Structural Reliability Assessment Using a Hybrid Algorithm of Artificial Neural Network and Particle Swarm Optimization Algorithm

There are several sources of uncertainties existence in the loads, parameter of strengths, and simplification of complex model of structures in civil engineering problems. Thus, it make reliability analysis and estimation of failure probability of structure inevitable. In some reliability problems it is difficult to find an explicit form for the limit state function. Even occasionally due to discontinuity in the limit state, derivative of limit state needed in the estimation of design point seems impossible. In this study a new algorithm based on the hybrid form of Artificial Neural Network and Particle Swarm Optimization algorithm (ANN-PSO) has been developed for reliability assessment of structural. The proposed method firstly involves generation of training datasets to establish an ANN model, then approximation of the limit state function over the trained ANN and finally estimation of the failure probability using the PSO algorithm. Numerical results show that the proposed method has a good agreement as compared to the other methods such as time- consuming Monte-Carlo approach or First Order Reliability Method (FORM) that needs the derivation of the limit state function in its algorithm

Keywords:

Reliability , Limit state function , Artificial Neural Network , Particle Swarm Optimization