基于RBF网络的重力坝结构可靠度分析

计算重力坝结构可靠度时，功能函数往往是高度非线性且为隐式的，此时经典的计算方法，如JC法、Monte-Carlo方法会由于计算量及计算时间的大量增加而不再适用。此时，应用神经网络方法是一种较好的改进方式，其中RBF网络具有收敛速度快，适于非线性拟合且可实现全局收敛的特点。本文探究了RBF网络在重力坝结构可靠度中的应用模式，并结合实例分析了其合理性。

RELIABILITY ANALYSIS OF GRAVITY DAM BASED ON RBF NETWORK

The performance function tends to be implicit or nonlinear in the evaluation of gravity dam reliability, making it difficult to apply some classical methods, such as JC method, Monte-Carlo simulation etc., as they are supposed to be too time-consuming. One possible solution to this problem may be the introduction of artificial neural networks, among which RBF is featured with faster convergence, better precision and can realize global convergence to some extent. In this paper, the application of RBF in gravity dam reliability is investigated, with some examples presented to convince that it’s reasonable to put it into use.