Hybrid optimization of curvature continuous stacking line on the highly loaded diffuser cascade

The compressor is a critical component of aero-engines. In order to improve the performance, the compressor ratio of single-stage compressor is getting higher and higher, which will lead to high back pressure gradient and losses. To solve this problem, there are many techniques applied, such as cantilevered stator, tip clearance and slotted airfoils. However, traditional design methods are experience-dependent and time-consuming. This paper proposes a hybrid optimization method to optimize the stacking line of compressor cascade and reduce total pressure loss on both design and off-design conditions. The approach employs various surrogate models and a multi-infill strategy, outperforming traditional optimization methods using a single surrogate model and a single infilling strategy. The results show that compared to the original blade, the optimized blade has a 34.6 \(\%\) $\%$ lower mass-averaged total pressure loss at the design point, while the static pressure ratio increases by 2.43 \(\%\) $\%$ . This paper innovatively combines deep learning-based surrogate models, the hybrid optimization algorithm, and the curvature-based blade shaping method to optimize the blade shape, shorten the blade design time, and ultimately reduce the losses significantly.