To improve effectively the shortcoming of time-consuming large in using directly Monte Carlo method, the theory of artificial neural networks and Monte Carlo algorithm combined with analysis of finite elements (simplified as the FE-RBF-Monte Carlo method) were applied to dynamic reliability analysis under wind load. It is done to organically combine with the finite element numerical analysis, high-powered RBF-NN and Monte Carlo method, then the feasibility of this method was proved by an example. The analysis indicates that the method is feasible, effective, and of high accuracy; it improves effectively the shortcoming of time-consuming large in using directly Monte Carlo method. The method provides a new way for dynamic reliability analysis of a complex catenary. 

catenary, dynamic reliability, neural network, wind load

针对直接利用Monte Carlo方法进行接触网结构抗风可靠性分析耗时长的缺点，将有限元数值分析、高性能的RBF神经网络和Monte Carlo方法有机结合，提出了分析接触网结构抗风可靠性的FE-RBF-Monte Carlo方法，并用实例证明了该方法的可行性。分析表明：该方法精度高、简便易行，有效解决了直接利用Monte Carlo方法耗时长的问题，为复杂的接触网结构动力可靠性分析提供了一条新途径。

接触网, 动力可靠性, 神经网络, 风荷载