Global-local hierarchical anti-vibration optimization of turbine blade based on adaptive multi-Kriging model

This study proposes a novel global-local hierarchical anti-vibration optimization method for turbine blades, integrating two complementary strategies: global structural vibration avoidance to enhance resonance margins and local structural vibration reduction to minimize vibration stress. The framework employs distinct optimization approaches for different dimensional challenges. For high-dimensional multi-objective global optimization, an adaptive multi-Kriging method based on parameter sensitivity analysis (PSAM-Kriging method) is proposed. PSAM-Kriging method combines dimension reduction through parameter sensitivity with cyclic adaptive infill points to improve computational efficiency and predictive accuracy. Traditional adaptive Kriging method (A-Kriging method) is retained for low-dimensional local optimization. Experimental validation on a newly designed turbine blade demonstrates the effectiveness of the proposed method. Global optimization achieved a 1.4% improvement in resonance margin for dangerous resonance order, while local optimization reduced maximum vibration stress by 24.99%. The integrated approach significantly enhanced the resistance to high-cycle fatigue (HCF) of turbine blade, yielding a 37.14% increase in HCF strength reserve coefficient. These results confirm the methodology's capability to synergistically address different vibration mechanisms through hierarchical optimization. This hierarchical optimization paradigm offers valuable insights for complex structural design challenges in turbomachinery applications.