Study On The Target-oriented Preference-strength Commensurability And Relevant Decision Models With Variable Weights

Multiple attribute decision making(MADM)is widely used in a variety of fields including society,economy and management.The study of MADM has always been an important hot topic in decision analysis.From the perspective of decision making process,the key process of MADM can be divided into two parts,one is to determine preference-strength values(PSVs),the other is to determine attribute weights and get comprehensive value of alternative by combining with the PSVs.For the PSVs,different from the traditional MADM method which only uses the simple vector normalization method and other normalized methods to solve the uncommensurability among various attributes,Bana e Costa et al.has proposed the Macbeth method which aims to measure the PSVs on multiple attribute dimensions with the same interval measurement.For attribute weights,the traditional MADM methods based on fixed weights with the assumption of attribute preference independence have the problems of failure when solving practical problems.To solve this problem,two kinds of MADM methods based on variable weights which can reflect the point-dependence preference correlation or the ordinal-dependence preference correlation have been proposed.However,these methods mentioned above have some defects in theory and application when solving MADM problems.For example,the set of ‘ Good ’performance values given by MACBETH approach on multiple attribute dimensions do not have the same preference meaning that guarantees the commensurability on attribute performances,and the determination of state variable weights for multiple factor decision-making method based on variable weights(MFDM-VW method)is subjective arbitrary and the use of fixed weights is logically inconsistent with the core idea of method construction,and so on.If the defects and shortcomings of these theory can be solved,it will have important theoretical and practical significance to guide people’s decision making.At first,this paper aims at the drawback of Range-Frequency Theory(RFT),i.e.,the range value(RV)is of unreasonable in reflecting the subjective preference,proposes a modified range value(MRV)based on the reference point and value function with Cumulative Prospect Theory(CPT),and presents a modified RFT model which based on the basis of MRV.An experimental research,which consists of prospect value judgment experiment and subjective judgment experiment based on AHP judgment matrix,adopts to test the science reasonability for the modified RFT model.The experimental results indicate that reference point,subjective value parameter and loss aversion parameter have significant effects on subjective judgment,confirm that the modified RFT model which comprehensively reflect these effects can outperform RFT model in fitting data,and show that the parameters of the modified RFT model reveal high stability in different fitting methods.For the set of ‘Good’performance values given by MACBETH approach on multiple attribute dimensions do not have the same preference meaning that guarantees the commensurability on attribute performances,a targets-oriented PSV(TOPSV)function,which takes targeted alternatives as decision targets,and a methodological procedure for measuring the commensurable PSVs in multiple attribute decision making,are presented by data envelopment analysis and Markov theory,and based on the development of CPT and RFT.The application of the procedure to an experimental decision has directly verified its effectiveness and indirectly verified the science reasonability of the TOPSV function.Aiming at the shortcomings existing in MFDM-VW method,such as without considering measuring the commensibility of PSVs,subjective arbitrary determinating the optimistic coefficient y subjective arbitrarily,and logically inconsistent between the use of fixed weights and the core idea of method construction,this paper proposes a target-oriented MFDM-VW(TOMFDM-VW)method,which based on the TOPSV and Swing weighting,can overcome the above-mentioned shortcomings of MFDMVW method.The case study shows that the TOMFDM-VW method can give the ranking result which is very consistent with the decision maker’s qualitative view of preference dependence correlation and objective facts.Compared with the MFDM-VW method,the TOMFDM-VW method can reflect the specific preference of the decision maker more appropriately,so it has stronger scientific rationality.Considering that multiple attribute decision making method with relative variable weights(MADM-RW method)does not measuring the commensibility of PSVs,the TOPSV method is introduced to realize measuring the commensibility of PSVs,and on this basis,a target-oriented MADM-RW(TOMADM-RM)method is presented.The example analysis shows that in the case of different target alternatives,the TOMADMRM method not only can reflect the specific preference reasonably,but also can reflect the changed ranking results towards different target alternatives solution effectively.Thus,scientific rationality of TOMADM-RM is verified.For the MADM problem in the case of fuzzy decision information,a targetoriented ordinal-dependence preference decision making(TOODM)method based on probabilistic linguistic preference relation(PLPR)is proposed.Firstly,the TOPSV function and PLPR are introduced to measure the PSVs with fuzzy decision information.Secondly,the parameters of TOPSV function and capacity of attribute set is determined by PLPR judgment matrix,which reflects the decision makers’ preference information flexibility and truly,and the comprehensive values for selective alternatives under the framework of ordinal-dependence preference correlation is determined according to Choquet integral with fuzzy decision information.An application of the problem shows the feasibility of the application of the TOODM method in the case of fuzzy decision information.Finally,combined with the actual decision making problems of wind power equipment bid evaluation,this paper applies and demonstrates the TOPSV method,the TOMFDM-VW method and TOMADM-RM method.The application process and analysis results show that the TOMFDM-VW method and TOMADM-RM method are feasible,and have stronger practicability than the MFDM-VW method.