Sliding Mode Variable Structure Controller for PSO-RBF Hypersonic Vehicle

Abstract

The dynamic model of hypersonic vehicle has nonlinear and uncertain characteristics. The traditional sliding mode variable structure needs to combine the optimization algorithm to suppress the chattering problem. The sliding mode variable structure controller based on RBF neural network parameter adjustment can eliminate the chattering problem of sliding mode control to a certain extent, but the uncertainty of network self-learning effect will affect the convergence efficiency of parameters, in order to improve the approximation effect of the network. The RBF neural network and particle swarm optimization algorithm are combined organically, the particle swarm optimization algorithm is used to optimize the hidden layer basis function width of the RBF neural network and the self-learning, self-adaptive and self-organizing ability of the center improvement algorithm, and the PSO-RBF tuning law is designed to train and test the height step command and the speed step command. The efficiency of the parameter convergence of the sliding mode variable structure controller algorithm is improved. Through a large number of numerical simulations and controller algorithm comparison analysis, it is verified that the sliding mode variable structure controller of PSO-RBF tuning parameters has strong robustness in suppressing various uncertainties due to interference.