Research On The Methods Of Grey Decision-Making And Its Application

Grey decision-making is an important part of decision theory, the grey decision-making methods are studied . Main results are as follows:(1) The initialize transform operator of decision-making is studied. At first, on the basis of current literatures, the nonlinear transformation " Rewarding Good and Punishing Bad " operators under the condition of the real numbers, interval gray number and the three parameters are studied, the operators can expand the range of index to negative while the use of nonlinear function to further strengthen the distinction between the degree of indicators to improve the quality of decision-making. Then the weakening operators under the condition of the real numbers, interval gray number and the three parameters are studied, the operators can weaken the gap with the indicator data on the effect of the negative impact of decision-making, and improve decision accuracy. When building a decision model, one of the operators can be chosen according to the situation.(2)The gray target decision making is studied. First it is defined that the ideal optimization scheme is the positive clout and the ideal inferior scheme is the positive clout of the grey target and the optimization and the positive off-target distances are comprehensively considered, and base on the spatial analysis, the comprehensive off-target distance is defined. The distance between two three-parameter interval grey numbers is defined. A new way to compare two three-parameter interval grey numbers is proposed and a decision model of grey target with three-parameter interval grey number is established. Then the grey target decision-making model for ABC classification of inventory items is studied, an optimization model is established according to the data of typical inventory items to establish the classification criterias, according to which the other inventory items are classified.(3) The gray group decision-making methods are studied. First determining weights method is studied. The entropy is used to determining the weights in the gray risk decision. Then the projection formula of interval numbers based on that of the the precision numbers is constructed, and the distance of the vector of the interval numbers is defined, and the weights of the decision-makers are defined according to the total distance between the decision-maker and all the others. Considering when the group of decision makers is large, getting the decisiong conclusion after clustering the decision makers can improve precision, the grey incidence analysis of the decision information is proposed, and a grey cluster model based on the grey incidence degree is proposed. The highest and the lowest decision matrix is defined, and the grey correlation degrees between the decision matrix of each decision-maker and the highest and the lowest decision matrix are abtained, through that the optimal membership degree of each decision-maker is gotten. At last, using the normal model the decision-makers are distinguished to three classes, and the weights of the decision-makers are obtained and the decision result of each decision-maker class is gotten, and the final result is gotten through that.(4) The grey group decision-making model which can discriminate and reject the invalid decision value is studied. First the weights of the index are gotten through a optimization model, and the highest assessment of all the decision schemes is obtained, and the invalid decision value is discriminated through the distance with the highest assessment, instead of the method which reject the highest and lowest decision value simply and which is used widely, and the new weights of the decision-makers of the new decision-maker group is gotten, and the result through the grey target method is obtained. Then structured interview is studied, a model is proposed to discriminate the invalid decision value considering the decision maker's preference information on indexes, and unbiases the valid decision value, and the grey incidence model is used to evaluation.