摘要
涡轮盘作为航空发动机的关键部件,在高温、高转速的严酷条件下工作,低循环疲劳成为涡轮盘的主要失效模式。而且涡轮盘的结构复杂性,在进行疲劳可靠性分析时,直接使用Monte-Carlo法的计算量非常大,而传统的响应面法精度达不到计算要求。Fourier正交基神经网络具有很强的非线性逼近能力,在对涡轮盘进行疲劳寿命分析时,采用Fourier正交基神经网络和Monte-Carlo法相结合的方法,并与传统响应面法和Monte-Carlo法进行对比。结果表明,前者不但可以满足精度要求,而且效率高,在涡轮盘等复杂结构可靠性分析中具有很好的工程应用前景。
Turbine disk is the key component of aviation, working tigue has become the main failure mode of turbine disk. Because in high temperature and high pressure, the cycle fa- of the complexity of turbine disk, the computation of directly using the Monte-Carlo method is very large,while the traditional response surface method is not up to the re- quirements. The Fourier orthogonal neural network has strong capability of nonlinear approximation. In the analysis of fatigue life of turbine disk, using the Fourier orthogonal neural network with Monte-Carlo method, compare with the method of traditional response and Monte-Carlo, the result show that the former method can not only safety the require- ment of accuracy, and high efficiency, and has Rood prospect of en^ineerin~ application in such complex structures.
出处
《装备制造技术》
2014年第10期132-134,共3页
Equipment Manufacturing Technology
基金
国家自然科学基金(5137503251335033)