A Kriging-based decoupled non-probability reliability-based design optimization scheme for piezoelectric PID control systems

Abstract

When dealing with optimization problems, the introduction of uncertainty will greatly increase the difficulty of solving the problem. The traditional reliability-based design optimization (RBDO) algorithm coupling uncertainty parameters and design variables, and solve optimization with optimization algorithms. The traditional RBDO scheme is easy to understand and realize, but it will greatly increase the calculation time, and it often does not get ideal results when solving some problems with high time complexity. The Kriging-based decoupled non-probability reliability-based design optimization (K-BRBDO) scheme proposed in this paper decouples uncertainty parameters and design variables, establishes a decoupled optimization system based on Kriging model, and proposes a new adaptive learning strategy which involves two stages of enrichment to improve the accuracy of the surrogate model in the region of interest. Finally, the effectiveness and accuracy of the optimization scheme are verified by two numerical examples and an engineering example.