Dynamic Decoupling of the MIMO System Based on the Elman Net

Elman net is a kind of well-known recurrent neural networks. Because the original Elman net and modified Elman net can approach the dynamic system, it can be used to complete dynamic decoupling in the MIMO system. The online decoupling algorithm of nonlinear MIMO system based on the Elman net is proposed. As the target function of the Elman net, the generalized cross-correlation function is defined which can implement the online dynamic decoupling. The hybrid genetic algorithm is used to train the Elman net in order to compensate coupling effect. The effectiveness of the algorithm has been shown by numerical simulations combing nonlinear MIMO system.