Reliability-based design optimization (RBDO) is an important research area in modern structural engineering analysis and design. Sequential Optimization and Reliability Assessment (SORA) is a wildely used and efficient approach in dealing with the reliability-based design optimization (RBDO) problems. Nevertheless, its dependence on the First Order Reliability Method (FORM) during the iterative reliability assessment gives restrictions, especially when applied to nonlinear systems. However, because the method uses FORM for reliability analysis in the sequential process, its accuracy decreases when handling nonlinear problems.This paper uses the decoupling architecture of Monte Carlo simulation (MCS) to improve the assessment of reliability. By generating extensive random samples, the approach enables precise evaluation of system failure probabilities and reliability index under nonlinear conditions, addressing accuracy limitations inherent in FORM. Meanwhile, the Kriging model is used to replace the constraint functions in Monte Carlo simulation, aiming to reduce cmputation cost. This paper uses the Kriging model as a surrogate for constraint functions in Monte Carlo simulations, which reduces the frequency with which the expensive constraint functions are invoked during the simulation. The method is authenticated engineering case studies, which confirm its efficacy.The reliability based optimization design of complicated structure is provided by the conclusions as a reference.