Research And Application Of Belief Rule Base Inference Methodology For Two-sided Matching Decision

With the rapid development of the information age,the amount of data generated by all walks of life almost an exponential increase.People continue to strive for excellence in data research,the data can bring endless benefits to pursue.Therefore,the research on data processing has been a hotspot in recent years.Matching decision-making method plays an indispensable role in data processing.It is not only in the commercial field cut a striking figure,but also the target tracking and data fusion applications in the extremely critical steps.Matching decisions are the process of dealing with a group of one or more objects in another group with one or more objects in the process of a reasonable matching.In this process,we need to give full consideration to the satisfaction information about all the objects to be matched.The ultimate goal of the decision is to find a two-sided matching relationship so that it can satisfy the needs of the objects to be matched as much as possible.Although the problem of matching decision has attracted wide attention to scholars,some strategies have been put forward to solving the problem.However,when the information carried by the object is imprecise or incomplete,the existing matching decision-making method is unsatisfactory.In view of this,the belief rule base reasoning method is introduced into the field of two-sided matching decision-making,which has the ability to comprehensively utilize and process all kinds of uncertain information.Its reasoning performance is affected by two aspects,one for the parameter training process,and the other for the rule base structure.Based on the above two factors,this paper study and optimizes the process,and finally modifies the parameter training method to make it better integrated into the two-sided matching decision problem.Specific research work is as follows:(1)The problem that the method of parameter optimization based on swarm intelligence is not ideal,this paper introduces the artificial bee colony algorithm and proposes a new parameter optimization strategy.By adding the Gaussian perturbation factor,the algorithm avoids getting into the local optimal solution at the beginning of the iteration,automatically shortens the perturbation domain in the later stage of the algorithm,and preserves the current optimal solution,at the same time,further excavates the precision of the solution.Finally,the validity of this method is proved by using multi-extremum function fitting and oil pipeline leak detection.(2)The belief rule base constructed by the linear combination method cannot give full play to the performance of the weight of the antecedent attribute,and propose two-value judgment method,first set the posterior evaluation level of the rule only two,Only to make two choice decision.Second,set up multiple belief rules library to deal with several sub-problems at the same time.Finally,the results of multiple sub-problems are merged by means of decision making.The experimental data are verified by the classification data.(3)In order to solve the multi-attribute two-sided matching decision problem with uncertain information,an improved adoption of cutting and linear mapping method based on the belief rule base is proposed.The method can solve the problem that when the input value of the BRB system reaches the threshold,the output will be induced to be wrong,and the input value will be regarded as indeterminate by forced truncation.When the adoption of cutting method is not enough,the linear mapping method is used to reduce the adverse effects on the results.Finally,the experimental results show that the proposed system has better reasoned ability.