Real-time synthesis of linear state observers using a multilayer recurrent neural network

Presents a multilayer recurrent neural network for real-time synthesis of asymptotic state observers for linear dynamical systems. The proposed recurrent neural network is composed of two layers of artificial neurons. By solving two matrix equations using the two-layer recurrent neural network, the proposed recurrent neural network is able to determine the output gain matrix of a Luenberger (asymptotic) state observer in real time. The proposed multilayer recurrent neural network is shown to be capable of synthesizing asymptotic state observers with prespecified poles for linear time-varying dynamic systems. The operating characteristics of the recurrent neural network for state observation are demonstrated by use of two illustrative examples