摘要
涡轮盘是发动机的关键部件,在高温、高转速的条件下工作,因此对其进行疲劳可靠性分析具有重要意义。由于涡轮盘结构热-机械耦合分析的复杂性,对其进行可靠性分析时,直接用Monte-Carlo方法计算量太大,常规的多项式响应面方法在精度上又难以满足要求。径向基函数(radial basis function,RBF)神经网络具有很强的非线性函数逼近能力,在涡轮盘低循环疲劳可靠性分析中采用RBF神经网络结合Monte-Carlo的方法得到了疲劳寿命的概率分布,并与直接用Monte-Carlo模拟和响应面方法进行了对比。RBF神经网络结合Monte-Carlo的方法具有高精度、高效率的优点,在涡轮盘等复杂结构可靠性分析中具有很好的应用前景。
Turbine-disk is the key component of aero-engine, it works under the conditions of high temperature and high speed, thus to carrying out analysis on its fatigue reliability possesses important significance. On account of the complexity of the analysis on the thermo-mechanical coupling of turbine-disk, while carrying out its reliability analysis the amount of calculation is too large by using the directly Monte-Carlo method and the conventional multinomial response surface method is also difficult to satisfy the requirements on precision. The neural network of radial basis function （RBF） has quite strong nonlinear functional approach ability, adopting the combination of RBF neural network with Monte-Carlo method the probability distribution of fatigue life-span was obtained in the reliability analysis of low cycled fatigue of turbine-disk. A comparison has been carried out between the direct adoption of Monte-Carlo simulation and the response surface method. The combination of RBF neural network with Monte-Carlo method has advantages of higher precision and higher efficiency, which bears good application prospects in the reliability analysis of complex structures of turbinedisk etc..
出处
《机械设计》
CSCD
北大核心
2009年第5期8-10,14,共4页
Journal of Machine Design
关键词
涡轮盘
热-机械耦合
径向基函数
低循环疲劳
可靠性
turbine-disk
Thermo-mechanical coupling
radial basis function （RBF）
low cycled fatigue
reliability
