基于深度学习的网状指标体系作战效能评估方法研究

Research on the Operational Effectiveness Assessment Method of Mesh Structure Index System Based on Deep Learning

随着装备体系越来越复杂,传统的树状指标体系不能完整表征装备体系的作战效能,需构建网状指标体系。常用的层次分析法不能解决网状指标体系作战效能评估问题,提出了基于深度学习的作战效能评估方法,将装备体系的效能评估结果分为不同的效能等级,装备体系的效能评估问题转换为效能等级的分类问题。将该方法应用到某数据中心的作战效能评估中,利用历史数据和仿真系统生成的5000组数据训练神经网络模型,测试准确率为99.3%,预测准确率优于Bagging分类法、随机森林分类法等机器学习分类算法。将某次作战试验实测数据输入模型,预测结果为该数据中心的作战效能为“好”,与实际作战使用情况较为一致。

As the equipment system becomes more and more complex,the traditional tree structure index system can't represent the operational effectiveness assessment of equipment system.The mesh structure index system is needed to be built.The common used analytic hierarchy processing(AHP)method can't solve the problem of the operational effectiveness assessment of mesh structure index system.A new operational effectiveness assessment method based on deep learning is proposed.The operational effectiveness assessment results of the equipment system is divided into different effectiveness levels.The operational effectiveness assessment problem of the equipment system is converted into the classification problem of effectiveness level.This method is applied into the operational effectiveness assessment of a certain data center.5000 sets of data training neural network model generated by historical data and simulation system are used,the test accuracy is 99.3%,the prediction accuracy is higher than that of such machine learning methods as bagging classification method and random forest classification method,etc.The measured data of a certain operational test is input to the neural network,the prediction results show that the operational effectiveness of the data center is good,and it is consistent with the actual operational status.