机器学习下随机机翼模型的复杂非线性动力学与控制研究

机翼模型的随机动力学与控制问题已经引起了学者们的广泛关注。然而，传统的动力学分析与控制理论和方法主要依赖于一个所有参数均已知的数学模型，且对一些复杂情形尤其是高维或随机问题计算效率较低。为了突破传统理论和方法的瓶颈，本项目拟以经典的机翼模型为研究对象，结合人工智能领域中的机器学习方法，从数据驱动角度研究随机扰动下机翼模型的复杂非线性动力学与控制问题。首先，基于风洞试验或仿真数据，构建神经网络模型对随机及非线性因素进行精确化建模，实现机翼模型的反问题研究。其次，发展高效的机器学习分析方法，探索随机机翼模型的复杂动力学行为，揭示随机扰动对机翼模型的作用机理。此外，通过神经网络模型构建有效的控制器，实现机翼模型的振动控制。最后，对提出的机器学习方法的可靠性及泛化能力进行讨论，旨在建立可靠且适用于复杂随机扰动下机翼模型的新理论和新方法，为我国飞行器设计与制造行业的发展奠定基础。

Stochastic dynamics and control for the airfoil models has attracted extensive attentions from scholars. However, traditional theories and methods for dynamic analysis and control mainly depend on a mathematical model that all parameters are known, and show a lower computational efficiency for some complex situations, especially for high-dimensional problems. To break through the bottlenecks mentioned above, this project intends to investigate complex nonlinear dynamics and control of the classical airfoil models with random fluctuations from a data-driven perspective by combining the machine learning techniques in the field of artificial intelligence. First, with wind tunnel test or simulation data, a neural network model is constructed to accurately model the stochastic and nonlinear factors, and then inverse problems of the airfoil models are studied. Subsequently, several efficient machine learning methods are developed to explore the complex dynamics of the stochastic airfoil models, and reveal influence mechanisms of the random fluctuations on the airfoil models. In addition, effective controllers are constructed via a neural network model to achieve vibration control of the airfoil models. Finally, reliability and generalization of the proposed machine learning algorithms are discussed in order to establish the new theories and methods which are reliable and applicable to the airfoil models subject to complex random disturbances. These results and methods in this project will lay the foundation for the development of design and manufacturing industry of the aircraft.