Hybrid evolutionary algorithm for global structural reliability assessment

Abstract

Composite laminate structures are prone to probabilistic catastrophic failure due to uncertainty in material and strength properties. The Reliability Index Approach (RIA) estimates the failure probability as a function of the reliability index and is formulated as a minimization problem. For highly dimensional problems and discontinuous implicit structural response functionals, gradient-based optimization methods suffer from premature local convergence, or even divergence. A new hybrid evolutionary algorithm is developed, to achieve global convergence of the RIA problem with practical efficiency.