Global Sliding Mode Variable Structure Control Based on Neural Network

A global sliding mode variable structure control method based on neural network reaching law for a class of nonlinear uncertain discrete-time systems was proposed. Parameters, e and δ, which were determined previously in the conventional reaching law, were regulated adaptively by two feed-forward neural networks (FNNs) respectively. System model was estimated by radial basis function neural network (RBFNN); as well as, system global robust sliding mode control was realized based on the shifting sliding surface. Simulation results show that good tracking performance is obtained; meanwhile system state is always sliding on the sliding surface, so system chattering is eliminated and good robustness is achieved.