On a new recurrent neural network and learning algorithm using time series and steady-state characteristic

This paper proposes a new recurrent neural network and the learning algorithm using time series and steady-state characteristics of nonlinear dynamic systems. Recurrent neural networks are often trained using only time series of systems, but sometimes other information about the system to learn can be obtained. Nonlinear steady-state characteristics of systems are important information to improve performance of recurrent neural networks. Furthermore, this paper shows the computational results to verify the performance of the new recurrent neural network and the learning algorithm.