Optimal control for vertical take-off and landing aircraft non-linear system by online kernel-based dual heuristic programming learning

Vertical take-off and landing (VTOL) aircraft system is a non-linear complex system with multi-variable large-disturbances. Hence it is a difficult task to control it efficiently. For the first time, this study applies the most advanced kernel-based dual heuristic programming (DHP) algorithm into solving the optimal control problems of VTOL aircraft systems successfully. Kernel-based DHP (KDHP) algorithm is developed by integrating sparse kernel machines and approximately linear dependence analysis into the critic learning of DHP algorithm. Kernel machines are used as approximation structure for the critic module in the proposed KDHP algorithm, while multilayer perceptron neural networks (MLPNNs) are used as approximation structure for conventional DHP algorithm. Compared with conventional MLPNNs with manually designed structures, kernel method can obtain better generalisation capability. Hence to a certain extent, the proposed KDHP algorithm has better learning efficiency and generalisation capability. Moreover, the simulation and experimental results of VTOL aircraft system control problems demonstrate that the optimal controller based on the proposed KDHP algorithm is effective.