基于PSO-LSSVM的飞机维修保障费用预测模型  

AIRCRAFT MAINTENANCE SUPPORT COST PREDICTION MODEL BASED ON PSO-LSSVM

作　　者：陈高波[1] 

机构地区：[1]武汉工业学院数理科学系,湖北武汉430023

出　　处：《计算机应用与软件》2010年第9期181-183,共3页Computer Applications and Software

摘　　要：考虑到飞机维修保障费用数据样本容量小和难于预测的特点,提出用最小二乘支持向量机LSSVM(Least squares support vector machine)来预测飞机维修保障费用。采用粒子群算法PSO(Particle Swarm Optimization)优化LSSVM的参数,并同偏最小二乘回归(PLSR)的预测结果进行了比较。结果表明,PSO-LSSVM预测模型可调参数少、速度快,预测精度比PLSR有显著提高。Least square support vector machine （ LSSVM ） is suggested to predict aircraft maintenance support cost for the consideration that the data of aircraft maintenance support cost has characteristics of small sample capacity and difficult to predict. In this paper we applied Particle Swarm Optimization to optimizing the parameters of LSSVN ,and compared the prediction results with that of the partial least square re- gress （PLSR）. The result showed that PSO-LSSVM has few parameters to tune and has fast trained speed,and its prediction precision is no- ticeably higher than PLSR.

关 键 词：最小二乘支持向量机 粒子群 费用 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]