Research And Application Of Group Decision Risk Evaluation Based On Preference Features

Since the beginning of the 21 st century,with the rapid development of China social economy,various natural disasters,man-made disasters,and public safety disasters have caused huge security risks to our society and people.Therefore,the evaluation and decision-making of risk events is of great significance.To evaluate risk events in a complex environment,it is irrational to rely on an individual or a person in a field to make decisions.Events are often affected by multiple perspectives.Decision makers are susceptible to complexity when evaluating solutions Influence of environment and personal subjective preferences.Based on AHP and group decision theory,this paper studies the problems of incomplete decision information,unsatisfactory consistency and expert preference that are difficult to quantify,and large deviation between group decision and individual decision results.Evaluation and decision-making,the main work is as follows.(1)In complex multi-dimensional decision-making problems,decision makers are uncertain about the comparison of certain indicators and do not evaluate them,resulting in incomplete information in the decision matrix.Aiming at the situation that the judgment matrix information is incomplete,a weighted least squares method for fitting missing values and ranking vectors is proposed.An adjoint matrix of an incomplete matrix is defined to eliminate the influence of missing values on the fitting result;the missing values are predicted by setting the amount of errors,and the geometric ordering is performed using the ranking vector obtained by the fitting and the ranking vector of the complete judgment matrix,thus we can get the final sort vector.Through the comparisons on a same example,the obtained results are satisfactory,which illustrates the feasibility and effectiveness of the algorithm.(2)Aiming at the problem that the decision matrix does not satisfy the consistency,the consistency correction algorithm is proposed for AHP reciprocal judgment matrices.Thisalgorithm uses the H-convex combination to adjust the inconsistency matrix,and corrects the inconsistency matrix through the modifiable range given by the expert,so that expert opinions are fully retained and the correction process is more flexible.Finally,the experimental results show that when the correction ratio is 20%,the similarity reaches85.93% and the reduction degree reaches 99.3%.In the condition of ensuring the same ranking,the decision information of the original judgment matrix is fully retained,making the correction more reasonable.(3)In the risk assessment of actual events,a matrix-aggregated intelligent decision-making model based on H-convex combination and expert preference is proposed to solve the subjective preference problem in the evaluation process.The H-convex combination matrix algorithm is used to aggregate the individual judgment matrices to obtain a group decision matrix,which eliminates the inconsistencies and non-reciprocity problems in the assembly process;the defined expert preference quantification formula is used to obtain the corresponding expert's preference weights,which are brought into H-Convex combinations as power exponents to resolve the impact of expert preference differences on decision outcomes.Compared with the two traditional models,the relative error of the group decision matrix after assembly was reduced by 9.3% and 12.29% respectively,compared with the two traditional models.Compared with the second scheme,the difference reaches 98.22%,which is better than the two traditional models,which reduces the uncertainty of decision results and makes decision-making easier.