Airfoil Design and Performance Optimization of Rotorcraft Robots through the Integration of LSTM-PSO and Multilayer Perceptron Technologies

This study designs and employs the Long Short-Term Memory-Particle Swarm Optimization (LSTM-PSO) algorithm, supplemented by a multilayer perceptron model, focusing on the optimization of rotorcraft airfoil structural parameters in low Reynolds number environments under Mars’s unique atmospheric conditions. A new airfoil was successfully developed through the optimized structure. The advantage of this algorithm lies in significantly enhancing the lift-to-drag ratio under ultra-low Reynolds number conditions and achieving a significant reduction in computation time by using a multilayer perceptron to replace traditional Computational Fluid Dynamics (CFD) simulations. This strategy not only reflects an effective improvement over traditional design methods but also demonstrates the tremendous potential of using machine learning technologies to handle complex optimization problems in the aerospace field, offering new insights for finding complex optimization solutions in dynamic and uncertain environments.