基于优化最小二乘支持向量机的襟翼趋势预测  

Trend Prediction of Aircraft Flap System Based on Optimized Least Square Support Vector Machine

作　　者：王旭辉[1] 黄圣国[1] 曹力[1] 施鼎豪[2] 舒平[2] 

机构地区：[1]南京航空航天大学民航学院,南京210016 [2]中国民用航空总局安全技术中心,北京100028

出　　处：《南京航空航天大学学报》2007年第3期388-393,共6页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：国家863计划资助项目(2006AA12A108)

摘　　要：将最小二乘支持向量机(LS-SVM)应用于飞机襟翼状态趋势研究。首先,通过分析飞机襟翼故障与襟翼动作耗时参数的关系,提出了利用动作耗时趋势来确定该系统未来状态的方法。然后,使用最小二乘支持向量机建立耗时回归预测模型,采用最终预报误差(FPE)准则确定回归模型嵌入维数,提出了自适应网格搜索法,优化最小二乘支持向量机的建模参数,从而实现比现有方法精度高,泛化性能好的预测模型。训练和测试样本取自飞行数据记录系统(QAR)中译码襟翼参数值。与神经网络模型的比较实践表明,该方法具有实用价值。Least square support vector machine （LS-SVM） is used to predict the trend of the aircraft flap system. Firstly, by analyzing the relationship between time-consuming of the flap action and the fault of flap system, a model based on the time-consuming trend of the flap action is used to predict the trend of the flap system. Secondly, LS-SVM is used to establish the time-consuming model. The final prediction error （FPE） principle is used to optimize the embedded dimension. Then, an adaptive grid search （AGS） method is proposed to optimize the model parameters of LS-SVM. The training and measuring data set is obtained from quick access recorder （QAR）. Finally, the one-step and n-step prediction results of LS-SVM and RBF neural network are compared. The method shows that LS-SVM regression model is feasible and accurate for the trend prediction of the flap system.

关 键 词：襟翼系统 趋势预测 支持向量机 参数优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]