Research And Application Of Uncertainty Situation Assessment Method For Simulation Syste

As the simulation system becomes more complex,the uncertain behavior of autonomous agents becomes more and more frequent during the simulation process.How to correctly characterize and evaluate this uncertainty,and how to introduce uncertainty methods in simulation modeling,evolutionary deduction,etc.These key issues have become a new research direction of system simulation.Methods such as rough sets and cloud models are the current mainstream research uncertainty tools.Based on the improved uncertainty method,this paper conducts in-depth research on the three problems of uncertainty variable representation,uncertainty evolution and uncertainty situation assessment in the simulation system.The specific work is as follows:Aiming at the problem of uncertainty variable characterization,this paper adopts the method of combining rough set and cloud model to characterize and reduce uncertain data.A filling algorithm for missing data is proposed,which is added to the rough set attribute reduction algorithm based on information entropy.The improved information entropy rough set attribute reduction algorithm has better applicability to uncertain samples with missing data.Validate attribute reduction methods by case.Aiming at the problem of uncertainty evolution,this paper proposes an uncertainty evolution method based on gated graph neural network based on gated recurrent unit(GRU).The method improves the GRU gated cyclic unit,adds uncertainty logic gate,and uses the membership degree of the uncertainty data as the evolution parameter to study the model evolution process of the uncertainty parameter matrix.The effectiveness of this method is verified by comparative experiments.Aiming at the problem of uncertainty situation assessment,this paper proposes an improved cloud model-rough set combined uncertainty situation assessment method.In the process of calculating the digital features of the evaluation index cloud model,this method further calculates all top-level index cloud models to obtain a comprehensive cloud model digital feature,and uses this parameter as a grade evaluation parameter,which reduces the need for separate evaluation and calculation of each index in the original algorithm.steps,as well as the similarity calculation steps for evaluating the cloud model,simplifying the original cloud model-rough set situation evaluation algorithm.Through the case,the feasibility of the method is verified.