基于改进的LS-SVM地空导弹生存能力的预测  

Prediction of Surface-to-air Missile Survivability Based on Improved Least Square Support

作　　者：王改堂 王斐[2] 黄超凡 郑松海 赵娜 叶锦函 

机构地区：[1]中国兵器工业第203研究所,西安710065 [2]河南科技大学信息工程学院,河南洛阳471003

出　　处：《弹箭与制导学报》2014年第6期43-46,50,共5页Journal of Projectiles,Rockets,Missiles and Guidance

摘　　要：针对最小二乘支持向量机(LS-SVM)缺失的鲁棒性以及参数难以选择的问题,提出了改进的LS-SVM预测建模方法。该方法首先从样本特征重要程度的角度出发构造特征加权矩阵,然后利用该特征加权矩阵对LS-SVM进行加权,最后采用了人工鱼群优化算法选择模型参数。通过对基准数据集的仿真验证了该方法的有效性和可行性,并将其预测地空导弹生存能力,获得了满意的结果。A novel prediction modeling algorithm based on improved LS-SVM was proposed to solve the problem of LS-SVM algorithm lacking robustness and hard parameter selection. Firstly,the method structures feature weighting matrix from the feature importance of sample,and then uses the matrix to weight for LS-SVM. Finally,the paper uses artificial fish swarm algorithm to choose parameters of the model. Simulation results verified the effectiveness and feasibility of this method by using some real-world datasets,and it was applied to develop a prediction model for surface-to-air missile survivability,and the result was satisfied.

关 键 词：最小二乘支持向量机 特征加权 人工鱼群算法 地空导弹 生存能力 

分 类 号：TJ762.21[兵器科学与技术—武器系统与运用工程]