FIRST OREDR RELIABILITY METHOD OF STRACTURS USINGA IMPRVED HARMONY SEARCH OPTIMIZATION

Abstract

The first order reliability method (FORM) is widely used to estimate the failure probability of structures. Appropriate evaluation of the reliability index is more important to estimate the failure probability in FORM. Generally, the iterative mathematical formula of FORM (i.e. Hasofer and Lind- Rackwitz and Fiessler (HL-RF)) was shown the instable solution of reliability index in highly nonlinear problems. The harmony search algorithm can be estimated the reliability index for concave and convex reliability problems due to capability and not need the gradient vector of random variables. In this paper, a global-best harmony search (GHS) algorithm with small harmony memory was proposed. In proposed GHS, a dynamical bandwidth, which is computed based on number of random variables, is proposed to adjust the harmony memory by a random generation with Normal distribution function. Accuracy and robustness of the proposed GHS have been compared with the HL-RF and stability transformation method (STM) through several limit state functions to that have been taken from. The results indicate that the HL-RF formula of FORM was not converged in several nonlinear examples. The proposed GHS was converged as similar as the STM results but it is more efficient )there is required less iterations to converge than STM). The GHS algorithm has top performance both of accuracy fast convergence rate.

Keywords

• first order reliability method
• reliability analysis
• improved harmony search algorithm