentropy and technique for order preference by similarity to ideal solution. Servo configurationsystem for numerical printing and die-cutting machinery of corrugated board (SCSNPDMCB)is developed by iSIGHT software and five alternatives of servo configuration are obtained bythe SCSNPDMCB. The five alternatives are ranked by the proposed method after calculation of importance weights of criteria based on maximal deviation model of intuitionistic fuzzyentropy.Finally, by the research fruit of AIS based QFD, the artificial immune based prototypesystem for quality function deployment and evaluation (AI-PSQFDE) is developed toillustrate the correctness and the validity of the proposed method. The AI-PSQFDE consists ofthree subsystems such as the system of immune product planning and evaluation based on VC,the service system of immune product planning and evaluation based on Matlab and the agentof immune product planning and evaluation based on JADE and Ontology.