A research about a MIMO system identification algorithm based on ANN using slide mode variable structure

Applying sliding mode variable structure control to train neural networks is proposed in this paper, which can not only increases learning rate but also improves the stability of neural-network. Furthermore, this method helps to promote the generalization capacity of neural-network. In order to improve the generalization ability, two sliding mode objective functions are designed based on previous researches for a new neural-network learning algorithm. Simulation analysis shows that the proposed algorithm increases the generalization ability and robustness of neural-network, meanwhile, enhances the identification accuracy.