摘要
首先介绍了近似方法中的二次响应面方法、径向基神经网络方法、Kriging方法和增强的径向基函数方法。结合测试函数和某机翼模型为例,分别用这四种方法构造测试函数和某机翼结构和气动学科的近似模型,并总结这四种方法的特点和在对各种模型近似中的适用性。结果表明,Kriging方法和增强的径向基函数方法的近似是非常精确和稳健的。
The paper investigates square response surface method, Radial Basis Function Neural Network method , Kriging method and Augmented radial basis functions method. Taking example for a test function and a straight airfoil, making use of the four methods to construct the approximation models of test function and the airfoil structure and aerodynamic disciplinary, and summarize the four methods＇characters and availability in every model. It is found that Kriging method and Augmented radial basis functions method are very accurate and robust.
出处
《航空计算技术》
2005年第3期5-8,16,共5页
Aeronautical Computing Technique
基金
国家自然科学基金(10377015)资助项目
关键词
二次响应面方法
神经网络方法
KRIGING方法
增强的径向基函数方法
近似模型
square response surface method
radial basis function neural network method
Kriging method
augmented radial basis functions method
approximation model
