Research On The Modeling Method Of Operational Simulation And Effectiveness Evaluation Driven By Bayesian Network

With the development of modern high technique, the mode of modern warfare is gradually evolving from the simple countermeasure between weapons to a more complex mode -- the countermeasure among systems and systems which is characterized by the cooperation of multi-platform and multi-system. For the reason of complexity of cooperation or counter measure, the result of modern warfare often takes on a character of uncertainty. In this condition, traditional simulation effectiveness evaluation method will facing the problem of too much work on designing and building simulation models. Meanwhile, large number of models will lead to low efficiency in computation. Therefore, it's important to propose a simulation model design method to fulfill the evaluation need. We call this the persistent problem of evaluation and simulation model design. In addition, a lot of useful information is contained in the simulation output data. Researches on how to build evaluation model from these data and the applications of the model in weapon system effectiveness evaluation is very important.To solve this issue, this thesis introduced Bayesian Network into the operational effectiveness evaluation issue and studied the operational simulation model and effectiveness evaluation model driven by Bayesian Network. In the research, this thesis presented the methodology and some principles of simulation model building oriented to the prior Bayesian Network informantion, gave the s designing and building method of simulation model based on SMP2.0. To describe the uncertain prior information, this thesis presented an"Uncertain Prior Structure Knowledge Description Method", and use Belief Map to syncretize the prior information from expert group. In the research, the thesis also gave a method named"Uncertain Prior Knowledge Bayesian Network Structure Learning Method"based on improved MDL score and simulated annealing method, which could properly solve the Bayesian Network structure learning issue with prior expert knowledges. Finally, this thesis gave an example of operational effectiveness evaluation of anti-ship missile coordinate attack and acquired satisfactory results.By study and research, this thesis offered a new way of evaluating the operational effectiveness of weapon systems, and it is also of the reference value for other correlated issues.