数据驱动的两级轴流涡轮多自由度气动优化设计

To address the challenges of having multiple design variables， long evaluation times， and poor global search capability of traditional surrogate-assisted algorithms that rely on complete substitution of accurate function evaluations in the turbine aerodynamic optimization process， a pre-screen surrogate model assistant differential evolution （Pre-SADE） was employed. This algorithm was combined with the directly manipulated free-form deformation （DFFD） method to achieve multi-degree-of-freedom parameterized control， resulting in a data-driven three-dimensional aerodynamic optimization platform for multi-stage turbines. Taking a two-stage axial turbine as the research object， 44 design variables were selected for combined channel-blade row aerodynamic optimization design. The results show that after optimization， the turbine’s design point isentropic efficiency improves by 1.10%， flow rate increases by 2.16%， pressure ratio decreases by 0.94%. Besides， the shock wave intensity decreases， the radial secondary flow is suppressed， and the internal flow losses are reduced. Furthermore， the turbine’s design speed characteristics are improved under all operating conditions. This platform not only guarantees optimization effectiveness but also significantly reduces the number of design variables and real evaluations， making it suitable for multi-stage turbine aerodynamic optimization problems with multiple degrees of freedom.