基于粗糙集-神经网络编队协同空战决策系统

Cooperative Team Air Combat Decision based on Integration of Rough Sets and Neural Networks

为了提高空战决策的实时性和准确度,将粗糙集理论和神经网络引入到编队协同空战战术决策研究中,提出了应用SOM网络-粗糙集-BP网络集成进行编队协同空战战术决策的方案:应用SOM网络离散化决策系统输入数据的连续属性值;利用粗糙集数据分析方法,从数据中提取出规则将输入映射到输出的子空间上;在这个子空间上用BP网络进行逼近。2∶4编队空战实例仿真结果验证该方案的可行性,在数据充分的条件下,该方案可以推广应用于其它空战决策系统。

Considering the ability of rough sets theory on reduction of decision system and that of neural networks for clustering and nonlinear mapping,a new hybrid tactics decision system of rough sets and neural networks for cooperative team air combat is presented. Firstly, the continuous attributes in air combat decision system were discretized with self-organizing map neural networks. Then, the input to the model is mapped into the output subspaces by using rules acquired from rough sets. Lastly, the output of the system is approximated by BP neural networks in the subspaces. Simulation results for 2 via 4 cooperative team air combat show the method can reduce the cost of decision system and increase the efficiency of decision. With enough sample data, the solution can be applied to other decision systems in air combat