A Multiwavelet Neural Network-Based Response Surface Method for Structural Reliability Analysis

摘要: A new multiwavelet neural network-based response surface method is proposed for efficient structural reliability assessment. Although network can be used approximating nonlinear functions, its application has been limited to small dimension problems due computational cost. The expands the of moderate by introducing a series intermediate nodes, and number these nodes determined theory. Thus, multidimensional function learning problem transformed into one-dimensional problem. Four examples involving one stochastic finite element-based illustrate effectiveness method, which indicate that more up 10 random variables than classical multilayer perceptron-based method.