Non-gradient probabilistic Gaussian global-best harmony search optimization for first-order reliability method

The performances of first-order reliability method (FORM) are highly important owing to its accuracy and efficiency in the structural reliability analysis. In the gradient methods-based sensitivity analysis, the iterative formula of FORM is established using the gradient vector which it may not compute for some structural problems with discrete or non-continuous performance functions. In this study, the probabilistic Gaussian global-best harmony search (GGHS) optimization is implemented to search for the most probable point in the structural reliability analysis. The proposed GGHS approach for reliability analyses is performed based on two main adjusted processes using the random Gaussian generation. The accuracy and efficiency of the GGHS are compared with original harmony search (HS) algorithm and three modified versions of HS as improved HS, global-best HS, and improved global-best HS based on a mathematical and three structural problems. The obtained results illustrated that the PGGHS is more efficient than other modified versions of HS and provides the accurate results for discrete performance functions compared to original FORM-based gradient method.