Decision-making Method With Tacit Objective Based On IEC And RBF

In the real decision-making, there exit such decision-making problem in the reality management domain that is important but hard to solve: decision-making problem with tacit objective function (DMPTOF).Its decision making objective functions can not or is difficult to be defined explicitly. And how to define the preference of decision maker in advance in the problem is also difficult. It is suitable for human to affirm his preference progressively during the decision making process. Interactive genetic algorithms combining traditional evolutionary mechanism with human's subjective evaluation can solve them effectively. But human's fatigue is a key problem of interactive genetic algorithms which restricts their abroad applications in complicated decision-making problems. In order to solve the above problem, this dissertation researched the theories and key technologies of interactive genetic algorithms and solving case of such decision-making problem.From the perspective of the decision-making problem with tacit objective function, this dissertation mainly contents include:(1) Firstly, decision-making problem with tacit objective function (DMPTOF) studied by this dissertation is introduced. The characters and the investigative difficulties of the problem is analyzed. The mathematic form of DMPTOF is given. The requests to solve the problem are discussed, IEC and RBF is the fit technology to deal with DMPTOF. The solving process framework based on IEC for DMPTOF is set up.(2) Interactive genetic algorithm (IGA) that is a branch of IEC evolves from genetic algorithm (GA), the concept of the IEC is introduced and the core issues are analyzed; Introduced RBF network function (RBF) Structure, analyzed Network of Radial Basis Function Approximation ability, learning and generalization ability, studied Radial Basis Functions learning algorithm.(3) A proposed interactive evolutionary computation (IEC) with RBF network (RBF) to solve the problem of DMPTOF algorithm, the algorithm based on a small number of users on the evolution of the individual evaluation, using extraction RBF Operators the evaluation of the characteristics of individual preference to people on individual machines fitness evaluation. The algorithm overcomes interactive genetic algorithm low global search capability, users easily fatigue shortcomings.(4) A material case of DMPTOF, which is fashion design problem, is studied. The coding and solving ideal of fashion design problem is analyzed and discussed. The implement flow of fashion design system based on 1EC and RBF is studied. The functions of the developed system are discussed.