Reliability-Based Optimization of Impeller Structure Using Exponential Function Approximation

This study proposes a decoupling approach based on first-order exponential function approximation for reliability-based optimization. It establishes an explicit relationship between the failure probability and design parameters within only one implementation of reliability analysis, without the repeated reliability analysis involved in traditional approaches. Then, the obtained explicit exponential function approximation is used to decouple the reliability-based optimization problem and transform it into an equivalent deterministic optimization problem, where a general optimization algorithm can be adopted to solve steadily with a neglectable computational cost. The proposed approach is applied to conduct the reliability-based optimization of a centrifugal compressor impeller design. The results indicate that the proposed method can provide significantly improved computational efficiency while maintaining optimization accuracy comparable to that obtained by traditional double-loop optimization methods. Meanwhile, sensitivity analysis reveals the relative importance of design parameters to structural reliability.