基于支持向量机的坦克驾驶模拟训练结果分析  

Analysis of Tank Driving Simulation Training Results Based on Support Vector Machine

作　　者：邓青 薛青 罗佳 DENG Qing;XUE Qing;LUO Jia(Training Center,Academy of Army Armored Forces,Beijing 100072,China)

机构地区：[1]陆军装甲兵学院演训中心

出　　处：《兵工学报》2019年第9期1953-1960,共8页Acta Armamentarii

基　　金：武器装备预先研究项目(41404060205)

摘　　要：利用坦克驾驶模拟器进行模拟训练是提高装备操作技能的重要方法。针对以往模拟训练忽视训练数据采集分析和提高训练质量的问题,提出采用支持向量机(SVM)对坦克驾驶模拟训练结果进行分析的方法。为了解决SVM参数选取难的问题,提出一种自适应粒子群优化(APSO)算法对SVM参数进行优化选择,设计动态权重参数并赋予相关惯性,实现粒子动态自适应。引入多位置查询机制和极值点信息以维持不同粒子平衡点的多样性,通过迭代选择与优化目标函数实现对参数的自动寻优。基于APSO算法的支持向量机(SVM-APSO)应用到某型坦克驾驶模拟器的训练结果分析中,结果表明SVM-APSO能克服多维影响因素对训练成绩分类带来的不利影响,实验结果在精度和时间上都有明显优势,验证了SVM-APSO在坦克驾驶模拟训练结果分析中应用的可行性与有效性。Training by tank driving simulator is an important way of improving equipment skill.A method of training effect analysis of tank driving simulator based on support vector machine(SVM)is proposed for the problems resulted from ignoring the collection and analysis of training data and the increase in training quality.In order to solve the issue of choosing SVM parameters,an adaptive particle swarm optimization(APSO)algorithm is adopted to determine the SVM parameters.Dynamic weight parameters are designed and the related inertance is entrusted for realizing the self-adaption of particle.The multi-location inquiry mechanism and the information from extreme point are used to keep the balance dot diversity of different particles.The SVM parameters can be automatically optimized by iterating and optimizing the object function.After the SVM based on particle swarm algorithm is used for the training effect analysis of tank driving simulator,the adverse effect of multidimensional factors on training score can be overcome.The experimental results show that SVM can be feasible and effective in the training effect analysis of tank driving simulator.

关 键 词：坦克 支持向量机 驾驶模拟器 模拟训练 结果分析 粒子群优化算法 

分 类 号：E92[军事—军事装备学]