B-spline Functional Link Network Adaptive Control for Hypersonic Vehicles Using Swarm Intelligent Optimizer

摘要

This paper proposes a new recurrent neural network (NN) adaptive control method for air-breathing hypersonic vehicles (AHVs) in the presence of dynamical parameter uncertainties and disturbances. The control law mainly consists of the optimal generalized predictive control (OGPC) algorithm, and an adaptive adjustment of B-spline recurrent functional link network (BRFLN). The BRFLN, a new NN presented here, is applied to approximating dynamical uncertainties and disturbances. Moreover, a guaranteed global convergence algorithm, the stochastic particle swarm optimizer (SPSO) is employed to learn the self-connection weights of the BRFLN, which can improve the adaptability of the BRFLN. It is proved that the closed-loop system error is uniformly ultimately bounded. Finally, simulation results show that the controller designed attains a satisfactory performance for the AHV attitude tracking.