The nested optimization method is commonly used in reliability modeling and analysis of non-probability hybrid structural. In order to solve the problem that the nested optimization structure takes a long time due to the complexity of the reliability model, a de-nesting optimization method based on dimension reduction for hybrid structural reliability analysis is proposed. The reliability function of the hybrid structural is reduced in dimension at the design point and approximated as a superposition of multiple one-dimensional functions. The process of solving the extremum of the inner layer function of the traditional nested optimization structure is transformed into the extremum of multiple one-dimensional functions. The optimization problem is transformed into an unconstrained optimization problem through the penalty function. Finally, combined with optimization, the reliability indices are obtained by multiple iterations. The Engineering examples show that the proposed method simplifies the calculation process compared with the traditional method, which effectively eliminates the traditional optimized nested structure.