小样本条件下基于GA-BP的CO方案评估  

GA-BP-Based Evaluation of Course of Actions in Cyberspace under the Condition of Small Sample

作　　者：陈登 陈楚湘 周春华 邢积超 CHEN Deng;CHEN Chu-xiang;ZHOU Chun-hua;XING Ji-chao(PLA Strategic Support Force Information Engineering University,Zhengzhou Henan 450000,China;Army Academy of Artillery and Air Defense,Zhengzhou Henan 450000,China)

机构地区：[1]中国人民解放军战略支援部队信息工程大学,河南郑州450000 [2]陆军炮兵防空兵学院,河南郑州450000

出　　处：《计算机仿真》2023年第12期8-15,共8页Computer Simulation

摘　　要：基于对作战方案评估及网络空间作战理论的分析,构建了网状网络空间作战方案评估指标体系。由于评估指标体系指标间关系非独立且非线性明显,同时可供量化提取的评估数据样本容量较小,依据统计学基本原理,提出无放回抽样样本扩充方法与GA-BP神经网络模型相结合的评估模型。解决了非独立、非线性评估指标评估问题和GA-BP面对小样本时随机性大、评估精度不高的问题。经验证,基于无放回抽样样本扩充方法优化的模型评估精度高、稳定性强,能提供正确的网络空间作战方案评估结论,为指挥员指挥决策提供可靠依据。Based on the analysis of course of actions evaluation theory and cyberspace operations theory,a network-based cyberspace operational plan evaluation index system has been constructed.Since the relationship among the indexes in the evaluation index system is not independent and nonlinear,and the sample capacity of evaluation data available for quantitative extraction is small,this paper proposed an evaluation model combining the sampling expansion method without replacement and the GA-BP neural network model according to the basic principle of statistics.This method solves the problems of non-independence and nonlinearness among evaluation indexes and high randomness and low evaluation accuracy of GA-BP when facing small samples.It has been verified that the model optimized by the method of sampling without replacement has high evaluation accuracy and strong stability.This method can provide the correct evaluation conclusions for cyberspace operation plans,and provide a reliable basis for commanders to command and make decisions.

关 键 词：作战方案评估 网络空间作战 评估指标体系 无放回抽样 样本扩充 

分 类 号：E917[军事]