Efficient optimization of noise-reducing orifice plates in nature gas pressure regulators based on adaptive multi-scale sampling-kriging model

The structural optimization of noise-reducing orifice plates in natural gas pressure regulating valves is crucial for engineering noise control. However, conventional optimization methods suffer from low efficiency and high computational costs, limiting their practical applications. To address this issue, this study proposes an efficient optimization method based on an adaptive multi-scale sampling Kriging (AMSS-KG) model. By establishing a dynamic sampling adjustment strategy, the method enhances model training efficiency: increasing sample size in the early optimization stage to thoroughly explore the parameter space, while reducing samples in later stages to focus on optimal regions, thereby balancing computational costs and prediction accuracy. Results demonstrate that compared with traditional KG models and the parallel q-EI strategy, the proposed method improves optimization accuracy by approximately 2.7% while reducing computational cost by 54%. This approach not only significantly enhances the optimization efficiency of noise-reducing orifice plates in natural gas pressure regulating valves-providing an effective solution for their customized design-but also offers a generalizable adaptive sampling framework that can be extended to other surrogate-based optimization problems, presenting a cost-effective strategy for engineering design optimization.