Joint optimization of aerodynamic and mechanical of a transonic compressor rotor based on neural network algorithm

This paper presents a joint optimization of aerodynamic and mechanical of a transonic compressor rotor, considering performance under variable operating conditions. The thickness of the pressure and suction surfaces and the camber line are parameterized using B-spline curves. Optimization is performed on four sections at 15%, 50%, 75%, and 100% span of the blade with a neural network algorithm. The internal 3D computational program is combined with a structural strength solver to quickly evaluate aerodynamic and mechanical performance. The optimized solution is verified with a more refined 3D mesh calculation. The efficiency at the design point is increased by 0.6%, the surge margin at the design speed is slightly improved by 0.2%, and the natural frequency of the blade is increased by 5%, still meeting the avoidance rate requirements. The average stress at the blade hub is reduced by 7.82%, while the peak stress is slightly increased by 4.92%, both meeting the static strength assessment criteria.