摘要
支持向量机是一种新的机器学习方法。它以统计学习理论为基础,从结构风险最小化原则出发,具有很好的泛化及推广能力。通过采用支持向量机对飞机襟翼结构在声激励试验中获得的总应力、总加速度及总声压数据进行分类,分别获得结构响应间的线性关系和非线性关系的样本点以及线性与非线性界限。最后利用多元线性回归模型分别建立线性关系和非线性关系的仿真模型。研究表明,支持向量机在模式分类中具有良好的分类性能,在解决航空结构声疲劳问题中有较好的工程应用前景。
Support vector machine is a new kind of machine learning method. This machine learning method based on statistical theories and structure risk minimum principle, has favorable capabilities of universality and generalization. SVM is adopted to carry out classifier function for the measured data（total stresses ,total accel- eration and total sound pressure） of an airplane shell construction experiment, excited by higher pressure level noise in this paper. The limitation of the linear and non - linear relationships both structural responses are ob- tained. At last, by regression analysis, the simulation models of the linear and non - linear relationships both structural responses of the aircraft flap under acoustic excitation are given out. SVM has excellent classification profermance in pattern recognition and of good prospect in the area of engineering application of aircraft structure acoustic fatigue.
出处
《沈阳航空工业学院学报》
2008年第3期15-18,共4页
Journal of Shenyang Institute of Aeronautical Engineering
关键词
声激励
支持向量机
RBF核函数
应力响应
模式识别
仿真
acoustic excitation
support vector machine
RBF kernel function
stress response
pattern recognition
simulation
