决策树集成方法在反舰导弹效能评估中的应用  

Decision Tree Integration Method Application in Effectiveness Evaluation of Anti-Ship Missile Weapon System

作　　者：姬正一 陈阳 沈培志 韩先平[1] 齐鸿坤 JI Zheng-yi;CHEN Yang;SHEN Pei-zhi;HAN Xian-ping;QI Hong-kun(PLA,No.92941 Troop,Liaoning Huludao 125000,China;PLA,No.92943 Troop,Liaoning Huludao 125000,China;Naval Aviation University,Shandong Yantai 264001,China)

机构地区：[1]中国人民解放军92941部队,辽宁葫芦岛125000 [2]中国人民解放军92493部队,辽宁葫芦岛125000 [3]海军航空大学,山东烟台264001

出　　处：《现代防御技术》2021年第4期15-23,34,共10页Modern Defence Technology

摘　　要：在反舰导弹效能评估方法中,针对存在主观经验和计算时间成本高的问题,提出了随机森林和梯度提升回归树2种决策树集成方法。通过构建3层19个分量的反舰导弹效能评估指标体系,改进的ADC(availability dependability capacity)评估模型建立了不同状态反舰导弹武器系统参数数据样本240份,切分数据集后采用归一化处理,结合交叉验证和网格搜索等参数优化方法,得到了2个较为理想的决策树集成效能评估模型。在仿真试验测试验证中,模型的评估准确率较高,验证了该方法的实用性,为反舰导弹效能评估提供了新思路。Methods based on random forest and gradient boosting regression tree are proposed in order to improve the problem of subjective experience and high cost of calculation time in the effectiveness evaluation of anti-ship missile.By constructing a 3-layer and 19-component anti-ship missile effectiveness evaluation index system,240 operational parameter data samples of anti-ship missile in different states are set up by the improved ADC evaluation model,then the data set is segmented and normalized,combining with cross validation and grid search,two ideal decision tree integration effectiveness evaluation models are obtained.In the simulation test,the accuracy of the two models is high,which proves the practicability of the method and provides new ideas for anti-ship missile effectiveness evaluation.

关 键 词：反舰导弹 效能评估 决策树集成 随机森林 梯度提升回归树 机器学习 

分 类 号：TJ761.14[兵器科学与技术—武器系统与运用工程] N945.1[自然科学总论—系统科学]