A proposed Improve Algorithm for Nonlinear Conjugate Gradient Optimization Methods the Objective of the Computation of Structural Reliability Index

Document Type : پژوهشی

Authors

• b Keshtegar
• m miri
• naser sahabakhti

10.22067/civil.v24i2.28783

Abstract

Accurate calculation of reliability index is very important for the reliability analysis of structures. In some limit state functions with nonlinear characteristic and several local optimum design points, the computational reliability methods may not appropriately determine the failure probability, and iterative reliability approaches may be converged to local optimum design points. In this paper, formulation of nonlinear conjugate gradient methods such as: Fletcher & Reeves (FR) and Dai & Yuan (DY) were proposed for computing the reliability index. Then, the efficiency, capacity and robustness of the conjugate gradient were investigated through several reliability problems. The FR and DY algorithms were diverged in some examples. Therefore, a new modified nonlinear conjugate gradient method was proposed by using the hybrid optimization algorithm to ensure accurate convergence of reliability index. The results illustrated that the modified nonlinear conjugate gradient methods were converged with less run-times and number of iterations in comparison with the FR and DY methods, and robust, efficient and stable algorithms are to calculate the structural failure probability.

Keywords

• Optimization Method
• Nonlinear Conjugate Gradient
• Reliability Index
• Hybrid Optimization Method
• Failure Probability

---