基于加权支持向量回归的抢修时间估计模型  

BDAR Time Assessment Model Based on Weighted Support Vector Regress Machine

作　　者：尤志锋[1] 石全[1] 熊飞[1] 

机构地区：[1]军械工程学院,河北石家庄050003

出　　处：《现代防御技术》2014年第4期160-166,共7页Modern Defence Technology

摘　　要：已有的抢修时间估计模型大都印有平时维修的痕迹,不能很好的反应战场抢修的随机性、多样性、时效性等特点。分析并设计了影响抢修时间的因素及其赋值方法,用复杂性来度量抢修任务本身的属性。将抢修时间估计问题转为抢修时间对其影响因素的非线性回归问题,引入在处理小样本、非线性问题时有较大优势的支持向量机,利用遗传算法对支持向量回归的参数进行优化;实验结论证明模型的估计精度较高、泛化能力较强;从一个新的角度估计抢修时间,结果更合理,能为抢修决策以及抢修训练提供良好的帮助。The existing time assessment model for rush repair time are all based on the maintenance in peace time.It cannot reflect the random,timeless and multiform of battlefield damage assessment and repair（BDAR）.The influence factors of BDAR task were analyzed and the assignment method was given.The task complexity is a substitute for the task＇ s essence.Then the BDAR time assessment can be transformed to the regress of BDAR task complexity vector and its time.To deal with its small sample,nonlinear characteristic,the support vector regress machine was introduced.And the genetic arithmetic was applied to optimize the parameters of SVR model.The veracity is well proved by the instance.At last the BDAR time can be assessed on a new concept,and it can help the BDAR decision and training simulation.

关 键 词：复杂性度量 加权支持向量回归 遗传算法 抢修时间 

分 类 号：E92[军事—军事装备学] E911[兵器科学与技术—武器系统与运用工程]