In order to calculate the reliability of large scale civil structure without explicit performance functions

a method combineding neural network with particle swarm optimization to analyze structure reliability is proposed in this paper.The structure mechanical analysis is carried out by ANSYS.The highly nonlinear mapping performance function is simulated by BP neural network.The penalty function and the particle swarm optimization are applied to evaluate the reliability index.A program based on the proposed method is developed by integrating Visual C++6.0

standard APDL（Ansys Parametric Design Language） and mixed-language in MATLAB.Numerical examples show that the method can be used to evaluate the reliability of complex structure with implicit performance function.