动能弹对混凝土靶侵彻深度的PSO-SVM预测  

Prediction of Penetration Depth of Projectiles into Concrete Targets Based on PSO-SVM

作　　者：潘强[1] 张继春[1] 肖清华[1] 邹新宽 石洪超[1,3] 

机构地区：[1]西南交通大学土木工程学院,四川成都610031 [2]自贡市城市建设投资开发集团有限公司,四川自贡643000 [3]成都工业学院建筑与环境工程系,四川成都610039

出　　处：《高压物理学报》2018年第2期106-113,共8页Chinese Journal of High Pressure Physics

基　　金：国家自然科学基金(50574076)

摘　　要：目前混凝土毁伤效应中侵彻深度的预测对防护工程设计与建设有着重要的指导意义,传统的预测方法存在样本需求量大、预测误差大等问题。根据支持向量机原理,采用粒子群算法优化模型参数,提出了预测动能弹侵彻深度的粒子群-支持向量机方法,并编写了相应的计算程序,通过援引实测数据验证预测的准确性。结果表明:该方法对于小样本、非线性预测有较大优势,相比于传统的灰色理论预测,其预测相对误差较小(最大相对误差为3.18%);随着训练样本量增多,最大相对误差逐渐减小,且变化速率逐渐减缓,但计算量增大。因此,粒子群-支持向量机方法用于动能弹侵彻混凝土靶体的深度预测是合理可行的。The prediction of the penetration depth of concrete in concrete damage effect is of great significance to the design and construction in protection engineering.However,the traditional methods for this prediciton involve such problems as requiring agreat supply of samples,or suffering from a large prediction error,and so on.Inthiswork,following the theory of the support vector machine（SVM）and according to the parameters optimized through the particle swarm optimization（PSO）,the PSO-SVM for predicting the penetration depth was proposed.The corresponding programs were written and the prediction was verified by the experiment data.The results show that the PSO-SVM method has a great advantage for small samples and non-linear prediction.In co m parison with the traditional grey theory,the relative predicted errors through the PSO-SVM method are smaller（the maximum relative error being 3.18%）.As the number of the samples increases,the maximum relative errors decrease and the changing rate slows down whereas,however,the amount of calculation becomes larger.Above all,it is feasible to apply PSO-SVM method to the prediction of penetration depth of projectiles into concrete targets.

关 键 词：粒子群优化 支持向量机 混凝土靶 侵彻深度 预测 

分 类 号：O385[理学—流体力学] O24[理学—力学]